Good morning, everybody. Good morning. Wonderful to have everybody here. Let me start with some basic housekeeping items, if it's okay with you all. First and foremost, please make sure you keep your badges visible as you'll use the restrooms and be moving in and out. We want to make sure we get you back in the room as easy as possible. The restrooms are just outside the hall here. If you've not attended before, there's Wi-Fi in the room. There should be little placards on your table helping you get logged in. If you have any problems with that, please go to the back room so that people can help you. What I'll do now is take you through, I know what is always the most exciting slide of the day. Please indulge me here. I'm going to just do a read-through of this briefly.
Statements in this presentation relating to Oracle's future plans, expectations, beliefs, intentions, and prospects are forward-looking statements and are subject to material risks and uncertainties. Many factors could affect our current expectations and our actual results, and could cause actual results to differ materially. A detailed discussion of these factors and other risks that affect our business is contained in our filings with the SEC on our most recent reports, Forms 10-K and 10-Q. Copies of these filings are available from the SEC or by contacting our investor relations department. All information set forward today is current as of today, October 5th, 2017. Lastly, we undertake no duty to update any statement in light of new information or future events. Additionally, we will be using, in particular, Mark's presentation. You'll see non-GAAP financial measures being used. Want to make sure that you're aware of that.
Lastly is also to understand, especially as Thomas, Steve, Reggie, and some of our other executives are speaking, there will be some discussion around future products and technologies, just to make sure it's only for your information purposes only. With that, thank you for your indulgence. Let's go through the agenda. We'll start with a technical discussion. Steve Miranda will come up and speak with you about SaaS business, what's going on there. Thomas will then follow to speak about platform and infrastructure as a service, give you an update on the technology. Thomas will be bringing up a number of customers here. The chairs, if you're wondering why they're here, they'll be having a panel discussion, and we'll have some Q&A with that as well. We'll take a break around 1:15.
Now, if you haven't, I know some of you are on East Coast or farther from Europe or Central. If you're on a different time zone, if you're hungry, please make sure you get yourself a snack. The lunches will be ready at 1:15. I know those of you on East Coast time, that'll feel late, so please get yourself a snack. Lunch will come in at 1:15. We'll come back from break around 1:40. At that point in time, Reggie Bradford, who's here, will come up and speak with you about the Oracle Accelerator Program. I think you're going to find it pretty interesting in that we'll talk about how people are developing on Oracle. After that, Doug was here with us last year to talk about the Accelerated Buying Experience.
He'll be giving you an update, not only on the Accelerated Buying Experience, but how we're transforming the company as a whole to basically live and work and play in the cloud. At that point, Mark will come up, talk about the business. He'll invite Safra up. They'll do a Q&A with you all. Then we'll take another short break. Afterwards, Larry will come up. We'll have Q&A as we always do with Larry, and we'll conclude the day. With that, can we start the next video?
Responding to the market conditions for a finance function requires a level of agility that the historical operating model between business and IT does not necessarily support. We are no longer interested in customizations where we are looking for a one way of doing things. We want to be upgraded every six months so that we never have to wait five years to get a functional upgrade and all the costs and pain and challenge that come with that. We wanted a public cloud solution. We were historically an SAP house. We began the journey with them, and in the process, found that SAP were offering private, you were offering public. They were offering customization, you were offering vanilla. We're going live with Fusion GL. We're also going live with Oracle DRM. We're going live with an accounting hub solution, a reconciliation solution.
We have a lot of components that we're bringing live. That is the first time I've ever been part of a project where we didn't have to spend nine months doing requirements. It was a transformative operating model that would change how business and IT work together and allow us to focus on what we're good at. You start with the solution as a vanilla configuration-based solution, and you iterate from there. The net cost of delivering that is fractional to what would have cost us in the past. You think about businesses are spending $60 million-$100 million on these types of programs, and we're operating at a fraction of that. We're seeing these increases in agility and reductions in costs enabling us to be more creative and more aggressive sometimes in solving problems and solving more than one problem.
Can only see us adopting more in the future.
Ladies and gentlemen, please welcome to the stage, Senior Vice President of Applications Development, Steve Miranda.
Okay. Good morning, and thanks for everyone. I'm going to go through the slides very quickly. I know you have them for content. In going through them, the couple of key points that I want you all to remember. First and foremost, if you've been to the previous three days, what we showcased in just about every one of our sessions was not Oracle talking with our customers, but actually our customers, just like Matthew from Lloyds, talking to our customers in terms of the benefit. With the SaaS, we feel extremely confident recommending to our customers across the board that they move forward with SaaS solutions, not based on things that we feel and believe, but actually showing them other customers that are similar. I'll go through that.
The second is we think we're going to extend the lead in applications and really across the board, particularly with our investment around machine learning and the speed that the SaaS applications and the cloud platform PaaS, IaaS, and SaaS deliver going forward. The third is a little bit how we're transforming to become a services company to make this result really come home with our customers. Let me go through some highlights. Any place you look across the SaaS portfolio, which we are talking about for this part of the session, we are in the lead. We have 97 of the top 100 U.S. advertisers using the Oracle Data Cloud. We have grown our customer base in HCM significantly larger than our nearest closest competitor.
We are the leader in B2B and B2C marketing as the measures you see there, both in terms of customers and actual usage. We far outpace any vendor in the marketplace in terms of users, customers, different sites, and phases for our ERP cloud, which we think is substantially differentiated and substantially unique. What do we hear from our customers as far as the reasons why they choose us over competition? First and foremost, across the board, most customers today are moving to SaaS for speed. It's all about speed of innovation, speed of reaction, speed of either disrupting others in their industry or speed to be avoided in that disruption. I'll give you a number of examples as we go through today.
When you get into Oracle SaaS, first of all, we have the most complete SaaS suite of anybody on the market, not only because we have CRM, HCM, ERP, supply chain, but also in those areas, the depth of that solution. I'll explain why that becomes important for customers. Next, probably what gets overlooked is though we have the broadest, it doesn't mean we don't have the best of breed in every solution going forward. When we compete head-to-head against HR, head-to-head against Salesforce automation, head-to-head in financials, we have strengths there that we bring to bear best-in-class applications. Part of that speed, the emerging technologies, which I'll talk about today, things we've delivered in the past as far as user interface, things we've delivered this past year, like IoT applications and machine learning applications, also how we're going to extend those going forward.
Next, our global reach. Not only the global reach in terms of the breadth of Oracle's products supporting globalizations and local rules and regulations, but the global reach of our broader infrastructures. Oracle's consulting, Oracle partners, the training that we can afford, the language we can do, support that can take global customers, really allowing for global customers today, but also customers who are expanding globally. You never have to fear that you're going to end up in a place where Oracle is not already there and able to support you completely with the infrastructure. At this stage, we see many examples of this. Now, in the apps world, it's a little bit complicated. To illustrate when I say many examples, when customers ask us for references or examples of success, they ask us for ERP customers or CRM customers or HR customers, sometimes big customers, sometimes small.
Public sector organizations, commercial organizations, non-profits. Sometimes we're the PeopleSoft as their starting point, sometimes the JD Edwards, E-Business Suite, Siebel, and sometimes NetNew. What we try to illustrate here is just about every dimension when you slice and dice that, do you have a financial services customer who used to be on PeopleSoft HR, who is in Europe, who's moved to the cloud? Yes. Do you have a telecommunication customer who's in the U.S. or who's global, who used to run E-Business Suite, who moved to the cloud service? Yes. You see examples like Cablevision, MoneyGram, HSBC, Siemens, Hearst, Lennox. We can show you the whole spot. Just about every one of these customers spoke on our behalf today, not only on that they've selected us, but on the real business benefits they've achieved by moving to our cloud already, and then expectations going forward.
Really, we made tremendous progress over the last few years, not only filling out the product, but in customer adoption. I think you guys see that in the numbers. Now it's bearing out in terms of customer benefit across all of those dimensions. The important thing that we talked about today or at this past conference was not only the speed of where we've been, but the speed going forward. At this conference last year or two years ago, I talked about how it wasn't a question of if our customers were moving to SaaS, it was a question of when. I talked about the benefits of that. There's some cost savings by us doing the labor and hosting things.
We think it's more secure because we keep you always up to date, and we enhance the technology, as Larry talked about on Sunday, on Tuesday. There's substantial benefits. Just functionally, the speed of innovation is unmatched. That's in part improvements from our engineers, but it's also in part the way the model used to work, or if you were a legacy ERP vendor only. Even if you were producing new capabilities, your customers are only upgrading and adopting that once every five, seven, 10 years at best. Today in our SaaS ERP, in SaaS HR, and SaaS CRM, we update 100% of our customers twice a year. The importance of that has really borne out over the last couple of years.
New technologies like blockchain, like IoT or Internet of Things, and machine learning capabilities, we might have been talking about those a year or two years ago, probably not two years ago, and it certainly would not have been as mainstream or technology advanced as it is today. If you are on a five-year, seven-year, 10-year release cycle and you have technology changes being introduced in weeks and months, you need a technology investment and a platform to react to that. This is what we're seeing not only from the customers so far, but as we talk to them going forward. I'll talk to spend a bit on machine learning in just a couple of moments. The way we do this, of course, is with the complete stack of applications. Let me give you the complete stack of technology.
From the apps perspective, why are all these important? If you are going to take a marketing system and move that to the cloud, a SaaS application, we provide that in our Marketing Cloud. However, now you can supplement that Marketing Cloud with data to do more targeted marketing. I use this example last year. All of you probably see this in your everyday lives. You go shopping for anything online, you go to another website or any other website, completely unrelated, and you automatically start seeing advertisements for that. That's targeted marketing, very likely powered by the Oracle Data Cloud and Oracle Marketing underneath that. However, our marketing apps typically don't run standalone. Even if you're an end-to-end Oracle shop, you have bespoke applications that you've bought either from an ISV or a third party, or you built yourself.
Those applications, if they run on the Oracle stack, can be run in our Platform as a Service because our Platform as a Service runs standard-based technology, Java, the Oracle Database. You can run your marketing app in the cloud, supplement it with a Data Cloud, add other applications. Oh, by the way, if you have completely other workload related to marketing that doesn't have anything to do with the technology stack, you need to move that to the cloud, our IaaS infrastructure allows you to take 100% of your workload and move it to the cloud.
When customers look for an end-to-end solution, though they may be looking at a Marketing Cloud application, really what enables them to get the big benefits of the cloud is the complete stack, from IaaS for full workload, PaaS for extending the application or third-party ISV applications, our SaaS apps, and all of those supplemented with our Data Cloud going forward. Hopefully, you get the sense of the power and the leverage of the stack. When we go through it today, I am going to talk about five sections. I will talk about ERP/EPM Cloud, Supply Chain Management and Manufacturing, HCM or Human Capital Management, HR, CX or Customer Experience, basically sales, service, marketing, and the Data Cloud. Those will be the sections that I carve up the rest of the time.
Now let me just spend a couple of minutes talking about new technology and innovation going forward before I give the individual product pillar updates. I talked about some of them that are really become more commonplace in the world today, both technology-wise and by customer demand and customer adoption. Talked about AI/machine learning, IoT, and blockchain. Just drill down first on machine learning or AI. I think it is important, and I would like to refer to it more as machine learning than AI or artificial intelligence. The reason why is I think artificial intelligence sort of implies there is a little bit of magic to it. It is artificial. You do not know what is going on. Machine learning, though maybe a more boring marketing name, I think better reflects what is happening.
I think it is important to understand what is happening because we think we have a unique ability to deliver machine learning. We feel that there are four necessary components. First is data. Without data, the algorithms that the machines learn on and they use to progress, there is really nothing for them to process. There is no way for the machine without any data to derive a suggestion or a recommendation or an offer to you. Between the SaaS applications we have and the Oracle Data Cloud supplementing that, we have a unique set of assets that no other SaaS vendor has for data to drive the machine learning. Second is you need data science, both a combination of data scientists and algorithms, computer science algorithms, to have the machine learn and process that data to suggest. Third, you need actual applications and use cases to surface those recommendations.
Let me give you an example. We have a new application we call Next Best Offer, which is essentially an e-commerce add-on app. If you go to any e-commerce site today, if you are registered and you bought something from that site, it will probably say things like, "We know you bought product A. We recommend you buy product B." However, when you take that simple use case, you add external data, meaning you have not bought anything from that e-commerce site. You have never even visited that site. We know from your digital profile, other websites you have visited, other things you have shopped for, other things you have bought, other things other people like you have bought, we can all of a sudden give you a recommendation based on that learning.
When you service that through our e-commerce app, if that recommendation worked, meaning you bought more, we use machine learning to tune it so that other people like you get that recommendation the same way. If you did not buy it, we tune the machine learning algorithm to go forward, and kind of iterate on it. The machine learning, the application improves as you go along with that source of data. Finally, to deliver all this, it's a vast array of data, which we'll talk about more in the Data Cloud, as well as the technical [audio distortion] You need data, decision scientists, the applications to actually surface that recommendation and feed it back in so the application algorithm actually improves, and the horsepower to our platform to run all these things.
With this unique combination of assets, we built a host of AI applications, one which I mentioned, Next Best Offer, Next Best Recommendations. If you think about it in the back-office space or in HR, if any of you today went to any customer's website to look for a job it's oracle.com/jobs or /careers, et cetera. You'd be presented with probably two drop-downs. One is a job type. Do you want an IT job, a sales job, a finance job? Another is a location. Are you going to work in the U.S., in Europe, where is it? Pretty universal. You would see the same thing that I would see. If we went to an e-commerce site, we'd see very different things, again, based on this recommendation engine.
We're taking the exact same techniques, data, machine learning, our recruiting app, and surfacing those capabilities for recruiting to give you a better recruiting engine. Based on the results of success, we're going to tune that going forward. Substantially differentiated with the machine learning and the data from any competitor we have in HR or ERP or CRM. This year, what we announced is we're taking that capability and making it much more pervasive throughout the applications. Not just add-on apps, but in everything you do in the application. The simplest example being something like navigation. When you sign on to the system, if we have a consultant and they're signed on to the system Friday afternoon at 4:30 P.M., chances are very likely that they are there to enter their time card.
When I sign on at 11:45 P.M. on November 30th, and my benefits enrollment's due December 1st, it is very likely that I'm signed on to do my open enrollment type of work. Taking that same data, machine learning techniques, horsepower, and making it pervasive through the applications, including some very commonly thought of user interfaces to these like chatbots for service and service use cases, augmented reality for IoT use cases. In the case of IoT, we've built out specialty IoT applications that sit on top of our and anybody else's manufacturing application that ingest the vast amounts of data, whether it be for your own products that you deliver to customers or to the machines or hardware that you use to produce your product. We could do things like predictive maintenance.
We could do things like tuning your supply chain, detecting anomalies in the products that you're producing, and combine that with supply chain planning, again, to iterate your entire life cycle going forward. Combine that with new technology for augmented reality, allowing you literally to walk around the shop room floor, scan with an iPad or another device, the machines, and detect which machine you need to look out for repair or preventative maintenance. The ability, again, to deliver all these things with speed enabled by SaaS applications and SaaS application updates. The next one gets talked about a lot. Again, the same concepts, is blockchain. We're taking blockchain technology, once again, that we built in the underlying stack and that we inherit at Oracle from the applications division, and applying blockchain to things like supplier payments or payments in general, and also for contract and contract negotiations.
This isn't about cryptocurrency, it's about taking the blockchain technology and capabilities and applying that to real-world business cases and delivering that to our customers within our existing applications. Now let me go pillar by pillar to give you updates on what we've highlighted in each product area, and then some of the differentiators we have there. First off, before I get into any individual pillar, across the board with every application I'm going to cover, as with every release, Release 13, we have a substantially new user interface, keeping it contemporary and doing improvements. There's really two dimensions to think about this. One is contemporary. Again, you all experience this in your everyday lives. You update to your new iOS or your new mobile device ID, and you'll see new navigation patterns, new UIs, slight new colors.
It's critically important that our applications stay contemporary, so when you use Oracle apps, it looks just like you're using a consumer device. That's always contemporary. Second is, since we host our applications, we collect a vast amount of data on how customers are using them, which features they're using, which features they're not using, which enable us to tune the application in ways we could never do before by learning and knowing exactly the use patterns that are being used and make those most optimal in the UI. Brand new UI across the board for the applications. Then, of course, I mentioned the adaptive intelligence or machine learning applications. Again, I showed a few examples, but those are going to be pervasive across every pillar. Within the pillars, first starting with EPM and ERP.
If you look at EPM and ERP, things like the SKUs, if you will, or the individual products, GL, AP, AR, expenses, fixed assets, et cetera, we have everything we've always had and anybody's had in an ERP suite of application. That work has been done. Now we've gone to the stage of really enhancing and going beyond what we've even had in our traditional products, E-Business Suite and PeopleSoft. You'll see where we've added localizations, countries like Brazil and India and Argentina. Those localizations were never part of E-Business Suite. You always had to customize or extend. In not only the product dimension, but now in the localization dimension, we've actually gone beyond what we ever had before in our cloud apps, what we had in our on-premise apps with our cloud apps. In EPM, we've now completed that portfolio.
We've always had budgeting, planning, consolidations, but now we've added profitability and cost management, tax reporting, planning, and automatic reconciliation. Once again, if you compare either E-Business Suite or PeopleSoft Hyperion, and you just look at the features those traditionally had on-premise, we've substantially surpassed that going forward, just on the apples to apples. When you add things like adaptive intelligence, more mobile, more BI, it's a substantially more compelling offering for our companies and customers. You see that with customers, again, big and small. The example here being Qualcomm, a very large, long-time customer of ours who've moved their ERP platform. We see this again, big customers, small customers, public sector, global, local, et cetera. I talked a little about the importance of the depth. Let me use this first slide as an example.
When you compare SaaS applications versus our closest ERP competitor, which is typically SAP, and the speed. Still today, the vast majority of the SAP applications are in the on-premise or R3 mode, which are in this three, five, seven, 10-year, at best, upgrade cycles. Regardless of what they're saying in terms of speed of innovation, their ability for their customers to adopt that is tremendously handicapped. If you're in a five-year, even give them credit for a five-year update cycle, impossible to do things like get blockchain technology or IoT technology, which were never even introduced going forward.
The next thing that the breadth allows us to do is, as customers like Qualcomm implement the core of our ERP solution, core financials GL, AP, AR, it also enables the ability to further expand our portfolio and solution within those customer base to come back after that initial success and add components to the picture. The next part I'll cover is our supply chain management. Supply chain is probably the best area illustrative of not only the speed, but now the completeness of solutions. If you go back to your slides of what we presented at this same forum, 2012, we had supply chain, but it was really around the financial part of supply chain, procurement, inventory costing. 2013 and 2014, we started to add components that you could surround your manufacturing base with. Product lifecycle management, transportation, product development.
2015 and 2016, we really got to the core of manufacturing. Manufacturing, order management, pricing, configure price quote, that you, a manufacturer could go live with our solution. Now at this conference, we've introduced eight new products, which just gives us full, complete suite. Maintenance, warehouse management, quality, service. You could read the rest. The result of that is we've gone from service companies moving to our ERP cloud to now core manufacturing. Customers like NCR, who the heart of their business is manufacturing, moving their business to the complete ERP cloud, including now supply chain and manufacturing. This includes both large traditional customers of Oracle, like NCR, but also net new customers like Carbon, the 3D printer just down the road, who started their journey with their first ERP system, moving to our ERP cloud inclusive of the manufacturing.
That's really the heart of their business. Next, let's move on to HCM. With HCM, it's very similar to financials, where we have the complete set of what you traditionally think. Core HR, benefits, payroll, time and labor, et cetera, on the SKUs. We've added substantial new features that surround that. Sourcing and recruiting, including the machine learning recruiting that I talked about in my example earlier. Governance, risk, and compliance for segregation of duties. A volunteering app, so that when companies have a donation matching type of programs, you can register your program, you can track donation manager, and you can track volunteering or encourage volunteering, which more companies are doing these days as their workforce shifts. Employee health and safety.
Not only wellness programs incorporated into HR, but also disaster preparedness and disaster recovery or reaction to it, to have your HR group be able to find employees when events happen and you need to track down where people are. Modernizing some older products like learning management, which used to be training, which many of you go through whatever manager training or compliance training you have to. Supplementing that old kind of next, answer a question training with video training. The way a modern person learns. A YouTube style, including social to recommend videos, record videos, and have that pervasively throughout the system. Very similar to the finance case, hundreds, if not thousands of customers now moving to the platform, big, small, global, and otherwise. Companies like AXA, who've really implemented global.
They started across Asia, but now taking their whole entire set of applications that were on PeopleSoft and moved those over to our HCM cloud. In that example, reducing their employee, new hire employee timing for training from three weeks to less than two days. Once again, when you look at the completeness of suite and the breadth of suite, really two dimensions stand out here versus our primary competitor in Workday and why we're selected. The first is we have a clear and demonstrable advantage when there's HR and financials in the picture. You saw in the video which preceded me, there's at least 35 examples, in the Oracle use case, 35, where there are integration touchpoints between HR and financials, and this unified platform becomes critical.
Even if you ignore that fact, within core HR, we have a substantial lead on the depth and the breadth of the products and the number of products. When you look at things like learning management, when you look at things like recruiting, when you look at things like employee wellness, and benefits, substantial advantage as far as the depth and the breadth of those products when we win on a daily basis versus Workday. Next, let me move to CX. CX being our customer experience. Really sales, service, marketing. The trend we're seeing in the market here is that the lines are being blurred more and more every day, particularly through consumer demand on what's a sales opportunity, a service opportunity in marketing, and especially getting blended with social.
It's reasonably common these days to have your support experience with a third party or a supplier of yours personally to happen through a social network. If you want to get a fast response, those of you who are going home today or tomorrow, you want to get the fastest response from your airline, I would not call their 800 number. I would immediately go to their Twitter account. That is where you'll get faster response going forward. That is commonplace. You have now a multi-channel service organization that's not just a call center or a listening center, but a monitoring center of social. The side effect of that is now brands are engaging with direct consumers subsequent to that service engagement for more marketing.
They now have a one-to-one relationship with you through that service engagement. They use that either directly or indirectly for marketing. If you're a B2B company, they feed that engagement directly into a direct sale, where it becomes part of Salesforce automation. The term you'll probably hear customers talking about is moving towards a much more digital transformation or the digital transforming themselves. What really that means is changing their entire buy-side experience from customers, sales, service, marketing, into a much more integrated approach and a much more digital approach, including social and working more one-to-one with you, their consumers. Another term you may hear companies talk about that's reflective of this is traditional B2B companies trying to get closer in touch with their consumers directly and becoming B2B2C companies.
Pirelli, for example, spoke on our behalf at this conference. They talked about how, yes, they do business with dealers, and they're still going to do business with dealers like they're B2B, but they also want to have direct contact with the end customers who are buying from those dealers. They're using this digital transformation as a way to enable that. Another example being Mazda, 20% faster resolution time in terms of their service request using our CX platform, incorporating that into their marketing channel as well because they now have closer, tighter relationships with their customers. When you compare this against our traditional SaaS competitor with Salesforce, first off, take everything I talked about before CX off the table. They have no ERP, no HCM, to speak of at all.
Important in the ERP, realize what that means is when you need a 360-degree view of the customer, but you do not have order management, you don't know what the customer has bought or what they have paid for or what they owe, it is impossible to have a 360-degree view of the customer. Don't think of CRM, well, that's all you need for the customer. ERP is a pretty critical part of it, we believe. When you drill down into the individual components, both the depth and the breadth, the integration for our Marketing Cloud, CPQ for configure, price, quote, Commerce Cloud, that's online service. When you add to that the Data Cloud, which supplements these in the examples I gave of before, substantially more depth and breadth than Salesforce.
When you look at an integrated solution, that's really where we come into play and really where we win customers who want to transform digitally. The last part I'll cover is the Data Cloud. Sorry, I know I'm going fast, but I've almost got a couple of minutes left. With the Data Cloud, first an explanation for what it is. With the Data Cloud, these are advancements we made this year. We're getting more audience information from channels like Snapchat and Pinterest. What it is over 4 billion user digital profiles. That means we track your cookie activity, be it on your laptop, on your home computer, on your mobile device, through your Twitter account, through your Facebook account. We also have other data, real-world transaction data from credit card companies and other partners of ours.
We create what's called an ID graph, that we, with high likelihood, understand not your name or any of your PII information, but on your digital profile, when you use one of those things, a laptop or a phone or a tablet, or a credit card or a Twitter device or Twitter handle, that there's a connection that we know that that person is the same person. When you interact with a brand digitally, and you may start a transaction through one channel, or you may just browse their website, we can have a better ability to advertise to you or to give you better offers through the other channels by utilization of the Data Cloud. These 4 billion profiles are by far the largest set of digital profiles used by marketers today.
Really, the customer base here is a who's who of brand and brand advertisers. We see examples like this all the time with Lowe's getting a 29 times return on investment in terms of improving their marketing and their digital targeted marketing by leveraging the Data Cloud underneath their go-to-market platforms. The last part I'll just say is all of this, the SaaS application improvements, the customer adoption, we've also changed and frankly introduced a lot of roles that didn't exist at Oracle before to be customer-focused. When we are now running our customers' business as part of our cloud, that's what we are doing. We are running our business. We are much more intimate partners with our customer. We've introduced brand-new roles like customer success managers, implementation success managers, customer feedback programs.
Because when the customer gets the software now, really we are incented, and we work with our customers to get live and get successful, and I think that's what you've seen with the results going forward. There's examples like Customer Connect, which is an online forum that we allow our customers, really as kind of a social network, to engage not only directly with our product managers but with each other. We have 2 events a week. These are product managers giving updates on new features, new functions, or customers presenting to us and other customers on innovative ways they're using the applications. We've grown that to now over 25,000 company members.
You'll see this reflected in quotes from our customers, not just the benefits we're seeing, but like Stenman says, "I can see the transformation in Oracle from a product company to a service company." Or Kaiser, who we're working in tight partnership with, our customers are starting to see those results.
From the benefits of the product they're using, but also in partnering with us so that we are ensuring their success going forward. With that, thank you very much.
10 years ago, when cloud infrastructure was really getting started, no one knew where it would go. It was really a kind of big, open free-for-all.
When we think about building a new cloud, there's some things that fundamentally we think that enterprises need that they don't have today. What we're talking about is best-in-class performance and fundamentally different ways in which we're building virtualization and building a software-defined data center. When you've run big clouds before, you quickly realize that the network is the platform.
Typically, people are used to plugging things into a single rack. What we've done with the physical network is take that rack-level performance, we've blown it out to where the entire availability domain gets that same level of performance.
It's always possible to satisfy a request from Dan with going from point A to point B on this core fabric. When you are interacting with your storage system, let's say, from a compute node, the performance is going to have the same characteristic. Very low latency, phenomenal bandwidth. The second element of this, we've got a great physical network, was that we decided to do off-box virtualization.
If you think about the way that every other cloud provider operates, they have to perform incredibly complex logic around management, around billing, around provisioning, around networking, around compute, around storage, all on the same host where your application runs.
What's very interesting and different about what we're doing with Oracle Bare Metal Cloud Services is that we give you the physical machine, we don't have any software running on it.
I think that was truly one of the core choices that gave us, I think, a very differentiated and very effective foundation on which to build. Another important area of investment for us is the way that we manage governance. Instead of having to have multiple accounts that are very difficult to control, can have a singular relationship with Oracle from a billing standpoint, from a pricing plan standpoint, et cetera, and yet let you internally regulate how your own user, your own project, use the cloud resource, and they benefit from the full cloud experience in terms of very low friction provisioning within the limits that you've set.
It really is first in the industry, having all the benefits of the cloud, all the benefits of bare metal. We've seen large mission-critical applications coming over and getting drastically better performance and drastically better assurances about security than they're getting anywhere else. You cannot beat that.
Ladies and gentlemen, please welcome to the stage, President of Product Development, Thomas Kurian.
Good morning. Can you all hear me okay in the back? Thank you for coming. I'm going to briefly talk about what we're doing with our cloud infrastructure and platform, as well as give you a glimpse of where we're building our software for the next 8 to 10 years. Okay? Whenever we build our software, we're working on it in 2 time dimensions. What you see with the Oracle Cloud infrastructure, platform, software, and data as a service, the engineering for that started in 2007, over 10 years ago. Okay? I'll talk about today, new things we're enabling with our cloud. I will also talk about briefly a glimpse of some of the new areas that we've been working on for the last 3 years that we think will represent very important areas of growth and innovation for Oracle.
From the area of applying artificial intelligence to systems to drive autonomous computing, the areas around new forms of human interface and how people interact with systems, different ways going forward, sensor-driven computing and how it can make business processes more efficient, blockchain. I'll talk about those to give you a sense of these areas we've been working on for the last three to four years. You're seeing the first introduction of capability at this OpenWorld, but they portend where we will see major capability over the next five to eight years. I'm going to start with cloud infrastructure. We're building cloud infrastructure on two important dimensions. First is the workloads that we see are necessary, and the scale of infrastructure that's necessary. If you look, when you went from PCs to mobile devices, the volume of devices grew substantially.
As devices get instrumented with sensors, we believe the volume of devices grow by another order of magnitude. Today, we manage over 20 exabytes of data in the cloud. Over the next five years, and an exabyte is a 1,000 petabytes, we think you will need to manage an order of magnitude more, meaning thousands of exabytes.
When you have that volume of data, humans have a hard time processing it. Increasingly, the data will be processed by algorithms or AI-based systems. We think that the mix of workloads that require those kinds of algorithms will increase substantially. At the heart of it, what we're building is very high scale, very high performant, radically different architecture data centers to enable this. If you look at a data center, you saw a brief video of it. What we've done is built a gigantic flat network, to which we attach compute of various kinds and storage of various kinds. They're attached in what's called a leaf-spine design, so that as the number of these compute nodes grow and grow, the bandwidth between different compute and storage nodes remain 100% the same, even as we attach more and more computers to it.
It's a giant flat physical substrate. If you look at Frankfurt, where we opened our data center region just yesterday, there are 9 million ports. These are not gigabit Ethernet, these are 25 gigabit Ethernet ports. The amount of traffic that that data center supports and the amount of compute that you can attach to it is very significant. Now for a customer, we take that giant flat physical substrate and allow the customer to come in through a graphical user interface or through an API and carve up a software-defined, fully private virtual data center. What do you do? You come in and set up an account. You then map the users you're giving privileges to create the data center. You define a virtual network. You define subnets. You attach compute of various flavors. You attach storage of various flavors.
You can set up load balancing in software. You can set up firewalls in software. You can do DNS in software. There's a whole bunch of additional services. Okay. What we introduced at OpenWorld were 5 new big announcements. The fastest compute servers in the industry. The fastest GPUs in the industry to run AI and machine learning algorithms. The fastest elastic block storage, all flash elastic block storage in the industry. The first cloud to provide, in any cloud provider, 25 gigabit Ethernet networking directly to the host. Our computers have dual NIC 25 gigabit Ethernet interfaces, so in and out, you can do 50 gigabits of Ethernet. The average computer in a public cloud data center runs one single network interface card running 1 gigabit Ethernet. It's about 50 times more bandwidth down.
For big data and various kinds of workloads, that's very important. We've also integrated the highest scale elastic DNS service. We're measuring DNS traffic from 300 edge locations globally. We correct 240 billion data points every day, we optimize the traffic, we have very low latency query times on DNS. Why that's important is if you get a DDoS attack on your site, you can push out the traffic globally and distribute it, which ensures that your site doesn't come down. If your North American data center goes down and you want to route traffic somewhere else, you can route very efficiently. You may wonder, why is performance such a big deal? In an on-premise world, when you bought a machine, you had the machine. If it was fast or slow, it didn't really matter once you bought it.
In the cloud world, the faster the machine, the less time you need to complete a computational job. The faster the machine, the fewer processors you need to run a job. Speed is inversely proportional to cost. Here at OpenWorld, we announced 3 new benchmarks. SPECint has nothing to do with Oracle software. This is a straight industry benchmark to see how fast compute is. You can see our numbers. Smaller in these charts mean better. TensorFlow is the industry standard AI algorithm. This is the industry standard ImageNet TensorFlow benchmark for AI. You can see our numbers. Again, smaller is better. This is an industry standard IOPS benchmark to see how fast you can run a workload. Just to be clear, the expensive thing in everything is not compute.
When we say four times faster, it means four times less compute, and people say, "Well, compute is $0.75 an hour, so that's $3.80 an hour." It's not really that. It's the cost of the software licenses that run on top of it that dominate the cost If you took a system that runs four times faster and it requires four times less processors as a result to run the same computation, you are saving four times on all of the software licenses and maintenance you pay on top of it. The computational comparison on cost is not just on IaaS cost savings, but on all the software license and maintenance you're paying on top of it. What's the value prop with IaaS? This is a typical customer workload. Cost is in cents per core of compute per hour.
On-premise, people pay $0.025 roughly for facilities. Something like $0.10 approximately for hardware. It depends. If you've got an incredibly old machine that's fully depreciated, you may be. On average, because most companies have a mixture of new and old, you pay for software licenses, mostly operating system and virtualization, on top of that, ongoing cost of maintenance. What IaaS does is it takes the bottom three pieces and it delivers it as a service. Then it takes the top piece, which is maintenance for people, and reduces that because you need fewer people. They are more highly paid. They're called DevOps people as opposed to IT administrators. You don't save all of that money. You will save significantly relative to running a system on your floor.
If you look at Oracle's economics, we had virtually nothing on the left side of that picture. The OSs were owned by other people. Facilities, we were never in the facilities business. Most people were running Intel from Dell or HP or somebody else. Every time somebody chooses our IaaS, we make money. Okay. The differentiators competitively. Fastest and cheapest. That's relatively easy to say. Right. If you cost the same amount and you go two times or three times as fast, you save a lot of money. I said, it's not just on the IaaS that you're saving, it's on the hardware above, on the software licenses above the IaaS that you have enormous savings. Best reliability and high availability. There's enormous number of differentiators we have in how resilient you can run a system. Here's an example.
There's a number of ISVs that have done press releases at Oracle OpenWorld, SAP, FICO, Calypso, Manhattan Associates, Informatica, MapR, Qubole, et cetera, running on our cloud. One of them is a company called Calypso Software. They're a company that sits and that provides the software that does risk management and financial trading for many, many, many of the largest funds in the world. They started a project with us last year. They actually met us at OpenWorld, started a project with us. They chose us over every other cloud. They went live in January. 70% of the North American business that they have runs on the Oracle Cloud. They manage over $3.8 trillion of assets on the Oracle Cloud. They chose us because of the mission-critical nature of the application and the degree of reliability and availability we provide. Deployment flexibility.
If you're a company in an industry where you're regulated, take AT&T. If you're a company in a country where you cannot put data outside the country's boundaries, take Banco de Chile. If you're an organization that has patient data or highly sensitive data, take Glintt, which is the largest hospitals management system in Western Europe. They all have the choice of running Oracle's infrastructure and platform behind their firewall in Cloud@Customer. No one offers this capability. We have over 200 customers with over 70 live using this. There are, even in an area like infrastructure, which people believe is commoditized, there are fundamental differences in what we're doing versus the competition. Let's look at platform. If you look at our platform-as-a-service vision, it is very different than what other companies look at with platform.
It started with a fundamental realization in 2008 that platforms were not managed efficiently at all. Okay. If you look at it, people install our software. They configure it. They back it up. They patch it. They upgrade it. They set up disaster recovery. They encrypt it. They monitor it. They do change management on it. Not only is it extremely laborious, it is extremely inconsistently done, and as a result of it, you find a lot of problems. Our vision for platform as a service is very simple: eliminate all human labor, 100% of human labor in monitoring, in configuring, in administering our software. Do it not by us hiring an army of cheap DBAs, but do it by having the software do it.
In doing that, we can fundamentally lower the cost of owning and operating our software at a very radically different price point than anyone else can offer. Not only does it lower the cost for people, but it makes operations run way more reliably. This platform vision allows us to build eight big areas of capability on top of the infrastructure as a service that I talked about. Today, in the interest of time, I'm going to focus on just three of them: data management, application development, and analytics. If you look at data management as a service, we run the world's best database, Oracle, better than anybody else. We also run a number of other databases. We have the fastest MySQL and MySQL clusters in the world. We offer Apache Cassandra as a service. We offer HBase.
We offer a number of other data management solutions for customers to manage structured and unstructured data at very large scale. If you look at what we're doing with the Oracle database, the major thing that we introduced at Oracle OpenWorld was a fully autonomous database. Think about running a data warehouse. It is a very complex thing to run a data warehouse well. There are very few DBAs who know how to administer a data warehouse well. You have to decide which of 300 types of indexes to create and manage. You have to decide what your partition allocation mechanism is. You have to decide how much to let data age out and rotate partitions out. You have to decide how efficient parallel query is running.
You have to decide how much allocation of memory to do because different nodes need more memory depending on the mix between read and write. As a result of it, if you're a line of business user, you don't get your own data mart. Now for six years, we've been working on providing capability in Oracle Database to make it fully self-managing. When I say fully self-managing, no human at all required. You load your data, you query it. Who creates the index structures for you? We do. Our software does it because we know every query that's happening on the system and what's the best way to optimize it. Who handles capacity addition? We do because we know how to do it better than anybody else.
Our software fundamentally knows where we need to add storage and how do we optimize, defragment the storage underneath the database and reallocate it. We're doing the same thing for transaction processing. I'll give you an example for transaction processing. For 17 years, we've been studying database health. There are two kinds of failure conditions, performance degradation and crashes. Performance degradations happen very incrementally. You can see them trending bad and unless a human tries to adjust it and makes a mistake, you have enough time to go optimize things. Health on the other hand, the median time between you're healthy and you're down is 4.2 seconds. Even if the software instantly discovers it and alerts the DBA, there's not enough time for a human to come online and resolve the issue. In a fully autonomous world, our software takes care of it.
We know that the system is not healthy and we'll resolve it, we will ensure that your system doesn't crash as a result. These fully autonomous systems do three important things for us. Dramatically lower the cost for customers, broaden our reach into many parts of the organization that would love to use our technology but don't have the skills to do it because it's now so much simpler. Third, dramatically improve the reliability and consistency in the way our software runs. How do you build applications against this? We support five styles of application development, recognizing the mixture of capability that customers want in their own organizations. Migrate and extend is lift and shift legacy applications. I've got an application that was written 22 years ago. I need to shut down my data center. You can lift and shift it into the cloud.
An example of what we do differently there, we offer bare metal as a standard environment. Many legacy applications were written before virtualization. They have a hard time running inside a hypervisor. In our world, you can lift it clean up to the cloud because you've got a raw physical server to deploy on. Cloud native is using a PaaS layer to spin up workloads. Container Native. We introduced it here at this conference. We've had it for a while, but we introduced it formally at this conference. There's a technology called Docker and Kubernetes that's becoming very popular for people building something called microservices. We have a fully managed platform. You simply tell us how big a cluster of Docker nodes you need and spin up that using Kubernetes. You don't have to deal with managing it. We offer that.
Serverless is another thing that we introduced here at OpenWorld and at JavaOne. That's for developers who just want to write functions. They just want to write their code, and they're like, "Frankly, we don't care what infrastructure you deploy it on. Scale it out for me. I want to write what's called a reactive program that listens to events, adds capacity on the fly. You manage the infrastructure for me." Visual development is for those who are truly non-technical and want to drag and drop and assemble applications. We support all of these, which means we are able to reach a very broad community of developers. We've also integrated this into open-source tools to do what's called continuous integration and continuous delivery, or as some people call it, DevOps. You check in code, you check into source control. You can kick off a build and test.
You'll get a notification if you're a developer on your Slack channel that the build is ready. You call Kubernetes. That'll spin up a Docker cluster. You can deploy your executable as a Docker image, and up and running. We've got an extremely clean way of supporting all of these different styles of application in one continuous integration, continuous deployment pipeline. Who's using this as an example? Banco Santander wanted to take their banking services and make them available to, let's call it millennials, who want to bank but don't want to use an app. They want to message with the bank from SMS. They built a mobile banking solution with what's called a digital assistant or a bot, front-ending their core banking. AXA Insurance moved their claims management, something called AXA Direct, onto our App Dev Platform to deliver claims management directly on the web.
You don't have to fill out paper and forms if you've had an accident or whatever it is you've had, and you're managing your claims. Back office. You're worried about who in the back office is doing this. BAE is one of the largest aircraft maintenance companies. Their entire maintenance, workforce scheduling for maintenance, repair, that entire system runs on our application development platform. I said, why are we taking this focus in autonomous computing? It's because of this chart. If you look at a typical customer running any system, Oracle is one example. Facilities cost, hardware cost, software operating system is a small percentage of the total. You have the Oracle license, then the biggest cost is the cost of people. Typically, there are six sets of people. Storage admin, network admin, sys admin, sys DBA, apps DBA, security admin.
With the autonomous solutions that we're building, you don't need almost all of that. You save dramatically in cost off the top line, not off the infrastructure pieces, not off the facilities' hardware and software, but off the total run and manage. This number that I've got in front of you is actually from a customer who spoke at OpenWorld. Performance is still important, even in platform, and you can see our results that Mr. Ellison shared in his keynote on Sunday. We're slightly faster than the competition running any other database, and we're also substantially faster running Oracle. When you look at our platform, the areas that we have fundamental differentiators. A unified platform. Just to give you a sense, between infrastructure and platform, since last OpenWorld, we've introduced 98 new services. 98 new services. Okay?
There's a lot of new innovation, but it's on one architecture. One architecture. It's not on a plethora of architectures. Fully autonomous. I think a long time ago when we started, people said, "Is PaaS just an image of your software that you can configure on top of IaaS?" That was never our vision. Our vision was always to eliminate the need for skilled people to operate our software. By doing that, to dramatically lower the cost, broaden our reach, and differentiate from the competition. Now, how easy do you think it is for somebody else to build an autonomous database or somebody else to build an autonomous Oracle Database? Impossible. They'd have to go back 17 years and observe everything we observed for 17 years. They'd have to go back in time in order to understand what conditions we see and to build the algorithms.
The algorithms that we use are extraordinarily specific in order to ensure that, for example, we're not doing any false positives, right? Like take health. You don't want to kill the patient if the patient is not sick. You have to be very accurate in detecting if the database is having a problem or not, because if you over-correct, you're going to mess up the system. That only comes from learning for the last 16 to 17 years what makes it healthy or not. Obviously, we have a lot of customers who use our platform to extend application software as a service. There are other differentiators as well. With that, I've got two more sections. Does it make sense to stop and take a few questions now or hold till the end?
We'll take a couple of questions now to break up the session, we'll come back and talk about our vision for analytics and the other pieces. Yes, please.
Thomas. Oh, I'm sorry. Hi, Thomas, John DiFucci from Jefferies. That slide you put up there was really interesting with the numbers. Thank you for that.
Yeah. Do you want to go back to that?
Yeah, that'd be great.
Could you go back to that? One more. I think back to the I think you're talking about the green, red, and blue one, right?
Yeah.
Yeah.
I'm color blind, but yeah.
Okay, yeah.
On the left, though, that Oracle license.
Yes.
When you're comparing these costs here, that Oracle license cost isn't spread out over the lifetime of the customer. It's like the upfront cost. I just want to make sure I'm looking at.
It's the license and maintenance together, blended. We did it averaged over five years.
Averaged over five years.
Yes.
Okay. Even if you were to take, if I said your customer was 20 years, and
Yes
took the average, that would be reduced
Yes
It would still be higher. Okay.
Yes.
I just want to make sure of that.
Yes.
Thank you very much.
Yes.
Hi, Thomas. Michael Turits from Raymond James. Can you talk a little bit about the mix of both demand and how you're approaching it between hybrid cloud and pure public cloud, and also managed cloud with your at customer and Oracle Cloud Machine?
Sure. Just to be clear, I'll cover what we're seeing as the mix between hybrid, which is defined as at customer and public, multi-cloud, which I'll explain in a minute what we do there. Currently, our mix is about 20%, 15%-20% cloud at customer, 80% public, meaning public is in our data centers. We also have a number of offerings that are multi-cloud. Okay? When I say multi-cloud, multi-cloud is you're using, take security as a service, and you say, "Hey, every cloud vendor says use me for identity management. Use me for system management." We've got six offerings in our cloud that run across a variety of other clouds as well. When I say other clouds, Salesforce, Office 365, Azure, Amazon, Google, Rackspace, pick your favorite.
If you map your users from on-premise and you say, "I want one place to manage the user identities in the cloud. I don't want to have an Oracle identity and an Azure identity and a Google identity and an AWS identity," you can use us as an identity system. If you have services that you've deployed on other clouds and you want one place to manage it, you can use us to manage that. If you've got a salesforce.com application that you want to integrate with somebody else's, and you want an integration bus to connect the two, you can use us for integration. Multi-cloud deployments, we're seeing about 15% of our base doing that, and we were the earliest in.
A long time ago, we recognized that most large companies will have multiple clouds. We built five or six different services that span this environment. Did I answer your question?
Yeah.
Okay. One last one or keep going?
Yeah. Hey, Thomas, Mark Murphy with JPMorgan. How do you keep this huge global community of Oracle DBAs fully engaged and fully in your corner while you're announcing the driverless database? If you're the driver that's becoming driverless, it might not be the most uplifting realization for them. Can you find some other way maybe to appeal to the business owner?
Of course. Well, the DBAs, I'll be frank, this is the same question we've addressed for many, many years, right? Let me explain myself. Prior to 8, prior to introduction of Real Application Clusters, you couldn't really run a very large database because the size of the database was one machine. People said, "When people have hundreds and hundreds of little databases, aren't you going to take away the capability from the DBA by consolidating the larger databases?" We said, "It's better for the DBA to have fewer larger databases than lots of little ones." Most DBAs loved when we came out with technology like clusters, because it gave them the ability to run a much more efficient infrastructure. There will always be work for the DBA to do. In fact, work that they far better enjoy than they do today.
Look, no one wakes up on a Monday morning, no DBA on Earth has woken up on a Monday morning and said, "I am so excited today because I'm going to patch Oracle all day." Not a single person. We've tried to find that person, believe me. When we do the work for them that they don't want to do, they have a lot more important things they can do. Take an example. One of the things a lot of DBAs would like to do is to look at if somebody's looking at data inside the database they're not authorized to see. I'm not talking about some rogue attacker from outside the company. A lot of the vulnerabilities happen are within the company.
You want to spend time analyzing the query patterns on data based on user behavior to see are people looking at data they're not supposed to see? Those tools, the average DBA never gets time to do that because they're so busy with day-to-day stuff that they never get to do that. We're not talking about not needing DBAs. We're talking about making them do other things that are far more valuable for a business than mundane things like patching. Okay? Let me go forward. There are two more sections. I want to talk briefly about analytics and then what we're doing with new technologies. What's our vision for analytics? If you look at analytics for the last 30 years, it has been what's on the left side. People look at numbers, they take those numbers, aggregate them in a set of charts.
They have a professional user do that analysis, and then those people then derive insight from it and tell the rest of the world, "This is what it is." Now, many, many things have changed, and we've introduced over the last four to five years a new suite for analysis. It's predicated on the very basic principle. Analyze any data. Not just numbers, images, sounds, audio, video, sensor data, textual data, Twitter streams, anything. Present it not just as charts, but for example, you can say, "Monitor revenue in the Northeast for me." Leave that with a digital assistant. That will run the analysis for you and monitor what data is changing. It can send you an SMS if that KPI changes, and it can leave you a voicemail telling you why it changed.
That's a very different experience with analytics than having a human being sit in front of a chart and run a report every day. Third, it is designed to make things extremely easy so that anybody can do the analysis. Okay? When I said anybody can do the analysis, you bring up data in an interface. You say, "I want to mix revenue and product." Now, underneath the covers, we're creating an aggregation mechanism called a cube in order to allow you to summarize revenue by product. Underneath the covers, we're generating what's called an MDX query on a MOLAP cube. Humans don't need to know that. We're simplifying that because we know what analysis you're trying to do. That broadens our reach with analysis to everyone in the company.
Not the 10% of the company that were power analysts, but the 100% of the people who want data. We've introduced a brand-new suite. You're welcome to look at Forrester Research, for example, that just came out with their latest cloud BI vendor comparison. We are the winner, and we're the only one in the leader segment. If you look at it, numbers, images, audio, video, various kinds of data can be replicated into a cloud. You process them, you summarize it and visualize it, and our goal is to offer an end-to-end platform. The heart of it comes down to two pieces, okay? When you move data in, you can spin up a scalable Hadoop Spark cluster. You can stream in data through something called eventing, and then you do these five things.
Imagine you've got images from Instagram of athletes and you want to compare how many mentions that a particular athlete's had. You want to actually first understand how many pictures people posted of a particular athlete. You need to do what's called entity extraction, which is process the image, extract the entity, say, "I know that this is a picture of" Pick your favorite athlete. You want to clean the data, you want to enrich it with data from your corporate systems, and you want to blend it with data from other sources, and then, most importantly, we are processing a very large volume of data, you will not have a human be able to understand all of it. We have algorithms that are custom-built to allow you to do statistical processing as well as AI on this data.
The output of that data can be put into our autonomous data warehouse if you want to compare it with other data from your corporate systems, or it can be processed. Okay? You can model it in an analytic tool. You can visualize it, as I said, in a variety of ways. We're not saying charts don't make sense. If you think of the human interface, right? People have five senses. All computers are restricted to two, your eyes and your fingers. Fingers to enter stuff in a keyboard and your eyes to see stuff. We believe you should use all your five senses. We're going to send you data on voice. You can say, "Here is why a particular number changed," not just, here is what the number is, but also why it changed. Because the system understands the patterns in the data.
Our goal with this is any data, any source, anyone can do analysis, and the system will monitor your business for you and tell you what changed and why it changed. The why it changed is really important. What's our competitive differentiators? All forms of data for analysis. Every form of data. End-to-end solution. Not just a warehouse, not just an analytic tool, but an end-to-end data pipeline. No one offers a full data pipeline. Ingest it, stream the data, cleanse it, process it, aggregate it, model, visualize it. If you use someone else's, you've got to assemble all these pieces from 15 different tools. Most importantly, why we were chosen as the winner in that Forrester review was self-service. All analysis should be enabled for everybody without waiting on anyone. Who are using this? I'll give you a couple examples. Yum! Brands.
They use us to allow a hotel manager or a restaurant manager to be able to look at what products in the restaurant sold, how successfully, at the end of every day. Their challenge was the following. They could take all that data from that store or restaurant, ship it up to their corporate data warehouse, standardize the data in the warehouse, 3 weeks later, they could get the restaurant manager to see the data. They wanted it the same day. Second choice, put a server with a database in every restaurant next to the boiling furnace in which you're cooking fried chicken. It's typically a bad idea. What they did was load your data, run your analysis in the cloud, 1,500 restaurants get their analysis every day from our cloud. The U.K. National Health Service wants an idea of how good customer satisfaction for their patients were.
Wait time, average time to schedule an appointment, which specializations needed more staff, et cetera. They get their analysis from our analytics cloud. This is an area where most people look at Oracle and think when you guys say analysis, you mean numbers with charts. Absolutely not. This has been a significant growth area for us over the last several years. Emerging technology. Steve talked about a number of this, but what I'm going to do is just give you a glimpse of why Oracle is uniquely positioned in these areas. The first one is evolution of business applications. If you looked at business applications in the past, you saw three things. A professional user, like an order management clerk, a supply chain expert, go to a screen called a form, enter some data.
It kicked off a workflow that then ran through a rules engine called a constraints engine that said, "Hey, this order should be shipped from that warehouse because we don't have it stocked in this warehouse, and that's the nearest one." That's what applications used to be. That's not where we see applications going to be. Where do we see applications going? Professional users to self-service. All business processes will go self-service, 100% of them. Everything. Accounting, ordering, HR, supply chain, manufacturing, everything. When you make that self-service, business rules that are encoded inside the application will be replaced by algorithms called AI-based algorithms in many cases. Let's take a very simple example of why we believe that'll be true. All of you do approvals and workflow, yes? If you look at the average company, for every 100 approvals, 99 are approved, one is rejected.
The average human is looking at 99 approvals to find the one that they will reject. Why can't a rules-based system capture that? A rules-based system cannot capture that because people are idiosyncratic. You take two different people, same list of approvals, they will reject different things. You cannot write a rules-based system to capture every human and what they're likely to do. An algorithm can figure it out. A learning system can look at your particular set of rejections and over time, flag just the one that you're going to reject. AI to us will replace rules as a centerpiece of all of these calculations. I'll give you three examples of things we're doing. First, we built an artificial intelligence machine learning platform as a service. That's part of our PaaS. You can run any algorithm you want on it.
We have built our own unique, extremely domain-specific algorithms for our apps. ERP, HR, CRM. Examples. In ERP, identifying what's a fraudulent transaction in procurement. Okay? Fraudulent transaction in procurement is I've got a kickback scheme going on between a supplier and one of my procurement people. You could write a set of rules, which used to be called governance risk and compliance controls. As you know, the smart ones always know how to get around the rules. Algorithms can find fraudulent transactions way better than any rules-based system. Pricing. What's the best price to offer somebody when they place an order? What's the best job to offer a recruit? There's 100 of these examples. What are we uniquely positioned at? To do it well in the applications, you need a platform that provides great scale and power to run AI.
I talked about our GPU facilities and Infrastructure as a Service and our PaaS for AI. We're the only ones who combine applications and platform, and in this space, that gives us an advantage. How domain-specific are these algorithms? They're not at all like what you hear about the consumer companies working on. Let me explain why with one simple example. If you're looking at pictures on your favorite photo site and the search is using an AI algorithm, and you enter, "Find Steve Miranda in a park," and it picks out a set of pictures. Whether you think it's Steve Miranda or not is an orthogonal question. You really don't need to guarantee how the algorithm found it was Steve Miranda in the park.
However, imagine you're in procurement and a transaction gets flagged by the engine saying, "This is a fraudulent transaction." As part of your, let's call it, discovery process, you will want auditability on that algorithm to know how did that algorithm find that problem. That is an extremely hard problem in AI. It's called traceability, and we are working on solving that. This is a very different space than consumer algorithms in terms of how domain-specific they are. Because we have a lot of history there and we understand the domain, we obviously know how to build a bunch of these algorithms. Blockchain, similar example. We offer a blockchain platform as a service that lets you build a blockchain. It lets you build smart contracts. It lets you build a Hyperledger. It handles distributed nodes. It allows you to do all of that.
On top of that, we offer in the applications themselves, capability that lets you leverage a blockchain without having you have to do a lot of work. Let's give a simple example. In intercompany accounting, intercompany accounting is two different business units in a single company reconciling a transaction. Okay? Often, the controllers of the two divisions argue over what's the real financial allocation of that transaction. What price was the intercompany charged between this business unit and another? With the Hyperledger and blockchain, that becomes irrelevant because whenever that transaction is executed, you can guarantee the integrity of the audit trail of that transaction. Therefore, you don't have this intercompany problem to deal with because all of those are calculated in the Hyperledger and tracked in the Hyperledger. We're going to apply it to a number of domains.
if you're an applications company, you need to have a platform allowing you to do blockchain. If you're a platform company and you want blockchain adapted in a bunch of these scenarios, having an apps offering helps you. Same thing with IoT. We offer an IoT platform as a service, as well as applications. People use both. If you're in Japan and you travel in SoftBank's ride-sharing service in Tokyo, it is monitored and the fleet is optimized using our IoT service. We're monitoring all of the little cars that are driving around, and we're optimizing where they're going. If you're in industrial manufacturing and you're a manufacturer called Noble Plastics, one of the largest manufacturers in the world, we monitor their industrial robots on the shop floor and do predictive maintenance and failure detection using our IoT supply chain apps.
we offer both the platform and apps. To do AI well, you have to have data, because garbage data fed to an algorithm, I'll guarantee you, will give you back a bad result. we started on this Data Cloud back a long time ago, in 2013, and most people asked us, "Why would you be building a Data Cloud?" Because we always knew that applications eventually would have rules replaced by algorithms, and algorithms needed incredibly accurate data in order to work well. in this area of emerging technology, it's very simple, our differentiation. In artificial intelligence, we have incredibly domain-specific knowledge that allows us to build algorithms for specific domains because we've had a long history in those application domains. We complement that with the largest, highest quality data in the Data Cloud.
For domains like blockchain and IoT, our strength in both platform and applications. Both platform and applications differentiates us from the competition. these are, I just gave you four or five areas. There's a lot of new stuff we're working on. I gave you four or five because most people have asked us, "What's beyond the cloud for Oracle?" we wanted to give you a glimpse of where we see the next five to 10 years. Okay? in closing, we talked briefly about the cloud. It has been a 10-year journey building our infrastructure platform, application software as a service, and data as a service. We now have a fully complete suite. There's always new stuff to do, but as I said, in infrastructure and platform, we introduced 98 services. Steve introduced almost 200 modules since last OpenWorld.
the rate at which we're introducing stuff, you can compare us to any other cloud vendor. We're doing a lot of new capability every year. Not only are we doing that in the existing modules, but we're also introducing a lot of new capability in these new areas. it's the combination of both of them that drives solutions into the customer base. It's the combination of the breadth of our suite and the new innovations that differentiate us as we go forward. with that, I'll take a couple of questions, and then we've got a customer panel. Any questions?
Hi, Thomas. It's Adam Holt from Oppenheimer.
Yes. How are you?
Good. A couple of questions about the IaaS and PaaS business. It seems over the last couple of quarters, in addition to doing a lot on the product side, you've also had some learnings about how customers are consuming, particularly IaaS business. Can you talk about some of the things that you've seen, specifically addressing maybe the usage of some of the cloud machines relative to some of the other types of infrastructure as a service?
Yep.
Then maybe longer term, given how big these markets are and how under-penetrated they are, do you think the IaaS and PaaS businesses can grow faster than the applications businesses over time?
What have we learned? I'll give you just one example to illustrate what we learned and how we translate it into an action, because often people say, "What did you learn? How do you convert it?" One of the things that we've seen that customers have really struggled with is they have two challenges that they struggle with. The time they have a budgetary cycle is distinct and separate from the time they're in a consumption cycle. Right? Because they say, "Hey, we've got a budget. The budget fiscal year cycle ends in September. My projects are really going to start in December." They have a difference in time between when they can spend versus when they consume. That's number 1. The second is people have an abstract view when they look at a workload, even an existing workload, let alone a new workload.
They have an abstract view of what services they want in the cloud. They say, "I think I'll need 20 cores of compute and 4 cores of database, and certain amount of object storage." And then when they actually prototype the application, they realize, "Oh my God, that's actually not what I need. My database is far more heavy than I thought my compute was required." That was one of the things that we saw a lot of customers asking us for, and we came up with a new scheme called Universal Credits. Universal Credits says you don't have to buy a particular SKU of our infrastructure or platform. You give us a dollar commitment per month on what you're willing to spend, and you have rights to any and all of the SKUs in infrastructure and platform.
Which makes it extremely flexible for people to consume and avoids them having these two problems. One is the consumption cycle being tied to the budgetary cycle. Because now you can budget and say, "This is what I think I roughly will be willing to spend." And then you can later on decide what projects you want and what components of cloud services you specifically need for the project. That was one of the things that we learned over the last few months observing IaaS. The second thing we saw was when our Cloud@Customer, when we're deploying at customer, we have to work with the customer's data center people in order to attach our compute into their core corporate network. We've now automated a lot of that to deploy much, much faster. Maybe one last question. Yes.
Hello, Thomas. Trip Chowdhry, Global Equities Research.
Yes.
Been here for almost seven days. I can clearly say that you have changed the industry tone in the enterprise software with many things you announced. I have two quick questions. First, in your machine learning algorithm that you have placed, in some specific instances, do you have human in the loop? If so, how do you think that could progress? Second, your customers seem to be quite excited by having bare metal GPU offering.
Yes.
Why do you have it, why AWS doesn't? Thank you.
Okay. Why do we have a bare metal GPU offering? Okay. GPUs are often They're called graphical processing units. They're designed to run AI and machine learning algorithms. When people deploy these algorithms, many are used for what I'd call intermittent computation, meaning you're not running the algorithm all day, all night. You're running it, you load a certain set of data, you run the calculation, and you're out. Virtualization introduces significant overhead. A typical hypervisor overhead is between 15%, 20%, even 30% in certain cases. A lot of customers say, "Can I get a raw physical GPU and be able to just run my algorithm and get out?
The faster I run the algorithm, the faster I get my results, frankly, the most cost-effective it is." The reason we're able to offer bare metal GPUs, as you saw in the video, we took virtualization from being done in compute and put it into the core network. On a single, fully virtualized layer 3 core network, we can attach physical processors, either regular compute or GPU bare metal. You can run hypervisors, or you can run Docker on top. That's one piece. Now, algorithms for AI and human involvement with it. Our algorithms are designed to address extremely domain-specific scenarios. I'll give you just one example to illustrate what I mean. I'm going to use an e-commerce example, okay? In e-commerce, one of the hardest problems with product recommendations is understanding what product to put in front of a customer, okay?
For many years, there's been this notion of collaborative filtering as the state-of-the-art. We've introduced a new algorithm called proximal control, which is far more accurate than collaborative filtering. Why did we introduce that algorithm? We introduced that algorithm because collaborative filtering works well when you have a very large number of products and a very large number of customers, okay? Because it averages the experience in a certain way across all the people looking at a wide diversity of product. If you're a company with a very small number of products, it has significant what's called AI skew, which is because the data set is small, meaning the number of products, it causes significant statistical skew in recommendations. That's one example of something we did very differently because small retailers ask us, "I've got a small population of products, a small set of customers.
I really need more accuracy, and the historical algorithms I've tried have got significant skew." What's the second kind of thing that we do? Imagine your historical recommendations had a failure in introducing something called novelty. What's novelty? Imagine your algorithm was based on what's everybody buying, and let me recommend based on what everybody's buying. The core problem you have is imagine you were Apple. You would have iPhone 7 is what everybody could buy. 8 was not available. 8 would always remain bottom of the tree because I'm recommending what everyone bought and what everyone used to buy was 7. I never pop 8 to the top. 8 always remains at the bottom. When you introduce a new product, as an example, you need to introduce this notion of novelty.
Novelty is, trust me, I know the algorithm is going to recommend this, but this is the product I want to put on top of. The third is to your question on human involvement. Human involvement, typically, a human comes in and says, "This is a new SKU. I need it to go at the top of the thing because a merchandising manager says, 'Trust me, I want to push that.'" Second example of human involvement is, "I've got a lot of product inventory over here. Even though it's not a popular product, I want everybody to see it because I need to get rid of that inventory." We allow humans to be assisted by algorithms, not to be replaced by algorithms.
Ladies and gentlemen, please welcome to the stage Senior Vice President of Oracle's Startup Ecosystem and Accelerator Programs, Reggie Bradford.
Thank you. Good afternoon. I'm Reggie Bradford. Really excited to be here. I'm new to this crowd, not new to Oracle. I'm going to take you guys through for the next sort of 25 minutes, how Oracle is approaching next-generation audiences including startup ecosystems, developers. I'm also talking about university programs. Before I get into that, I just want to kind of give you my background because I'm not a traditional enterprise executive. I'm actually a startup founder myself. I'm actually a three-time entrepreneur. My most recent startup was called Vitrue that Oracle generously acquired in 2012.
I had two startups before that I've dealt with many of you guys around, one in the communications industry and one in the health industry called WebMD. I bring a particular brand of experience around how do startups think about corporates and accelerators. I wanted to do something that would create value that I thought that startups would see. Just a typical safe harbor statement. When I was asked about nine months ago to approach this initiative, what I focus on is how do we take and create leverage in this incredible ecosystem we have called Oracle in 185 countries with 420,000 customers. How do we leverage that to create one unified voice?
Which has sort of been the challenge, I think, that I've kind of witnessed is we have great programs all over the world, but we don't really unify that in a single, cohesive experience across our developers, startups, universities, SMB. You've heard at OpenWorld all this week about all the initiatives, but think of this as a sort of a container to bring all that together, hopefully, under a unified umbrella, and do so on a global basis to capture the enormous opportunities that the cloud and globalization enables. At the heart of the future really is developers and startups that are really driving innovation around the world. You guys have seen that and heard that.
We fundamentally believe that enterprises that really rethink collaboration, both with other enterprises as well as with startups are those going to lead and be in the future and will be the ultimate innovators. Thomas talked about some of these massive transformations like blockchain. We fundamentally believe between our platform, our enterprise customers, and then also working with startups and developers, that's going to be the key to unlock value and continue to grow our position into those marketplaces in the future. If you could queue up the video, please. Really quick from the video. First of all, one key takeaway is that you don't have to be in Silicon Valley to develop a great startup. My three startups were all developed in Atlanta, Georgia.
I always felt this giant sucking sound to move to the Valley to bring my startups to market, but we were able to create great business there. There's startups developing all over the world. That's number one. Number two, and we want to be able to take advantage of that opportunity to help them and facilitate that. Number two is, I never joined an accelerator in any of my previous startups because I was like, "Where is the value in it?" These are kind of a dime a dozen in some respects. What we want to do is create an accelerator that I wanted to create an accelerator that I would actually want to join. When I was a startup founder, I was focused on how do I generate revenue and reference customers.
I started thinking to myself, Oracle absolutely knows how to sell to the enterprise. If we can help startups actually sell products into the enterprise, the world is our oyster. No one's been able to do that at scale around the globe. That's what we're focused on. First of all, in terms of the program vision, it's a next generation accelerator initiative. We want to reimagine the relationship between startups and the enterprise, fostering co-development and co-innovation. What I mean by that is, again, when I was in my startups, for example, in social marketing, we went to Apple, actually, and we went together and said, "What is your future view of how you see taking advantage of Facebook, Twitter, and Pinterest, and Instagram?" We said, "This is our product roadmap and our future. Let's develop that together." We see the same opportunity.
We've been working across our product development teams and our field teams to identify where are the gaps in the portfolio, where is opportunities for startups to innovate on top of our platform that we can also work with our enterprise customers, and really drive innovation across the industry and really push the boundaries on where we can all win in this together. Okay. A background on the program. It was actually initiated by Safra visiting with the Prime Minister of India back in, I believe it was April 2016. In Bangalore, we launched our first accelerator. We got it up really, frankly, probably too quickly, but we got it up and running in two months. Then we announced, with my participation, the global expansion in January 2017, so really only six months ago, okay.
We decided to expand from one location to nine total in the last six or nine months or so. It's been extremely busy. These are the markets that we just expanded and have launched in. As you can see, a big penetration across Europe, in the Middle East, Tel Aviv, Asia-Pacific, South America, and then we're also in the U.S. These are some of the stats that we'd like to just mention since we launched in January. We've had launch and promotional events in every single market. I've been to six continents in the last six months. We've spoken at numerous, not only Oracle events, but industry events. We've generated over 300 positive media articles around the globe about the initiative, with millions of social media mentions.
The program has been so well-received internally, we've got over 50 Oracle markets around the globe that now want to bring the initiative there, which is tremendous. We went live in all the eight markets, which people told me we'd never be able to do that quickly. We created these ecosystems of innovation across Oracle. Literally partnering with our innovation advisors, which is part of our field, to bring startups into enterprise customers. We're also ready for additional expansion phase 2, and a wider ecosystem strategy, which I'm going to be coming to you guys to talk about very shortly in terms of broadening this. These are just a couple examples of headlines.
My point here is not to tout that positive news stories turn into business, it is important and people are paying attention, and they see this as a very strong, important vehicle for us to look at repositioning Oracle, democratizing Oracle, contemporizing the brand into Millennials on down, et cetera, which I'll get into. Really quick on the program benefits. I picked this term up from Israel, but I've heard it before. It's a pay-it-forward model. Our philosophy is that there's plenty of venture capital in the marketplace. Any startup that has a product and a service with a decent track record can raise capital. That's not where the value is. We don't take equity. It's run by product development, so I actually work for Thomas. The reason why I think that's important is we have peer-to-peer relationship with the startups.
We really help them understand how they take advantage of platform as a service and infrastructure and really give them mentoring advice. We give them access to platform and infrastructure for free. We give them engagement opportunities, I mentioned, across the portfolio of customers. Mentoring and advising. We have PR. We had 15 startups here this week. We give them co-working space, and then if they need it, we connect them with customers. Really quickly on the program. It's a six-month program. Immersive experience facilitated by technical and business experts, both within Oracle as well as outside, angels and VCs and repeat entrepreneurs, et cetera. Selection process includes pitch and interviews. Then we have an alumni program once the startups graduate that they can continue to be part of Oracle Marketplace and connect with our customers and opportunities. Really quickly on who can apply.
We deliberately kept this open rather than focus specifically on a blockchain or cybersecurity because Oracle sells to just about every enterprise on the planet across every vertical. We wanted to see the diversity and be able to provide a tremendous array of startups that can address all different aspects of the portfolio. They have to have a large addressable market and a defined yet malleable product roadmap so that they want to be willing to work with us to develop further. It includes the team has to be together six months and a pitch and an overview. Cue up the video. Sorry.
Shippeo is a SaaS platform for tracking and managing road transportation.
Ambi Labs is a IoT company, we believe in using artificial intelligence to unlock the true power of all of that data.
Micro companies, people at the beginning of something, to work with large enterprises, to work with people who might become their customers to co-design new stuff in this rich, vibrant ecology is really important.
To build a foundation is to build a really kickass leadership team. We're highly focused on the big organizations that spend $hundreds of millions on innovation with millions of consumers under their belt.
Oracle itself and the organization there, but all the customers that they have and the links to people that they have, which gives us a global reach that we wouldn't otherwise have.
Oracle provides a gateway to customers, essentially.
At startup, it's quite hard to meet an enterprise in France. Much easier in the U.S. You can meet everyone. In France, if you are not backed up or supported by Oracle, you can wait long time, and as a startup, we don't have time.
Oracle has quite a big presence in Bristol. I know the people here. I've talked to them. They visited where I work. That's a good aspect too. To have a company that has a physical presence here is very important.
For us, it's very important to be enterprise-grade as soon as possible. To deploy all what's needed to do so as quick as possible, and Oracle can help us to do so.
The future is now.
Really quick. We had almost 3,000 startups apply for the first 40 slots. It's harder to get into Stanford University than it is to get in our program. The reason I bring that up is because We were able to attract much later stage startups that most of them have a substantial business model developed. One has 1,000 customers, for example, an SMB who wants to attract the enterprise. These are just a few examples. We brought 15 of them here to Oracle OpenWorld. They span blockchain, HCM, transportation, logistics, hotel, hospitality, CX, across the entire portfolio. Okay? The last thing is just the partners and alliances. One of the things that we're doing around Oracle is focusing on reimagining the workplace.
We're working with WeWork, for example, in several of our locations to think about space for the startups and how we collaborate together. These are local partners all across the globe. Again, we've just put this together, again, in a very short period of time. We've been developing very strong relationships. Okay? I'm going to now shift to universities and obviously, college students. Again, thinking about imagining the future, positioning ourselves for the Millennipreneur on down, which is the young adult that wants to maybe quit college and start a company. There's three things that we're focused on here, is to get cloud into the hands of very young adults, and gain mind share against the next generation of IT leaders and founders. Number 2 is this also applies to workforce.
Oracle being a place that's attractive for young adults as they're at engineering schools or whatever the case might be, to have Oracle be part of the agenda and then want to come work in here. Then three, I think, again, on the pay it forward side is, and I know Safra is a big proponent of this, is the providing activities and philanthropic activities, but really giving them a vehicle so that we can help provide higher education students computing resources to develop their skills in the marketplace, again, using cloud, and then educators, so they can do curriculum and create opportunities, again, built on the same technology that we use. The program benefits for cloud services, they get access to any combination of a wide selection of metered services in terms of infrastructure as a service and platform offering.
We have massive open online courses, attend courses with recorded lectures, so again, giving them content, and then community forums that they can communicate and connect together. We launched this, I would call it in a beta, with a handful of universities around the globe, some of the most prestigious, including Stanford here in town. We've got a tremendous pipeline of universities that, as we develop this, again, think of it similar to the startup ecosystem. Start with a small cohort, a few, and prove ourselves, get traction and a track record, understand what works and what doesn't, and then look to expand that. We've got a video to show real quick.
When we first started this project, the four of us, none of us had experience working with Oracle Stack. It was really an eye-opening experience. Then I think with Oracle, we've just been really lucky to have the team at Oracle backing us be so responsive and so great to work with. The Oracle tech stack gives us access to just so many tools we couldn't have possibly imagined to begin with.
When we initially started using Oracle services, we picked a lot of our technology stack, and it was pretty seamless to say, "Okay, well, given that you've already decided to write this in Flask and that you want to use MySQL instead of this, we can make that work across this set of specific solutions." Which is great to know that there's the flexibility of whatever technical decisions we make, we can still implement it on the service.
Just to summarize with those first two initiatives before I go to the third. This is about really awareness. I was with those students earlier this week. They're on the OpenWorld from Stanford, short trip. It's really not that we have a bad reputation with students or startups. We don't have the awareness that we need. It's about making available to those either startups or those university students. Once they get a taste, they start running projects. There's a tremendous change in their hearts about wanting to work with us. There's perception because I think we've been in the enterprise space so long now with platform infrastructure. We've got a product suite available to two people in a garage on up. Super excited about the traction and what we've seen so far.
Just for the sake of time, the third area I'm now going to talk about is developers. As you all know, with Oracle, developers have been at the heart of Oracle for years and years and years. This is not a new thing. What we're focused on now is a retool, reboot, extending the capabilities, and continuing to provide developers with a strategic approach, outreach, and engagement programs. Goal is to increase the awareness with developers around the Oracle Cloud. Winning the hearts and minds of developers versus other platforms, continuing to change the perception that might be out there with Oracle as a cool, contemporary, hip company, easy to use, and then continue to drive adoption, obviously, to cloud. First is a developer champion program.
Basically, if you're familiar with the term influencers, that's basically what this is identifying and working with who are sort of the leading edge folks out there that influence a lot of people on social media in the developer community. Experts in modern development such as microservices, containers, DevOps, machine learning, IoT, SQL, NoSQL databases. They contribute substantially to open-source projects. They're authors on contemporary development projects and approaches, they speak at lots of events, and we've brought them into our events. They draw a big crowd. These are some of the developer champions that we have created relationships with. I'm not going to go through them right now, consider them to be leaders in their space and influencers in the marketplace. These are some of the areas I mentioned that the champions are focused on. Their expertise area is around open-source.
I mentioned already IoT, big data. Thomas had talked earlier, machine learning and AI, serverless computing, database, Java, JavaScript, et cetera. This is our website interface for the developers, the portal, if you will, that they can access and connect with. The developers build projects, create capabilities. For this group, for us, it's not just about the platform and capabilities that we enable, making it easy to use, it's really about creating content, and planning that will enable them to build projects. Content is key around containers, microservices, and APIs, how to take advantage of those, making it accessible for them around AI and machine learning, developer tools and low-code programs, giving them access to languages in multiple areas, making it focused on education and sharing and really hands-on demonstration, making it quirky and nerdy. These guys are a little bit different, which is fine.
It's not just exclusive for us. We offer programs that all these programs developers can use, startups can use, but they can also work with other platforms. That's not really what it's about. It's about mind share. We revamped and launched our code event series, which is basically our roadshow to go out in the marketplace. We went to 20 cities in the last 12 months or so. We had over 10,000 developers that attended and another 350,000 participated online. That was a very good success in terms of the reboot in those markets. We try to go similar to the startup ecosystem. We go where the market is. These are some of the examples of cities that we've gone.
Just like where we go with the Accelerator, trying to line up where we have data centers and we have existing infrastructure and resources so we can go into a market and have a very strong presence and point of view and position across North America, EMEA, Asia-Pacific, and Latin America. These are just some photos of the events. You can see it's obviously not as formal as this group. It's probably a good thing.
Yeah, it's t-shirts and blue jeans and hackathons and a lot of fun type of activities where developers. Again, we see this playing extremely well around some of those big tentpole future areas Thomas talked about around getting developers, startups, our developers, and enterprise customers to work on, for example, a hackathon over the weekend to solve a real problem that exists in the education industry or with blockchain or whatever the case in the marketplace. We see tremendous potential here over the next couple of years to continue to extend and enhance Oracle's position in the marketplace with developers. These are some of the other programs that we offer, again, about education. MOOCs, which are massive online, open online courses. They're all for free. Again, providing value back to the community. A code online webcast and DevLive interviews. Again, sort of webinars.
A developer newsletter with monthly updates. Oracle Zip Labs, which is hands-on labs at conferences. We have a legends and luminary speaker series for developers. Obviously, across all these programs, we have a very tremendous, robust presence. Steve Miranda mentioned it around social media, but connecting and engaging with audiences across the globe in a real-time basis. There's 4 billion people on messaging platforms now around the world, so with mobile devices. Being able to connect and engage with them on the medium that particularly young adults want to be able to communicate with is extremely important and a big part of the developer outreach and engagement. Just to sort of summarize, because I've got to wrap up, but again, think about this as sort of a unified platform across all these different services.
One innovation across Oracle, leveraging the capabilities of the cloud with our enterprise expertise, the global footprint and capability that's unmatched in the industry, connecting that with startup ingenuity that really is a win-win for us and our customers and the startups and developers and university students. We think this is a tremendous recipe for success. I think finally, it also keeps us on our toes
We're moving very fast. Fortunately, I've had super support to launch these programs in six months or less. I think is a testament to the organization and some of the changes that you guys have hopefully seen in terms of, Doug's going to be talking about right now around buying experience and other aspects of this. This is sort of the tip of the spear about how we're going to be super competitive and continue to position Oracle for the next 40 years. Thanks very much.
Ladies and gentlemen, please welcome to the stage, Executive Vice President, Chief of Staff, and Head of Corporate Development, Doug Kehring.
Well, I got the longest title. That's the one thing I can take credit for here. You guys might remember I was here last year. I'm here again, I didn't do anything too badly. We're going to transition a little bit from the products and technology side to the more of the business and operational side of things. I did want to spend a few minutes here before we turn to Mark and Safra, giving you a little bit more of the inside baseball of what we've been doing on our own internal transformation. It really is amazing how far we've come when the questions you guys asked of the panel and to hear them say it from their own words about how Oracle has really transformed in terms of how we touch a customer.
It's really important because, I'm not going to repeat the speech from last year, I'll just highlight again. That we're moving from being a products company to a services company. Obviously, intellectual property underpins everything we do. As you move to that services side of things, it's no different than when you think about Airbnb, or you think about Lyft or any other company that's obviously disrupting an industry where it's really about a superior experience. When we think about delivering cloud as a service, the ultimate experience is going to be the predictor about whether we're long-term successful in this business. Let me start with just highlighting maybe the three pillars of our cloud strategy, because I think this helps you understand how we think about it internally. Clearly, technology has always been the driver for what differentiates ourselves.
Really, as we move to the cloud, it's these next two pillars that have become more important. I'm going to give you a little bit more color about the customer experience, and I'm going to introduce about what we're doing from an employee experience. Employees are now front and center of everything we do, simply because as you deliver a service, it's those individuals across our company that are touching customers every day. To the degree that they are motivated and highly engaged, they can deliver a much better customer experience than those that aren't. We think about those pillars. Obviously, on the technology side, it's really important that we deliver the right solutions that exceed customer expectations. You've heard from Thomas and from Steve about what we've been doing there. It's been an amazing amount of transformation on that side.
In terms of the best customer experience, the reality is, we're not going to be successful in the cloud unless our customers are successful. These are long-term relationships, just as the customer panel had described, the entirety of that customer experience and what we do to enable them to be successful will determine whether they stay, and they renew, and they keep doing the things that will enable us to expand our relationship with them. Finally, on that employee experience, that idea that we need to attract the best talent and make sure we retain that talent long term. It's a lot more competitive, as I think everyone knows, than it's ever been, right?
The competition for talent and technology, given how much money's available and how many startups and other companies are being funded, it's critical that we find new ways for Oracle to differentiate ourselves when it comes to our people. I'm going to get into the customer experience, but as I lay this out, I want you to think about our transformation along three different categories. One is people, two is processes, and third is platform. These are things that are fundamental to what we do when I go around internally inside the company and talk about how are we going to do something differently. If you're a startup, it's great because you start from scratch.
You don't have anything legacy to have to deal with, versus when you're an established organization, you have all of the, "This is the way we used to do it" that you have to overcome. When we think about overcoming, it's doing across those three areas. It's who are the people and how are you organizing in order to succeed? What are the processes? How do we reimagine and reinvent the way we do business, like the Accelerated Buying Experience? Third, what do you do it on? Which is obviously, we're going to talk a lot about how we leverage our cloud platform in order for us to have the agility to update our transformation that much more quickly. We move to the customer experience.
Again, last year we highlighted this, but I'll explain these 3 categories because as we've spent more time on this, it's evolved a little bit, and it's become a lot crisper. We think about a customer experience along the entirety of the journey, right? From the moment they're thinking about purchasing a cloud service all the way through to the renewal. That's the entire journey that they're going through. We have to make sure that it's a superior experience on each phase of it. The first phase is the discovery process. It's the idea of, "Gee, I'm interested in the cloud service. How do I find out what Oracle offers?
How do I try it out, pilot it, do things like that, where I'm discovering what it is that Oracle is offering in such a way that I'm prepared to purchase it?" In a lot of sense, that discovery process is something that often does not involve our sales reps, right? It's just like with our mobile applications on an iPhone. We don't go into the Apple Store in order to find out which of our next applications we want to purchase. We do the website searches. We look online. We do lots of things in order to discover what it is. We want to have that same rich experience in our ecosystem so that our prospective customers can find the Oracle Cloud services.
It moves to the cloud ordering process, in which, obviously, they're going to purchase from us, and part of the purchasing is to make that quick and easy. Finally, it's the utilization. The moment that they've made the purchase, what's that installation look like all the way through the renewal? How rich is that so that they want to renew it and expand it? The entirety of that process is really built around these ideas at the bottom, which is make it fast, make it simple, make it intuitive, and make it flexible. Built around the Oracle Cloud.
Whether it's discovery, using Eloqua to help us understand when folks are going through that process of analyzing and are ready to purchase, or the CPQ Cloud when we figure out what they want to purchase and when, all the way to the cloud utilization, where we're tracking them from a cloud service standpoint to understand what issues they're having or what bots we can put out there in order to push more information to them. Let me just highlight where we've come on the Accelerated Buying Experience. If you recall last year, I talked about the experience being one in which we want to target 90% of our cloud purchases. It's a simple idea of, with the customer, you determine what their cloud needs are once they've done all their discovery, you review the order with them, and then you click to accept it.
How have we done? The target has been reached. We're now almost two years into this, and we've reached our 90% target of our overall cloud transactions going through this Accelerated Buying Experience. In fact, in the last year-over-year, the volume of these transactions are up 3X. The deal velocity, which is how quickly we're getting these transactions done, has increased 50% since the launch of the experience. Ironically, it's costing us less, which we always knew would be the outcome of this because that's what cloud should do, but we need less people in order to deliver this experience than we ever had. It wasn't why we started it, but it's how it's come about. We don't end with that, which is all this time, we're continuing to identify ways to improve the experience.
We're in the process of rolling out the next version of our CPQ that's all about for the salesperson, how do we make it even more intuitive for them to do something? We're going to roll this video, and you're going to see an example of how a sales rep goes through a process of doing an order. Go ahead.
With CPQ Jet, sales teams can sell Oracle Cloud services as easily as one, two, three. First, the rep chooses the services that the customer needs. Next, he or she indicates the type of discounting required. Third, the sales rep selects the applicable agreement. Once the ordering document is generated, the rep sends it to the customer for review and click-through acceptance. Once accepted, the order is automatically booked, kicking off the activation process.
That's it. That's how you get an order done as a salesperson. Of course, if you have some options you might need approval with, we might kick off an approval process internally. A majority of the Oracle Cloud transactions don't require any additional approvals now, and that's the entirety of the process they go through. It's a very simple, intuitive process of going from screen to screen that helps them walk through how to pick what they need and how to get it ordered and sent to the customer for completion. Oh, sorry. Can you go back one slide? Let me introduce a new phase of our customer experience that we've been working on since we last met. This is around the cloud utilization side of the house, what we call the Superior Ownership Experience.
The Superior Ownership Experience is all about what does the customer feel as they go through the usage of the cloud service, all the way from how do I activate and get my cloud service up and running? How do my users utilize it? How do I continue to succeed and enable me to expand my usage? We think about it around these three pillars. The first is around rapid activation. How do we enable that quick use of cloud services? We spend a lot of our time trying to identify ways to make it repeatable, make it so that every customer that comes in can utilize the experiences of the last cloud customer so that they can get their cloud service up and running faster.
The second is, now I've got it deployed, how are my users learning how to use the cloud service, and how do I deal with issues that they experience just from a knowledge standpoint? This is where we're providing proactive journeys, where folks, again, through an intuitive process, can figure out how to utilize the cloud services more easily. Finally, this notion of easy self-service is really around the idea of what's my view look like as an administrator? How am I understanding all of the cloud services that my organization has and whether my users are experiencing any complexities that we might have to talk to Oracle about in terms of training and other things? This is them having a full dashboard view of what we have in place.
We're going to show you a video now of what it's like as you come in as a new user to the Oracle Cloud Service. After you've made the purchase, now you're going in to get started, this is going to be something that's focused around our infrastructure as a service. If we could roll that next video.
This video, Oracle is sending you a welcome email. When the email arrives, open it and scroll down to the Access Details section. Note the username, password, and the name of your Oracle Cloud account. Click Get Started with Oracle Cloud. Sign in to your new Oracle Cloud account with the username and password in the welcome email. After you pass through security, you're asked to reset your temporary password. This ensures that your Oracle account and your data will be safe and secure. The first thing you'll see is the Guided Journey page. It's like an information desk. Select a task you want to perform, and we'll ask you a few questions so we can guide you to the right place. You can visit this page at any time. For now, let's make our way to the My Services dashboard and have a look around at your new account.
The dashboard is like a terminal map. To see all the services, click Customize Dashboard. Scan the list of services and make sure the ones you want to use are set to Show.
Okay. That gives you a sense of the type of things that our customers are seeing as they order these cloud services. Again, it's really important that we push this type of help to customers. It's no longer about them having to pull from us, where they need to contact us and ask for things. It's these type of information pushes that we put out there, they're also simple, they're easy to use and discover. Around this whole superior ownership experience, there's other things we're doing, like communities. It's the idea that I'm a user of ERP, what about all the other customers that are using cloud ERP at Oracle? How do I learn from them in terms of what they've experienced and how to do things differently? Again, create communities, create forums for collaboration.
All of that is helping our teams push the assistance out to customers so that they have a great, rich experience, then the likelihood of renew goes up. When we think about it from a business and financial standpoint, the impetus of the value has now shifted from that upfront sale to that renewal. Rather than just selling to a customer in the traditional on-premise format of the license revenue and letting the customer go discover what they're going to discover with our products and knowing that we've gotten the vast majority of the value out of them, it's now shifted to that renewal process.
The superior ownership experience has become critical to enabling us to have that renewal be much more simple and much more likely to occur, along with giving the opportunity for folks like the customer success managers, who are in with the customers, seeing what their experience is, and knowing the type of things that they'd like to expand into so that we can generate new sales opportunities. When you turn to it and say, "Well, how are we doing on that front?" Again, it's something I didn't talk about last year, something we've been working on, though, for a little while as well. These are a couple different metrics that we track when we do our customer success interviews where we send out different questionnaires to folks and continue to get feedback directly from customers.
You can see things like, is Oracle proactively providing the help I need? Again, up 18% year-over-year. Am I getting the business value I expect from the Oracle Cloud? Up 27%. Would I recommend the Oracle Cloud to another? At an all-time high from when we started doing this tracking. This is a continuous vicious cycle of going out to customers, getting their feedback, incorporating into our processes, continuing to improve it on a constant basis with that idea that those customer sat type studies are the best indicator of where they're going to be from a renewal process when we get to the financial end of the cloud. That's a little bit that I described is all about what's the customer experience look like. I just want to finish my discussion today talking about the employee experience.
I started off the session talking about why that was important to us. Quite frankly, the way I've looked at it is, it's the same idea for an employee as it is for a customer, as you can imagine, which is ultimately, they have to have a rich experience and wake up and be extremely engaged with the product in order for a customer to renew or an employee to stay with your business. We at Oracle have been going through the same process of understanding, how do we become a great company to do business with from an employee perspective, as we have from a customer perspective? Similar to what I described as the customer experience, it's the same idea for an employee.
What's the journey like for any of us when we come on board from when I'm interested in being part of Oracle to when I become an employee, to when I start to become a contributor, all the way to when I'm a champion, I'm a leader, I'm an executive, I've promoted myself through? These are the three pillars that we spend a lot of time understanding how do we improve that experience so that employees are going to recommend to other prospective employees why Oracle is a great place to work, as well as how they interact with the customers so that the customers, again, feel a positive experience with Oracle. How do we approach it? We approach this, and Mark's been a huge advocate for how we think about this internally, which is around three ideas of listen, evaluate, and act.
Listen is the idea of going and surveying our employee base just like you would survey a customer base. As you bring that information in, you have to evaluate it, take the analytics, unpack it, understand what's important to employees, and how to change their experience just as we did with the customers. Obviously, act is the implementation of that. It's amazing, as you probably could have guessed, if the experience for a customer is very difficult, we were just the same for the employees. We knew how to make life hard at Oracle in order to get things done, like how do I hire somebody? How do I get a purchase order done? Those sort of ideas. We continue to spend a lot of time thinking about how do I simplify the process.
How do we remove complexity out of Oracle to make things go faster? How do I streamline approvals, take people out who aren't adding value, and how do I just generally improve the user experience so folks feel a richer outcome? Again, this is important for you to know, not because you think, well, Jesus, you should do this anyway, but more from the perspective of freeing up employee time. Make them more productive, make them more engaged. All of that contributes to what we believe is a better financial outcome for the company. I'll just, in the last few minutes, just highlight one of the areas we're spending time on and have launched. It's called the Accelerated Hiring Experience. The Accelerated Hiring Experience was this idea of how does Oracle go about getting offers out to candidates and getting them onboarded?
Again, it was one of these things where we did needle in a haystack versus not. One of the examples is, if you want to come to Oracle, and we're going to extend you a job offer, we had to make sure you cleared a background check. As most people know, this is not going to be that many people who don't clear background checks. It's, as they say, the needle in a haystack. Instead of giving somebody an offer and then doing the background check in the process of onboarding, before they started, we waited to send the offer out until the background check had been completed. This can be a multi-week process.
As a candidate, you're sitting there and you've got one job offer from somebody who just issued it to you because they're going to make your start date contingent on the background check. Oracle says, "We're not even going to give you an offer until you complete the background check." Ours was much riskier. As a result, when you weigh the two potentials, a lot of people chose the other offer and went forward. Weren't willing to wait for Oracle to conduct its process, in which case, most of the times, it didn't matter because almost everyone passes a background check. It was things like that where we changed the approach, and now we've reduced our offer time all the way from when you initiate the offer to when it gets out to an employee to a week.
Our onboarding of how employees feel about the experience they go through as they come on and become an employee has increased 50 percentage points, which is an astronomically high percentage, but that's how far we've come in a short period of time. Of course, as a result of all this, compared to the history, employees are feeling great. Our new candidates, as they've come in, feel great from an employee success and just a general goodwill standpoint. Again, these are things that are important to us. What we did about a few months ago, as we kicked off our new fiscal year, is we held a Global Employee Summit.
As part of that Global Employee Summit, where we talked about our strategy and what we're doing, is no different than what we did today, which is we want to convince investors why they should be long-term shareholders. We're going to do the same thing with our employees. Convince employees why you should remain long-term employees. As part of that, we asked folks, "Submit videos. Tell us how you feel and what your experience is with Oracle." I'm going to show a video to show some of the impact here of just how much folks are really positively feeling about the power of Oracle now and where we're headed from a cloud perspective. If you could start the video.
What makes me excited is our opportunity to really define the future of the cloud, the future of the software industry.
I want Oracle to become the global cloud provider.
We're a technology company, but we serve people.
It's all about the customer and what we at Oracle can do for him.
Whenever I see my customer success because of my work, I feel motivated to take on the next challenge.
I love the fact that Oracle has given us the opportunity to share our research with the world.
Building a team and building successful products is just so fulfilling.
We have a lot of autonomy and flexibility with our management team.
We're helping Oracle realize the promise of machine learning for our SaaS customers.
We have to change and stay on our toes all the time, so it's always really dynamic here.
What I love the most about working for Oracle is the chance to work with a team of highly motivated people all over the world.
Here, it's a fresh day, every day.
I know I can learn or implement anything I want.
I really love working at an organization that prioritizes learning.
You have the freedom to learn. You have freedom to innovate. You have ability to grow.
No matter what job role you're in or career level that you're at, everybody here has a voice.
Building the cloud.
This is a sea change of culture that's going on. Larry and Safra initiated this a few years ago. It takes a long time for this to trickle through, at times, for everyone to get the spirit of moving from a product to a service-led company. I think that's starting to give you proof points of how well it's going. The purpose of today was just give you an update on that transformation. With that, I think I'm complete. Thank you.
Ladies and gentlemen, please welcome to the stage Chief Executive Officer, Mark Hurd.
Yeah, I didn't expect a standing ovation. I can't think of a better way to end what an exciting week at Oracle OpenWorld than be with all of you. I'm just trying to get a little energy in the room. All right. To the stuff Doug described, I think there are a couple other metrics that might be worth sharing. Just to give you an idea, when we make an offer on the college campus today, and we obviously, I think most of you know, hire hundreds and hundreds of people every year. 87% of our offers are accepted, just to give you some context about how people want to join the company. Second, our time to fill a req. Doug talked about the time we take to onboard people. We've about halved the time to fill a req.
If it used to be 60 days, now 30 days. To the point Doug described about our surveys. We survey our people. We ask them, the point of listening, what's going well. One of the things we got three years ago was, it is really, really hard to join this company. Hard to figure out how to get things done, where to get things, et cetera. I have a big belief in engagement levels, because engagement levels, if you plot them over time, the higher the engagement level, the higher the productivity, and you can align the engagement levels really to financial performance. We had, like, a 30% favorable onboarding rate. Not good. You read all the verbatims.
We did a lot of work on the exact point Doug described, and we more than doubled it in two years. It's a big deal for our employees to speak to us, for us to listen, evaluate. Some of the things they say we're not going to fix, like, "We'd like to have the autonomy to spend $50 million." No. Some of the things they do say are extremely insightful, that we then go and fix and has a meaningful impact on the business, and it shows up in our numbers. All right. What I'm going to do is talk to you a little bit about a couple of subjects. I think first, let me reiterate what Ken said. Then I think what he said again. What he said yet again. I support every bit of that. All right.
Let's talk about SaaS. Can we go back one? This dynamic thing does not let me. No, I don't want to be on that chart. Next chart. There we go. All right. I'm going to talk to you a little bit about applications. I'm going to talk to you not just about SaaS, but I'm going to talk to you as much about the applications ecosystem. We talk about how the total application business is performing. I tried to do it in a way of, I get a lot of questions and tried to synthesize those questions, and hopefully what I do is give you more clarity, a little bit more insight into how the business performs. We'll talk a little bit about the database, not just our database business, but the database market.
We'll actually listen to a customer for a couple of minutes. Then Safra and I will do Q&A, as soon as we get through this. Okay. Let's talk about SaaS and SaaS revenue model. Let me try to take you through how the SaaS revenue model. I'm talking about how you think about revenue, how that works. We have a pipeline. I'm going to talk to you about the pipeline. The pipeline is really big. Okay? We have salespeople who take this pipeline, and they try and close deals. That is what we call ARR, annual recurring revenue. You would take that ARR, you would divide it by four because that would become a quarter. My only caveat on that is it takes some time. There's some latency between the receipt of an order and when it gets provisioned.
It's not a perfect model in terms of ARR showing up right away in terms of revenue. You add that to the revenue. There is a percent of the existing revenue base that is up for renewal. Okay? That number changes on a quarterly basis. You can imagine if we sold more into Q4 and the contracts were one year, that next year, you would have a lot of renewals to do. If it was three years, it would take three years for those renewals to occur. There is a percent every quarter of that revenue base up for renewal. Just to give you more clarity, each pillar has a little different renewal rate.
I can give you a renewal rate. But for example, three-year, four-year, five-year ERP ARR, I will still use the term ARR, have a really, really high renewal rate. Three-year, four-year HCM contracts have a really, really high renewal rate. One-year marketing campaigns, but still an ARR, have a lower renewal rate. The renewal rate is a blend of all of these. When you're growing, let's pretend our average contract length moves towards three years. Not completely three years, but that's roughly right. As you're growing, you'll have a little less than a third of that, provided you're growing, up for renewal. Then you'll get a percentage of that. Either look at that as a positive or negative. You take the part that's not renewed, you subtract that from the revenue base. Then you get the total SaaS revenue.
We'll do Q&A later, but I hope that was clear, because I get questions about this a lot, of can't I just add the ARR to the revenue? Doesn't that work? No. You have to do the renewal process as part of that as well. The contract lengths are continued, just as a trend line, continue to lengthen. Why do they lengthen? Because a bigger % of our ARR is now ERP. Bigger % of our ARR is now HCM, and those are stickier and stickier and stickier apps, ones that frankly, you just might view the renewal rate. By the way, let me be clear also about what we look at renewal rate. Renewal rate to us is literally a comparison of what was signed the first time compared to what renewed the second time. It does not include expansion. Okay?
If we renewed 80%, it would be to say it was 0.8 or 0.85 of the original thing we signed. An expansion would be in what we call new and expansion. That would show up in ARR. My objective for this chart is to make sure we're level set because there may be some of you who say, "I knew all this. I am no smarter than when he started." There are some who ask me about this frequently, and I just want to make sure we have absolute clarity and we're not confusing anybody. Okay. This chart is simply to say our pipeline has been big and it's getting bigger. This is, again, SaaS. Our pipeline is growing in the context of a lot of ERP, a lot of HCM, the contract lengths in our pipeline are increasing as well. This is SaaS revenue.
I'm going to talk to you about how the build goes in our applications revenue ecosystem. Our SaaS revenue has grown from 23 to 36 to 42 to 64. In terms of growth rate, it is at $1.2 billion and grew 38% last year-over-year, in terms of bookings. Our bookings last year, remember last year we made the statement, this is a subset, just to be clear, of the more than $2 billion worth of bookings we did last year. Last year around this time, we were prognosticating roughly around $2 billion. I got a lot of, "I don't think that's going to happen." It did. This is a portion of that. Okay? That's the ARR for SaaS. It drives application ecosystem revenue, this is the implication of what happens in the application ecosystem of total revenue.
What we've added in here is on-premise support, on-premise license. I'm a chart behind. I'm a chart behind. Sorry. I'm looking at the wrong chart. This is the chart I was on, I meant to be on. This is what's going on in our total applications ecosystem. This is, if you look at the chart from the other page, if you go back one chart, you see the 64%? Go back one chart if we can. That's the 64% from the previous chart. Go forward. That's the 64% in blue. The red is the addition of the on-premise support, on-premise license, that gives us 8% total revenue growth last year. In a second, I'll talk about Q1. I made the statement somewhere last year that this year our horizontal applications, viewed again as SaaS support and license, will grow roughly double digits this coming year.
I'm sure, I know I put the caveat on, if it's eight, nine, 10, 11, don't call me up and tell me I lied. This is roughly right in terms of what the revenue performance will be for this year. This is with SaaS now becoming a bigger and bigger part of the revenue pie. The market, depending on whose numbers, I tried to pull the consistent research, and I used IDC. This is not my number, 4%. I can find numbers at 2.8%, 3%. What I did was I didn't take one year. I took two years, tried to make sure I smoothed it out and got the four. I could have probably put three up here and had a research firm that would have agreed with me, I would have had to switch research firms as we go.
Either way you do it, we're gaining share. If you look at our Q1, which was obviously materially better than the eight, we gained material share in Q1. This will continue. This will continue in terms of us taking market share. I'm going to explain to you why in my next chart as well, we will continue to take market share in applications. This is what happens in our support business, we're way beyond the ideas of Mark and Safra's modeling or This is data over now thousands of transactions. This is what happens in our support business. Our support business is a good business. We get $1 and we got the high amount of gross margin. Many of you who've been shareholders in the company have known this model for a long time. Here's what happens when we convert it to SaaS.
Remember, we get $3 instead of $1 because we're also doing more work. It is in our data center. We do the hardware. Database revenue is in that number. Middleware revenue is in that number. We do a lot more. Labor is in that number. We get $3. Today, we get 67% gross margin on that, or make $2.01 instead of $0.95 in support. As our margins move towards 80% and above, we will make $2.40 or more as we convert our application support base to SaaS. We do this analysis every quarter in every pillar, we have very tight understanding throughout our management team of how we go to market on the conversion of support to SaaS. We want to convert our support base to SaaS. I get that question a lot. Are you concerned about that? No, I'm excited about that.
We are trying to drive this faster. That said, we are driving it at the pace that makes sense for the customer. We are not end-of-lifing our on-premise applications. I believe that to be a bad move for us. We want our customers to move at their pace. We give them choice. We also give them modularity, meaning you can move part of an app, you could move budget and planning as part of your ERP, move that to the cloud, convert it three to one, and leave your on-premise, your E-Business Suite on premise. You can do mix and match as you map your journey to the cloud. Today, in our SaaS revenue base, only about 35% of our current SaaS revenue is revenue that converted from our support base. The bulk of our SaaS revenue is from new logos.
That really hasn't changed that much over the course of the past couple of years. Okay? Let's go through a few numbers. ARR +48% CAGR between FY 2013 to FY 2017, was up 54% in Q1. I will not give you a forecast for Q2. I'm looking over. You don't want me to give a forecast for Q2? No? No. We have a big pipeline. Up 54% in Q1. Obviously, we grew in Q1 at a faster rate. I think I said this, in SaaS at accelerated rate than we did the year before. By the way, just to be clear, we're better at this than we were four years ago, three years ago, two years ago, and a year ago. Our sales teams are better. The participation rate of our salespeople is increasing. The quality of our references has increased.
I'm not trying to tell you we're terrific, but we're getting pretty good at this. By the way, I think I've told you all before, we had a lot to learn. We had a lot to learn. We've learned a lot, and it shows up in our performance. Non-GAAP SaaS revenue grew 68% in 2017. It was up 62% in Q1. The non-GAAP apps ecosystem that I just told you about that grew 18%, grew 18% in Q1. Okay? Again, that's a phenomena now of our SaaS revenue getting bigger and bigger as a percent of the total. Our support base, while in a bit of a decline, its biggest category of decline is moving to our SaaS. As a result, we get this level of growth. Again, I expect this market share gain in applications total market to continue.
Cloud short-term deferred revenue, you can see what happened. SaaS pipeline, the biggest base we've ever had as we go into the year and now we go into Q2. I just say, listen, the momentum I'm not going to take you through all this. I sort of said all this, but this applications business By the way, I will say as a percent of what I talk about when I see investors, I talk less about apps now than I used to, and it's probably because of these numbers laying out the way they have. This is a double-digit sort of billion-dollar business for us, all ecosystem included. This is a big deal, and we're now on offense, we're not on defense. This is not like it was four or five years ago. We're much better, much more skilled.
The references, I mean, the thought that we could sit here with a Bank of America moving its financials to I mean, I can go on and on with some of the references that we now have that we didn't have two or three years ago, and the quality of wins that we've got. The morale on our teams is high. These share gains are going to continue. Okay. I am going to move to talk a little bit about database. I do get more questions about database, and I want to share some numbers with you on database as well. I think we're going to get started with a video. What we've got is John Donovan, who is now the CEO of AT&T Communications, did a video for us. Obviously, we did some things, as we announced our deal with AT&T.
We can take some questions if you want about it later. John did a video with us, if we could roll, John. Let's go to that.
Long gone are the days of relying on traditional, hardware-intensive ways to build our network for this continued data explosion that's taking place on AT&T's network. Our network has seen a 250,000% increase in traffic come across it since only 2007. We see 168 petabytes of data travel over our network every average business day. To put that in perspective, at the end of 2015, we had only 114 petabytes a day. Our internal IT migration to the cloud is essential to not only stay on top of this demand, but continuing to deliver the best customer service and new products and services on an accelerated timeline. We're moving thousands of database and hundreds of petabytes into the cloud. Oracle is a leader in the cloud, and this collaboration aligns us to put together roadmaps that will be meaningful to AT&T and other Oracle customers.
We remain one of the most aggressive cloud and software companies in the world. Having a collaboration with a company where you have confidence you can do all of this together is a huge deal. Two key motivations in this new collaborative approach. First, it shift from our relationship from transactional to strategic, it accelerates our movement to the cloud. Working with Oracle, it allows us to rapidly increase the speed of our cloud migration across the enterprise while also improving cost structures. We're excited because this agreement gives AT&T global access to Oracle's cloud portfolio offerings, both in the public cloud and on AT&T's integrated cloud. This includes Oracle's infrastructure as a service, platform as a service, database as a service, and software as a service. AT&T is in the process of moving thousands of databases with hundreds of petabytes into the cloud.
Oracle is going to address our specific need. How do you tear down a massive database and regionally distribute it so you can be really fast in how you're managing your IT application changes that rest on top of this data? This will fundamentally change how AT&T will do things. We don't have to waste our energy and time now on each and every database trying to determine how we're going to migrate to this new architecture. Instead, we've worked closely with Oracle to build a world-class database roadmap to migrate an extremely large company from fixed databases to clouds. We're also looking at applications and within applications with capabilities like artificial intelligence and machine learning. Let me give you a couple of specific examples. AT&T becomes Oracle's first database as a service deployment on a customer premise.
This helps increase employee productivity, reduce IT costs, and provides new flexibility in how we implement software-as-a-service applications across our growing global enterprise. We also selected Oracle's Field Service Cloud to further optimize our ability to schedule and dispatch for more than 70,000 field technicians. We're able to combine the Field Service Cloud into our existing machine learning and big data capabilities, which will quickly increase the productivity, the on-time arrival, and the job duration accuracy of all of our field technicians. Take workforce management for our dispatch function. We have 70,000 trucks that we dispatch every single day. It's one of the largest fleets in America, and that can present administrative challenges. On the efficiency side of it, if we're rapid and nimble, we can change dispatch strategies tonight and have it implemented tomorrow.
We can start to think about not only what did we learn yesterday and how can we get smarter. We can bring in third-party data that allows you to do things like look at weather, neighborhood construction. Instead of using an average time to complete a job, now all of a sudden, your precision gets better and better and better, and our truck rolls become more efficient, and the company saves money, and this cycle continues every day. Our customers get faster, better service from us because of this migration. Our hope is that we'll significantly reduce the appointment window. It's one of those rare virtuous cycles. It's better for the technicians. It's better for the customers. There are three things. One is the liberation of data, just getting it to be usable to more people. The second is speed.
We believe that we should be able to innovate faster, in order to manifest that, you really have to materially speed up. The third thing is that we want to use data as a product. We want to move from the way we think of it, data generating, to data powering networks. It's subtle, but I think it's an important difference. This is new ground for everyone and should be viewed as one of the more exciting cloud-driven projects in front of both companies. We have a lot of work to accomplish in a short period of time. I know we have two of the world's best companies working together, and we're already seeing great progress in our short amount of time together in this new agreement. I look forward to hearing about the progress all along the way. Thank you very much.
Yeah. Thanks. That was a lot for John to do, and I really appreciate it. Let me just give you some stats about AT&T. John touched on what a big app it is, but it's not unlike many of our customers. There are 12,000 or so Oracle databases inside AT&T. A couple of exabytes worth of data across this enterprise, not quite. Huge. Inside that ecosystem, 500 databases are bigger than five, six terabytes. It's only 4% of the databases, but it has 70% plus of AT&T's data in 4% of those databases. The ability to modernize those, just the simple process of patching them. You want to go patch all that stuff? You want to take the company down and patch it? That's the kind of problem John's faced with. The only reason I wish he was here, I love the video.
The only reason I wish he was here, I know one of you would've said, "Did you consider something else?" You would've got a heck of a kick out of his answer. We'll get him out on the road shortly because this is a lot of work we've got going together, but it's exciting, and while it's big in scale, it's not unlike the situation that most of our customers face. All right. There are a few choices now, John touched on a couple of them, by which you can buy Oracle. We made this announcement of BYOL. BYOL again means bring your own license. I'm sure that was covered yet today. You can do it from on-premise to the cloud. Let me be clear, we don't care. We're nondenominational on how you buy, where you buy, and how you use.
We have lots of different ways to help you solve your problem. You can buy on-premise. You can move to the Oracle public cloud. You can bring your own license, as this chart describes, 2 paths. You can do it in the public cloud. You can do it in the private cloud. In the private cloud, we manage it for you. You can see Oracle managed in both boxes. The only difference in the 2 charts is the thing that says Oracle Cloud Data Center, customer data center. Your choice. Same stuff. Same OS. Same platform. Same version of the database. Same everything. Our labor. You're alleviating the risk of patching, of upgrading. We do the work. This is our database ecosystem, same way of looking at the applications market. I did this over 4 years. I am in a good mood today. Been a great week.
I don't want to call anybody, I get the question every now and then of, "You know, I think you guys are losing market share." I always find, "How did you come to that conclusion?" "Well, I looked at a thing from MongoDB, and they're going public, and they look like they got a bunch of revenue." I must say, I swear, I mean, you're kidding me with this stuff. I'm going to stop. I'm going to get back on target here. This is not thoughtful analytics. This is a big market. The market grows 3.5% per year. I've rounded up here to 4. Our 4-year CAGR, by the way, let me go through it again. On-premise support, on-premise license, Database as a Service is 5%. Let me compare 3.5%, I've rounded to 4, to 5%. That means we're gaining share.
That's how the math actually works. I'm not going to tell you again. That's how the math works. These are the players in the market. You can probably read Oracle. You might be able to read the next one and the next one, and you probably can't read the blizzard of players that come after that. This is not my chart. I just want to be clear. This is not from the Mark Hurd research firm. This is from IDC Research. It has names like Microsoft, and it's got Fujitsu. Something I find interesting is some of the names on here is actually Oracle reported by them. This is the reality of the market. The top 3 are still the top 3 from the year 2000. The market share of the 3 has changed 1% during that entire timeframe. Okay? These are the numbers.
By the way, I would love to have peeled apart like SAP, which is Sybase plus HANA, plus whatever else stuff they got in there, but this is an IDC report, I can't mess with it. This is the market. We have our biggest announcement, by the way, coming, I'll talk about that in a second. Why do customers stay with Oracle? You just heard it. I mean, it's just the best database. It's the fastest, it's the most scalable, it's the most secure, and it costs the least to run. There's no business case to justify. I get a lot of people say, "I hear," I always love these ones, "I hear people are moving off your database." I was like, "Who? Let's talk through them because those are the numbers." There is no business case.
When you get down to the point of what's the business case to justify the move, you may hear it, we don't. This is not what our people see in the marketplace. We have a big sales force, if you asked our sales force what are the issues they face with, they say. In terms of new workloads, this really becomes a lot more interesting. New workloads on where they're going to go. I'd argue we're gaining share in the most important announcement we've made in database was made Sunday by Larry.
If you wanted to see a field sales force jacked up about something, I used a joke about, if I put the value of this database into a note, I put it in the mouth of maybe even a dog, I had that dog run up to customers and hand them the note, I think I'd get business. I do. This is the way you motivate a sales org. I explained to them, the good news is, I have you guys. You can imagine how this went. This was a thrill a minute with these guys. Anyway, this has our sales force more jacked up than I've seen them in years. This is a group that, when you look at the numbers, is winning. You got to buy three features to make this a reality.
You got to buy multi-tenant, RAC, and to get the SLAs, you have to have Data Guard. Three features. You can bring your own license. It's automated. Automatically provision, upgrade, patch, tune while running. I don't know how much these words mean to you. To anybody in the database technical community, this is life. Most of our customers, if you don't have this data, when we release a patch and we implement it on our cloud, it will be 14 months later when that patch ripples its way through our user base. 14 months. We have customers, very large customers, who are very good, who've accelerated that process to be four or five months. Their CEOs say, "Four or five months, I'm exposed that you've got a patch that I haven't yet deployed through my entire Oracle base.
If I have a problem, whose risk is that? This changes that entire dynamic. That's as big a deal, I'm telling you today. I was at a meeting with one of the biggest banks in this country, with a CEO for an hour. We spent 30 minutes of the hour talking about patching. Patching and speed of patching, and what we could do to help them. I think you know we're guaranteeing at half the cost of Amazon. Not a hard guarantee for us to make. Reliability, as long as you have Data Guard, you get that SLA. We're gaining share. Our biggest release just made. Database ecosystem, we're up 6% in FY 2017, so that 5% I showed you is the four-year average. 6% is last year. That's the reality of what's going on inside our database ecosystem.
The stuff Reggie's doing, getting the new startup companies, our next release. I'm going to get you a little bit off of now just considering what is our DBaaS growth, but looking at this aggregated ecosystem, because you're going to see an impact on license sales as a result of what I've just described.
All right. Oh, Ken?
Hi. Joel Fishbein from BTIG. Just wanted to follow up on the proxy that was recently filed to talk about compensation. In there, you had some pretty interesting targets, long-term targets in terms of from both SaaS, IaaS, and PaaS, longer term models. I wanted to know if that's something that we can use to sort of benchmark you against and use to track against going forward.
Well, I can tell you that my husband and his wife will be benchmarking us, as will the children. Since it's their inheritance. I'll tell you, these are targets we're going for. We don't earn anything if we don't get there. We are very, very serious. I'm sure many of you have walked around OpenWorld. You saw customers here who we let you ask them the questions. You must see that the last piece of our strategy has now That rubber has hit the road, which is we are now ready to take production Oracle workloads into our cloud. That's why we've been so optimistic. That's why these things are in our comp plans. Because we are now ready. We'll be, surprise, an overnight success, frankly, 40 years in the making. The technology that Mark was talking about, that Larry announced on Sunday, means it's time.
It's time to bring the most important data of any company, and all that data is in Oracle databases. As you could see in market share, we're basically the market share of everybody else combined, okay? That number goes up every year, not down. Why is that? Because companies want their most important data in Oracle because it's way better. Now, our comp plan goes ahead and shows you what we think we can do I'm going to tone it down. I'll run out of energy.
I thought it was pretty good.
Did you want to answer that one?
No. I would say one thing that would be, just to add to it is, it isn't just the two of us, it's our board, it's Larry. When we put that out, this isn't just really a comp plan. This is us saying, "This is where we think we're going to drive the company." When you looked at each line, many of those lines sort of, I think, the way we're worked at is we can't just go crazy on one line instead of another. It's a set of things that work together to what we think at the time we achieve them. By the way, there's nothing that says we can't achieve them quickly. I also wouldn't take the timeline to say there's some of those things in there that we think we can do faster than the timeline you see laid out in the proxy.
We're motivated to go get these things.
Back over here.
Okay, thank you. Brian White, Drexel. On the autonomous database, maybe you could give us a feel for how this is different than other database cycles. Should we expect market share gains here? Are you going to reach new customers? Is this going to drive other pieces of the Oracle platform? Just some type of guidance on that would be great.
Listen, I don't plan to prognosticate anything other than to tell you I think this is going to be the most popular thing we've done in years. Just in years. I wish I could show you. Again, I love the benchmark of the enthusiasm of a sales force because they're motivated by simple things. They're not motivated by very complicated stories. This is not only amazing technology, it's easy to articulate. It's easy to describe. Our customer base. Listen, typical CEOs. This is a problem that's changed in the market over the past two or three years. It used to be, you got hired and fired, generally speaking, on revenue performance, earnings performance, cash flow performance, stock price performance, all this stuff. Now you have the added problem of risk. Risk of something happening, this poor situation at Equifax. You can go down many others.
The fact of what this does for you that nobody else can do. Who else is going to do this for you? When you've got a new workload, unless it's a tinker toy workload that means nothing and you don't care if anybody steals it, why would you pay more money to have a workload that's unsecure? I don't know. Maybe you've got some workloads that make sense. I'd like to pay more to have an unsecure workload. This is going to be a lot of fun for us over the course of the next period of time. Do I think it's going to drive more growth? Yeah.
Let me talk to you about the way I look at Oracle Autonomous Database. We have all wanted to change a tire while we're on a freeway. That's what this does. We've been working on this not for the past two years, but that's why I say it's 40 years in the making. For those of you who followed us a long time, you remember when we started with Parallel Server, then RAC. You're running multiple databases. Let's say you've got four running, like four tires on a highway. Imagine you take one down for any reason. You slow down a bit, right? Imagine if you now take that one, patch it, bring it back down on the highway, and move to the next one. That technology, multi-tenancy, and Oracle Active Data Guard. Remember, each one thought of differently.
Those three together, and also using the experience that we gain in running thousands, hundreds of thousands, millions, ultimately, of databases, allow each customer to get the benefit of the whole. Patching and upgrading databases while they're hot. That's been the issue. Do we believe this is going to spark a lot of database sales? Yes. Are we absolutely sure that it will spark a lot of options? Yes, many in our customer base have RAC. Not that many have Oracle Active Data Guard. Very few have multi-tenancy. We expect those customers to do one of two things. They will either add to their ULAs or buy additional licenses so that they can come to Oracle Cloud. They will use PaaS, where they will rent those same features. We think this is very big. For those of you who have wondered, why are we so optimistic?
We've seen this coming. BYOL, bring your own license. I know a lot of you are hearing us talking about it, and Larry talked a little bit about it on the earnings call. Also the second piece of BYOL, which is Universal Credits, where our customers buy basically credits that they can burn down any way they want. They no longer have to say, "I know exactly what I want when I buy it, and that's all I can get." Now they have Universal Credits that get burned down as they need them. Again, immense amount of flexibility. We think this is going to increase our market share because we have done something that no other database and no other cloud service of a database, as far as I know, is even trying. Yes, we're very optimistic.
Hi, Kash Rangan of BofAML. Hi, right over here. Congrats, Safra and Mark. The conference vibes are definitely more positive than they've been the last couple of years. Certainly something is going on in the customer base that we're able to pick up. A question for you, Safra. One is, if you're in the beginning stages of a big cycle where everything moves to the cloud, particularly Oracle Databases, is this not a very capital-intensive, lower margin proposition? Maybe it's not. Curious how we should think about your earlier vision for double-digit earnings growth rate, not only this year, but presumably into the future, in light of what looks to be a lot of capital that you would need. One for you, Mark. The product portfolio seems to be widening every year that we come here.
How are you changing or not changing the go-to-market and upgrading your sales force to be able to sell what looks to be an increasingly complex set of products? Thank you.
Okay. We've talked about this before. We've talked about the fact that we control virtually our entire supply chain. We, in fact, make our own systems. We also have the benefit of having an infrastructure that is optimized for our own workloads. As a result, we have an extremely performant infrastructure. Is it true that the more successful we are, the more capital we will have to deploy? Yes. Because of our economies of scale and our optimized infrastructure, I think we're going to make it extremely profitably. While, by the way, let me remind you, to the extent that the customer is bringing their own license as part of that ecosystem Mark was talking about, we still get support, and they buy more licenses.
We're very optimistic in our ability to manage our investment, even though it is to some extent ahead of my ability to recognize some of the cloud revenue. There's a lot of revenue coming in. I think it's pretty straightforward. You can even look just into our past, and you would've said, "Well, look at all those in the SaaS business." I think many of you wondered how we could possibly bring those kind of gross margins to the SaaS business. We had to build up. Remember, with SaaS, it's not only compute and storage, straight database. You have to run the databases. You have the middleware. You have a very large infrastructure. We were able, in that case, to make that very profitable. I think we start benefiting from those economies of scale.
It is true, we do make the capital investment often in advance of our ability to book the revenue from the cloud. In the platform side, we will get the benefit that I think many of you did not expect. You were thinking that in SaaS, like in SaaS, when you're no longer, let's say, using your PeopleSoft, you cancel your PeopleSoft support, now you're just buying SaaS, which is renting your licenses. With the platform, it's different, potentially. Many customers that already have large investments in the database and want the additional features or want more unlimited capabilities, they're going to buy more licenses. They're going to continue paying support, and they're going to deploy in our cloud.
That, I think, is an interesting point. You cannot compare everything in PaaS to SaaS for the reason that Safra described. The SaaS application support is binary. I'm either using HR on-premise, or I'm using it on the cloud. We may do some things to smooth it out as you transition, but it's a fairly binary exercise. Not so in database. Take AT&T as an example. AT&T will be using on-premise databases for years and years, even though they'll be using databases in the cloud as we go forward more and more. They will do that in the construct of a ULA and a license agreement, and that support will stay where it is and potentially may increase as they add options and bring those licenses to cloud. It's a different model in that respect.
In terms of the margin and capital you're describing, there is similarities. If you remember what we did in SaaS, and I can trust you we do, when we had a low gross margin, as we were building out our capacity several years ago. In Q4, I think our margins hit around 39% as we were deploying more capacity in start-up data centers. Once you plunk down that unused capacity, then from there, everything really is a percent of bookings as you add capacity going forward. I'm not going to add more capacity if I don't have customers. The analog of trying to take a Microsoft or an Amazon that doesn't have all the cogs that we do is a mistaken analog. We have a very different business model, a very different set of costs than they do.
As we go through scaling up, from there, it really is, to Safra's point, we'll pay more capital. When you look at an income statement, the expense gets offset With an offsetting booking that turns into revenue. To your question on the sales force, we haven't changed that much in our sales force, Kash. This year we made one change. We changed nothing this year in our applications sales force. We did not change many of our specialties in our tech ecosystem, but we did move what you would think of as our database rep and now call that a cloud platform rep. They are now remember, our principles are the same. We align our sales force by product, by buyer, by competitor. What changed over the past couple of years? Our competitor.
We've adjusted to align for that competitor, save that, we've really made no other significant go-to-market change. That is a change, though, because what we see now is lining up against cloud ecosystems, that would be a change we've made. Save that, everything else extremely stable. Just to give you one more stat, because I just want to make sure, because every now and then I get some email about, "I'm hearing you made a massive change in your sales force," which is just fake news. In the end, we had all of our territories, all of our comp plans, all of our training, all of our bosses done by, I don't know, June 10th, 11th, 12th. It's a stable sales force.
I would tell you, I'm coming out of this week, with a big percentage of them here, probably the most jacked up sales force across apps and tech that we've had in years. I hope you talk to them. I hope you talk to them while you're here.
What about it? That's where we're going.
Okay.
That's our plan. That's what we're planning on hitting. If we hit it, we hit it. If we don't, we're doing our best. We are very optimistic on having a great year.
Oh.
Hi. Hi, Sarah with Macquarie. How are you? I wanted to dig into that a little bit more on the BYOL. It seems to be probably the biggest area I'm getting incoming questions on. I know it's only been a couple of weeks. I do think that the impact on the financial model, if you could walk us through that a little bit more, Safra, would be really, really helpful.
Yeah. I think I tried to explain it. I'll try again. I don't know. Maybe I shouldn't explain it. Maybe you should explain it, because I already did.
Again, I'm not sure because I think I tried to explain it, but obviously I didn't do a very good job. As you know, it basically is you can now buy a license and you can bring it to the cloud.
That's it.
You buy the license. You have to procure the services in the cloud you need to run your license on. Think of it this way. Let's pretend you had an Oracle license and you said, "I don't want to run it on Oracle, but I want to run it on the cloud." I could bring my license and run it on insert name here cloud. I would then have to procure infrastructure from that cloud provider, some automation from that cloud provider, a suite of services, whatever they might be, in addition to the license that I'm bringing, and I would procure that from that cloud provider. I would then bring my license. I would continue to pay support to Oracle for that license. All we're saying here is you now have the exact same model with Oracle.
You're bringing that license, continuing to pay support for that license, and you're now procuring all of those services also from us. What you would do in the financial model to your question is you would keep support exactly where it was. You might increase your license if you wanted to buy those features as licenses and pay more support. Then you would add the services you procured from Oracle in the cloud, then you would add those together, and that would be the impact on the income statement. I don't know if that helped or you're more I helped confu Okay. Thank you, sir.
The alternative would have been to rent your license on PaaS. You would have canceled your database license support. You would have rented it. Our customers don't want to do that. Some want to rent it. That's fine.
You could. You can. We're going to say, fine, you can do that. We just don't think they want to pay twice.
Yeah.
We're trying to make it as simple as we can. This is the perfect analog. You can now take this bill of materials from this cloud provider, you can bring your license to that one, or you can bring it to us and compare our situation with our SLA, with our performance, our cost, our security against this provider. Let me just. If it isn't clear enough, we think we're going to win like every time. The impact on the income statement is those services from the cloud will then show up in our income statement as opposed to somebody else's. If a customer By the way, we're not dictating, again, I want to make sure it's clear. We're not telling the customer, you got to do it this way.
If the customer says, "I'm willing to give up the support and repurchase the license," so to speak, in a service in the cloud, God be with you.
Right.
Hi, guys. It's Adam Holt from MoffettNathanson.
Gosh. Hi, Adam.
How are you?
I'm good.
You look good.
Good to see you.
Um-
That's very nice of you. Thank you.
So-
Don't let us bother you.
A lot of great detail on the different elements of the model transitioning, if you will. If I could just boil it down to two simple questions. Throughout this shift, would you expect your core maintenance business to continue to grow? Over the next three to five years, would you expect the IaaS and PaaS business to grow faster than the SaaS business or vice versa?
Can I get the first one?
Sure.
Support's going to go up. Is that your question? Is support going to go up?
Yes.
Yes. We believe support goes up.
Okay.
What was the second question again? Is infrastructure going to grow faster than SaaS?
Yes.
Yes. Yeah. Listen, I don't always like the intramural competition, because I think we're going to continue to do really well in SaaS. Safra's always, because I go on and on about our apps business, I am so positive on our applications position. It's not easy for me. I'm probably not getting my enthusiasm across to you because we're winning, and our people feel it, and we see it in our pipeline, and we're getting better. Listen, I think I told this same crowd, or at least the majority were here three or four years ago. We're making a lot of mistakes here. Every day we're learning. We made a lot of mistakes, and we learned. In fact, Larry and I'd be on the phone sometimes.
He'd say, "Listen, this is how this works." He'd be like, "Jeez, never thought of that." Like, "Yeah, me neither." We learned these lessons the hard way. We learned it through acquisitions. We learned it from trying. We're past the learning phase, and we're on offense. I don't want to make the statement one that I think somehow I said infrastructure's growing and SaaS is not. I think our application business is going to gain significant market share. I believe at a point in time, we will pass SAP as the number 1 applications company in the world, bar none. I believe that to my core. Do I think we're going to sell more SaaS than [IaaS]? I do. I think we're going to sell more infrastructure, yeah, because of the very nature of the value proposition that we just described with Sarah.
It's too attractive of a value proposition over time. Just remember one thing about the support point. The support point in apps is still a different point to a degree than it is in tech. Because in apps, if you could ask me to wave a magic wand, I would move it to SaaS, and we're going to make a hell of a lot of money in doing that. I think that'll take time, though. If we look at our pipe, I don't think that happens next week. I don't think it happens next month. I don't think it happens next year. I think you will see a continual flow of our customers as we get to every new release, you will see a percentage of our user base moving more and more to SaaS, and that's not a bad thing. That's a good thing for us.
Hey, guys. Alex Zukin with Piper Jaffray. Maybe to your last point, Mark, how do you expect the slope of that curve in terms of new customers versus conversion to trend over the next one to two years? With respect to BYOL, you're giving customers a significant amount of choice, and I'm wondering how do you expect them to deploy public cloud versus Cloud@Customer?
Yeah. A couple of points. I think first, on your point about this percentage in our cloud, I think we're going to continue to attract a lot of upper mid-market new customers. Most of that market sits on the application platform of others. We talk a lot about the names you all know and hear about. Almost 50% of the market in apps is others. What's Infor? They have like, I don't know, 10 brands that are underneath that umbrella, and it's a blizzard of brands just in there, plus a whole bunch of other brands. Most of those are very old legacy applications. I think all of those move. It might be two, it might be three, it might be four. All of those will move. Who's in the best shape to get those? We have an entire automated fleet of ground-up modern applications.
I think we continue with a pretty strong new logo cadence. Do I think our customers will move? Yeah, I think they'll move, but I think you'll see our customers moving, maybe not so much moving everything at one time. You'll see a lot of our big customers moving part of an application at a time. A lot of people, like for example, in ERP, moving budgeting and planning. A lot of people in our base moving procurement, and moving it a piece at a time as they roll out. The truth of the matter is, I'm not trying to lock it into a certain %. We're trying to get into a position where we get the breadth of awareness. Our reference base, particularly in this country, is now so deep on financials. We're at 5,000 customers now, roughly speaking.
The depth of our references is such that we had a mid-market roundtable. To some degree, I wish you guys could see some of these roundtables that we go into with customers. The thought that they can now not do the work, and they can pass that work to us. Are you kidding. We had the founder of Lyft in my keynote. No data center, no IT. This is it. Oracle, that's what we do. This is what's in the market. I think we'll continue a very strong new logo cadence. You had one more question as well.
The BYOL.
My belief is what Safra said. You've been trying to get us to repeat it over and over again. What I think customers will do, on average, is more of them will say, "Let me buy multi-tenant. Let me buy Data Guard." Half our base, a little less, has RAC. Be a lot of people, rest of the base buying RAC. Then I want to take that with the cloud. I need to procure the services to go run. That's what I think the average persona will be of the customer that'll do it. But as we just described, you'll see all sorts of stuff that will occur. I think customers will go while trying to get this SLA. The fact I can buy Data Guard and get this SLA. What? Are you kidding? I think this will be a no-brainer.
The second that I can get RAC and I can get multi-tenant and now I can get You're telling me you're now going to patch all of this for me? Do I think some people will choose to do Cloud@Customer? That'll depend a little bit on geography, a little bit on industry. As you get to more of these regulated industries where there's a regulator that's got a thesis on how IT should work, you'll see more of this Cloud@Customer. Frankly, that's great for us, too.
Thanks. Kirk Materne with Evercore ISI. Mark, just to double-click a little bit on the ERP and financials in particular. It's been a very fragmented market for a long time, and it's been very difficult to get a customer to move off a financial system onto another alternative. What's going on that's sort of spurring customers to, I guess, re-look at Oracle or reinvestigate what you all have going on versus their prior decision in the last cycle, on the on-premise cycle? Just really quickly for Safra, if I can, we're pushing up against prior peak margins pretty soon. A lot of people would push back and say, in a cloud business, you can't get sort of past where you guys were. Just sort of a quick response to that if that's someone's sort of impression. Thanks.
Sure. I'll start. It turns out, like in financials, when you talk to CFOs, it turns out that CFOs actually would love to spend less money. That turns out to be actually a motivator for them. Second, getting out of the systems integrator hook is a really big deal. Most of these customers haven't had a new feature in years and years and years. Their ability to get mobility, their ability to get reporting, severely limited the opportunity to apply anything like machine learning, any of the new features coming, not present in their existing system. Most of those systems have been customized.
Any of those customers that are facing a release, an upgrade, that then have to repay for the same customizations they had to pay for before, the opportunity to have cost avoidance, the ability now to move into a modern ground-up application where I get integrated analytics, and I get somebody else doing the features for me, and it costs less all at the same time, very attractive. I would tell you that's probably something that's excited us. If I dialed it back three or four years, I think the speed. Jeff's here. Jeff makes a lot of sales calls with us, too. The speed and the number of people talking about this probably as high as we've seen it. Those are the core drivers.
I think you're going to have to watch this play out. The economies of scale that we bring to this are very profound, frankly. You've seen it with other companies, it's very important for us to continue to invest in the growth, but we can do both. I think for years upon years, we have told you margin goals and plans, we have always gotten to them. Always. I really don't think there's ever been a time where we've told you about a target of getting to a gross margin where we haven't gotten to it.
We believe that part is a lot easier for us to control because the whole cost of goods sold, the whole cost envelope, we understand it well enough so that when we mix that with the bookings and have a clear line of sight to the revenue, we think we can get there. We transitioned from the license business in the application side to the SaaS business. Yeah, the margins were lower at one point and got to where we said they were going to get to. Really, I don't think any of you really thought we would get there in this timeframe. Now, in the BYOL and the PaaS, IaaS ecosystem, there is a large pool of money for us to work with, and we are bringing so much value for our customers that it is quite obvious to them.
You got to ask a question today, I think it was Merck that happened to get the question. How much of your installed base are you going to bring to the Oracle Cloud, your installed base of your databases? I think she said substantially all. Okay? That's what we're talking about. When you think about the world's most important data and all of that, let's not say all, substantially all of it coming to us, you have to understand that the economies of scale will really spin out of control at that point. That's why we're so excited about this. That's why Larry's so excited about this.
Over here.
Hi. Thanks, guys. Thanks very much. Brad Zelnick with Credit Suisse. My question is for Mark. Mark, I'm hoping you can help us to frame how big of an opportunity, of an incremental opportunity, Autonomous Database really is. I think the perception over time is that Oracle releases a new version of its database, most customers are on maintenance. This time, if I'm correct, this is the first time in four decades that you have this kind of SLA around the database. If I'm also correct, those three options together are equal to about the price of enterprise edition of the core database.
When I put those together, can you maybe just give us a sense, I don't expect you to have numbers at the ready, but the penetration of those three options today within the base to see where the headroom is, and maybe in addition to that, if you can just share any feedback. You've been doing customer meetings all week. What are some of the wild responses and reactions that you've been getting? Thanks.
Well, first of all, I'm not falling for the modeling question. I think at the end, I'm really a big believer we need to do stuff as opposed to say stuff. I really don't like the let me give you a long-term model and tell you all these great things that are going to happen. The best thing we can do is just go do it and just go execute. That's what we're focused on is just, we're not trying to I reflect back, we've had two quarters of double-digit EPS growth. We've had whatever it was in Q3, I think was eight%, nine% EPS. We've moved the company forward. To Safra's point, I think a lot of people didn't know if we'd make this transition, what we have. We're now in yet another one of these exciting places.
SaaS was a big opportunity for us to transition our applications business. Now look at the broken field running we have ahead of us. I don't even want to prognosticate that because if I actually told you what was in my head, you wouldn't believe me. I'm just not going to say it because I think we are going to wind up. I've already said what I thought earlier. On the infrastructure side, what I believe will happen, without getting into numbers. By the way, I think your view is roughly right. It might even be a little better than what you described. Okay. Those are the core features, and you're in the right direction of what I think will happen. Those will become standard features that people will go out to buy.
I believe, again, as the average persona of our database customers, I would go buy those features. I would then bring them to the Oracle Cloud, and I would then buy the services I want, à la carte, as I brought my licenses to the Oracle Cloud. I would make a decision if I had to that said, I've got a regulator who's breathing down my back, who's making life tough for me, and I'd say, "Okay, Oracle, come do it for me in my data center." If I didn't have a regulator on my back, I'd go to the Oracle Public Cloud, and I wouldn't even mess with that. Many of our customers, I showed a stat earlier in the week. Year-over-year, the number of corporate data centers is down 15%. Customers want to get out of all of this infrastructure.
They want somebody else doing the work for them. It isn't just the fact that they'll get these features and get the SLA. The fact they transfer, Brad, I can't emphasize enough, all of this patching, all of this work to us is as big a benefit as anything else we could talk about. Are the customers today, I'd say, listen, I get the opportunity. I think without exception of all my speeches, I think I probably touched 650 customers this week, in various round tables. They mix between our apps business and our tech business. I think the most enthusiastic thing is exactly what we're talking about. Both people that have old applications that want to move, and they get the opportunity for somebody else to do the work, and these applications are really old.
On the tech side of our business, the fact that I can get out of doing all this damn work, lower my TCO, and get a better service while I do it. Let's face it, I haven't said their name much today. Amazon can't do this. They can't do it. It's not a question of how close. They can't do it.
Back here.
Hi, Walter Pritchard from Citi. Following up on the earlier question about the management goals. The margins, the gross margins, actually, the goal is quite low, I think 30%, which we look at those businesses and probably expect that as you get scale, substantial scale, you would get actually margin uplift from here. I'm wondering, what are the conditions or drivers that would bring those margins actually lower as you get substantial scale in IaaS and PaaS?
Lower than 30?
No, 30 or above.
Yeah.
They're way above 30 now. The goal is 30 or above. Oh, sorry. The goal is 30 or above.
Yes.
They're substantially above 30 today.
Yes.
We would expect if they're at almost twice 30 today and you're going to get five times bigger, that your margins would go up from where they are today, not down towards 30.
Yes. They will. It will.
You're saying it's okay. You'd like us to beat the goal.
Yes, we'd like to beat the goal.
We would, too.
There are a few ways.
Walter, we're in.
Yeah. You know. It's going to be at any one time, as we're doing. It really is dependent on how fast we grow. There could be a period of really hyper-growth where the margins are very close to 30, frankly, as we slow down, the margins will in fact, increase. There is that battle between it. The reality is that, even at our fastest growth rates, we should be at a point where 30 should be something we should be able to hit most of the time.
I agree with you that there's some goals in there that are higher and so forth than others. We expect to make them all.
Yeah. Yeah. Can I comment on one other thing, though? That we got a question earlier about when customers are in their own and all of that. I just want to talk about Oracle's own use of financials. I was going through an operations review with our financial team. Many of you don't realize, often we deploy something inside in our own private cloud, really often years before it's available in the public cloud. One of the applications that we had deployed privately a few years ago now was planning and budgeting. My own finance team, we have a lot of priorities, internal technical priorities, et cetera, this particular team very much wanted to go to the public cloud. Let me tell you what happened when we did.
Not only did the performance of the system improve by basically 100%, but because we were now in the most modern version, and it was implemented in the most vanilla, straightforward way, and it was the most current, we got so many additional features that we were able to deploy 91 full-time equivalents. People who had been working on the system, doing different things with the numbers to make them easily accessible and understandable by the management team, 91 full-time equivalents were released to go do other things. That's the kind of experience our customers are having. This year, we decided to put a PaaS, IaaS group of customers. Last year, I think we had a SaaS group-
We did
for you all. That's what's going on. I think we got a question about how SaaS is going. Let me just make sure we're clear. Thomas talked about version 13. This is an extremely important version, and I'll tell you why. It includes a lot of sophistication in the supply chain manufacturing piece of our cloud SaaS applications. Why do those matter so much? Many, not all, but many of Oracle E-Business Suite customers are discrete manufacturers. They have self-selected that they like suites, E-Business Suite. JD Edwards customers, they like it all packaged up together. They like it to work well together. This should be an extremely big year for those customers. Many of them have been waiting for this release. They've been fooling around the edges, but they're looking at deploying in the next couple of years.
Even though all we're doing around here is talking about IaaS and PaaS because our database install base is so enormous and so important, do not take your eye off the fact that the E-Business Suite crowd has not all shifted yet because some of them have been waiting for things, and that bell is ringing right now.
What would happen if we converted them from their support revenue? We would get three times their support revenue in SaaS revenue. Just if I didn't make that point clear earlier. This would be a good thing. In Q1, to Safra's point, we closed a lot of those, many of those deals we closed in ERP, some very important logos, were supply chain. This, release 13, is now the second release of manufacturing in the U.S., which is always a very important release for us. Now we're into a different dialogue with the rest of our ERP user base. With release 14, now you get everything in release 13 now goes global. You start to begin to get this ripple effect through our entire install base.
You just wonder what a double-digit billion-dollar apps company doing what we're doing in the marketplace is worth standalone. We're not doing that, right?
We-
We could. I mean, maybe.
We told these guys a long time ago, many of you have come back for more agony today, just like last year, you've been coming for the past 10-plus years, we told you that we're terrible losers. We tried to explain to you how we want to be number one in every market. The reality is, SAP has about 22% market share, we got 20. Many of you remember when they were five times our size. That is the only market in which we're in, really, where we had been in which we weren't number one. This is our move. Remember sitting here with the same really expensive, terrible lunch and us telling you how we would be number one in apps and that we were going to do it with this move to the cloud? This is what we were talking about.
Okay?
By the way, actually, I can't stop. Their model will work for a while. The licensing up your base, we know how to do that. That'll work for a while. Long run, these customers have to move to the next generation of applications. They have to move.
Maybe time for one last question.
Yeah, this thing here next to me says zero.
Zero. One more.
Hi.
That's it. You've made the decision, Ken.
Yeah.
Okay.
I'll take zero time.
Okay. Thank you.
We're going to address this.
Yeah, exactly.
Is that one. Okay.
Hi, Mark and Safra, it's John DiFucci from Jefferies.
Hi, John.
Mark, thanks for going through the cloud model, the steps for the cloud model today. I have a sort of a tactical question because I've been asked this question ever since you guys reported results, and it has to do with the guidance for cloud revenue for next quarter. Can you talk about why maybe the guidance is less of a sequential uptick than many of us may have had in our models? I have a quick follow-up after that. I think you hinted towards it when you talked about some latency sometimes between ARR and revenue recognition.
You have lots of moving factors. Again, I wouldn't try to get to, if you've got a spreadsheet Okay, I'm not going to do it. If you've got something out there and you're trying to just fill this in and map every number, there are variables you have to deal with, okay? Let me go back through the variables again. If we book, everything doesn't auto-provision at the time of a book. Depending on what got booked and what pillar where, there can be more implementation to do to get something, and the more ERP that we sell, this is good news by the way, but short term, it is a longer provisioning time to provision ERP. It can also hit to some degree in HCM. For example, it's shorter to provision in something like marketing. Some aspects of sales automation are faster.
For example, what Doug demoed CPQ, that takes longer to implement. There's different provisioning times.
Cloud@ Customer.
Cloud@Customer, for example, much longer-
Much longer
to provision. We've got to get data centers ready and all this stuff. It's not a perfect science. You get the impact of the seasonality of our renewals. Our biggest ARR quarter is still typically Q4, ARR. When the ARR comes back, if we have one-year renewals or two-year renewals, depending on when those are, you would see the most renewals. The cancellation rate potentially would see its highest. It's very different from the support model, would hit sort of in that end of Q4, Q1, Q2 timeframe. Again, I just caution you to try to make too perfect a science out of this because there's still a number of variables that go on.
The big issue is if you just end of day, look at our ARR over a couple, three years, go look at our revenue base over a couple of three years, go look at the renewal rate over a couple, three years, the revenue's roughly right. I mean, it's within chump change of being right. It may go up or down seasonally quarter-to-quarter, that model is the flow. After careful analysis, my determination of the best thing for our applications business is just to sell more.
Thanks.
Like big bucks. Okay.
I think Ken's going to let me ask a follow-up. There you go.
This is absurd.
No, you can't. No, it's not at all.
Yeah.
I'm sorry. Should we be thinking about this too, especially with that latency issue.
Wait a second. This is version two of the same question.
No.
Okay. All right.
It's just quick. Three years ago, there was a sales incentive thing where you would sign.
Sales incentive thing
ARR, and you'd sign ARR, but then you didn't start recognizing the revenue for six to nine months.
These are called promotions, not a sales thing. This is a promotion, yes.
Sales promotions. What happened for those of us who, because we model with only the information we have. We got less revenue than we thought, all of a sudden we got more revenue than we thought we would get a few quarters down.
Yes.
Should we sort of expect to see something like that this time around too?
Again, let me try one more time. Over the long run being defined as over several quarters, this will smooth. I'm trying to give you the caveats that occur within any given quarter. I'm just looking over here. Am I doing this effective? Thank you very much. I just like to look to the team to say, "Yeah. No, you're doing exactly the way." It goes this way. We have a revenue base. A % of our revenue base is up, I did it on an annual basis, is up for renewal. The % of ATR available to renew will fluctuate in a quarter, and it typically anniversaries relative to the ARR we sold in prior periods. If you know the ARR, what you don't know is the length of contract. The length of contract inside our SaaS base is lengthening. It's getting longer. Why?
ERP and HCM, very sticky applications are a bigger and bigger % of what is in our ARR. Contract length is going longer. To be able to figure out how the revenue works, you would have to know the ATR by quarter, and frankly, you'd have to know it by pillar, because each pillar has a little different dynamic. You layer on any new and expansion ARR on top of the revenue base minus whatever's not renewed, plus the ARR. You have the dynamic in the ARR of how quickly that's provisioned. That could be different based on pillar and different based on Cloud@Customer.
All of that, though, over the course of a year, should 18 months smooth out, even though you could have some dips and valleys and so forth based on these variables, it all works out.
Thank you.
Thank you.
Thank you. Bye, guys.
Please welcome to the stage Oracle Chairman of the Board and Chief Technology Officer, Larry Ellison.
The great thing about not being CEO anymore is I don't have to dress up. Whatever you guys want to talk about. Yes, sir. The lights are pretty. Can you dim the lights a little bit? I really can't see any faces out there. Okay. All right. I'm good.
Hi, Larry. Karl Keirstead at Deutsche Bank. Thanks for coming. Actually, I wanted to talk to you about the on-premise space. One of the interesting things I think about Oracle's numbers in the last couple of quarters is that it's actually been the on-prem license numbers that have surprised on the upside. I'm curious, as you're talking to customers, Larry, do you think there's any change in behavior or maybe the pace of cloud migration is slowing at all? Maybe you could explain what's happening, because we're seeing it not just with your numbers, but with some other infrastructure, software and hardware firms. Thank you.
Okay. I can speak to our numbers. Our applications migration to the cloud and our database migration to the cloud is totally different. It couldn't be more different. In the database migration to the cloud, we're expecting you to take your existing versions of Oracle, and we'll keep upgrading them, obviously. We just came up with a big announcement, the Oracle Autonomous Database. We're expecting you to migrate your Oracle Database workloads to the cloud. Just basically lift up what you're currently doing on-premise and move it to the cloud. We're not expecting any technology change whatsoever, any migration from X to Y, where you go from Oracle on-prem to Oracle in the cloud. Let me describe our application migration. We're expecting you to stop using SAP. Actually, let me just talk about Oracle. You stop using PeopleSoft and you implement Fusion Financials.
You stop using Oracle E-Business Suite. You stop paying support on Oracle E-Business Suite. You migrate and you move to a totally new product, Fusion Financials. You stop using JD Edwards. You stop paying support on JD Edwards, and you migrate to a totally new financial suite, Fusion Financials. Nothing could be more different. We'd be thrilled if E-Business Suite support went to zero and all of those customers moved to Fusion Financials. Our revenue would approximately triple, and our profits would probably double. More than double, probably. Again, couldn't be more different. There is no reason why our database support numbers should ever go down. They should go up forever. Especially now, you own the database, and you can take it if you want to. You can take it to Amazon, and if you take it to Amazon, you continue paying us support for those licenses.
You can take it to Oracle with bring your own license, and you continue paying us support for those licenses. There's really not much of a migration going on the database side of the business. Not a business migration. You're just moving your Oracle workloads from your data center to our data center. We're adding more value. Therefore, you're paying us more. Rather than building your own data center, you're renting our data center. Rather than buying your servers from HP, you're renting servers from us. Rather than getting disk storage from EMC, you're getting disk storage from us. You're not stopping anything. You're not stopping using the Oracle Database. You're not stopping paying support. In fact, theoretically, you're going to need more. You're going to need more options. If you want to run the Oracle Autonomous Database, you're going to need the multi-tenant option.
If you want to run the Oracle Autonomous Database, you're going to need the RAC option. If you want 99.995 reliability, you're going to need our disaster recovery option called Oracle Active Data Guard. By the way, Amazon doesn't have RAC, and they don't have Oracle Active Data Guard. They don't have these kinds of things. We think they're going to come to our cloud, not to someone else's cloud. We think we can run the Oracle workloads a lot better than anybody else. By better, I mean clearly more reliably, clearly more securely, clearly faster. Everyone said, "Yeah, but you guys are really expensive." Well, if you run five times faster, and we tested a whole bunch of Oracle workloads at Amazon, and these are real workloads from real customers whose names you've seen up here. These are not made-up database benchmarks.
These are real workloads customers give us to test our database. We average seven times faster than Amazon and five times less expensive. In other words, Amazon costs you 5x to run the same workload, the same database, the same queries, the same everything. You run it at Amazon, your Amazon bill will be five times your Oracle bill. Because if we run it seven times faster, for every compute second we charge you for, Amazon charges you for seven. This performance difference translates into a gigantic cost differential. They're not competitive. They're not close to competitive. I read articles that say, "Well, Amazon's got this Amazon RDS service, and they can automatically patch Oracle, too.
They can take the database down, take it offline, run some scripts against it to apply a patch, and then bring the database back online. They can do that. They do do that. That's not what we do. With Oracle, the database never comes down. It never stops running. By the way, there's real cost associated with taking these things down and making services not available. Forget the cost. It takes people also to run these scripts and patch the database and bring it down and bring it back up. With Oracle, it's autonomous. The database patches itself. There are no people involved. That's very inexpensive. There's no pilot error because there's no pilots. The amazing thing is the system never comes down. Never comes down. We patch while running. To patch the thing takes a fraction of a second.
When they do the patches or major upgrades, it can take hours and days. Let me say it again, hours and days. They're just not competitive. They're not even close to competitive. They will never have the Oracle Autonomous Database. They'll never have anything like that. With our BYOL pricing, we don't expect people to go through a transition. We'd expect them to take their existing Oracle licenses. They might have to buy a few extra options like multi-tenancy, if they want to run the Oracle Autonomous Database, which is the license business, if you will. People think of the license business doing well, meaning the on-premise business is doing well. They're close, but they're not the same. You buy license options. If you want to run the Oracle Autonomous Database in our cloud, you buy the multi-tenancy license to do it.
We're optimistic about our support going up. Our license business should be fine. We intend to keep selling Oracle database licenses with the bring your own license model. Set that aside. That is not the case over in applications. There's no Fusion database license. There's nothing like a Fusion database license. There's no support associated with a Fusion database license. There you move from a purely, if you will, on-premise product to a purely cloud product. There is this mid-range stopping point that we have, which you can lift up your E-Business Suite product as is, and you can run it on our Oracle Autonomous Database and infrastructure as a service. You can take SAP applications, for example, and move them to our cloud. You run the SAP application on our infrastructure as a service. They've certified SAPs.
You run the Oracle database that's behind the SAP application, you run that on our PaaS, our database as a service. You can lift and shift existing on-premise application workloads and move them to the cloud, but that's not SaaS. That's a combination. It turns out to be the application runs on infrastructure as a service, and the database runs on PaaS. I'd just like to distinguish those two separate businesses, which have very different transitional characteristics. Very different characteristics. Our tech business will always have a support component because of PaaS. Even if everyone moves to the cloud, we have 100% of our customers are running in our cloud. They'll still be buying licenses, buying more licenses and buying more options, paying support. Not the case if everyone moves to our applications. There will be no residue from those applications.
There'll be no more support fees, nothing. They'll all disappear. We moved into the SaaS business long before we moved into the PaaS business. We started working on our SaaS more than a decade ago. In fact, we were now, with the acquisition of NetSuite, we were the first cloud company on the planet Earth. The first SaaS company on the planet Earth was NetSuite. The second major SaaS company on the planet Earth was salesforce.com. I was involved with starting both of them. NetSuite was the first cloud company. It's, God, is it 20 years old now? Close? Amazon started about 10 years ago. Very different. You really can divide the cloud into two major segments.
The people who wrote applications for the cloud, where NetSuite was the pioneer, and then people who were renting infrastructure on the cloud, where Amazon was the pioneer about a decade ago. They are very separate businesses. We entered one long before we entered the other. The transitional characteristics are very different. Sorry, that was a very long answer to your question. Yes.
Hi, Larry. Kash Rangan, Bank of America Merrill Lynch. How are you?
Good, thank you. Good to see you, Kash.
Excellent. With all the shifts in the database technology landscape over the last 20 years or so, with this Autonomous Database theme that you're pivoting yourself around, you look at AWS, Google, Microsoft, they've made big investments in AI as well. How truly differentiating could this be for Oracle relative to the other guys? Thank you.
Okay. There's two parts. You're absolutely right. In fact, Google's really good at machine learning, I think. Amazon's really good at machine learning. Tesla's really good at machine learning. There are a lot of people that are really good at machine learning. Amazon's one of them. You know the problem, Kash? You know the problem? They don't have a database to automate. What's their database? Amazon Redshift? Amazon Aurora? We develop Amazon Aurora. They don't. Amazon Aurora is MySQL. We develop Amazon Aurora, and they just picked up a chunk of open source stuff that is, I think, 20 years out of date, and they rent it on Amazon. That's called Amazon Redshift. They can't really automate Amazon Redshift. Let me tell you some of the problems. One of the things that's very important, I think, is being able to upgrade or automate security patching while the database is still running.
That's a combination of a security feature and a database feature. For us to do that, we have to have RAC and our automated security patching. You need two features to do that. You need to be able to change something while it's running. The way we do it is. By the way, RAC is Oracle Real Application Clusters. It's taking a bunch of non-shared memory computers and having them operate on the same database. No one else has this. One of the unique things we have. In fact, the closest company to have something like that is IBM, and they had it with their mainframe version of Db2. They had something like RAC. Microsoft sure doesn't have it. Amazon, oh my God. Microsoft's 10 years ahead of Amazon in database. Amazon doesn't have databases. Amazon took a couple of public domain databases that were open source.
They made them closed source. They basically forked them and rent them. That's what they do. That's what Amazon Redshift is. No one's invested in these systems for a decade. To patch a database online, you have to have this feature called RAC, where you have multiple computers accessing the same data. Changing the same data and running at the same time. It's a very tricky problem. It's a very tricky what's called a locking problem, distributed locking problem. Anyways, very hard computer science. We've done it for a long time. Amazon doesn't have it. They have nothing like it. If it ever showed up, it would be developed by us as part of MySQL, certainly not by part of Amazon. Amazon has no expertise in database. Amazon runs their entire operation on Oracle. Do you think they run on Amazon Aurora or Amazon Redshift?
They're one of our biggest customers in the world. They gave us $60 million last year, support and license. They've been one of our biggest customers forever. They've had the cloud for a decade. You know who's not on Amazon? Amazon's not on Amazon. I know they're going to kill us. Everyone else is going to go to Amazon, but Amazon sure isn't. That technology, yes. You're totally right. They're really good at machine learning. By the way, and so is Tesla. There are a bunch of new companies that are very good at machine learning. They've applied the machine learning for digital assistants or self-driving cars or what have you. Applying it to database, you have to have a database. To be able to patch while the system is running, you have to have something like RAC.
To be able to provide real disaster recovery, you have to have something like Oracle Active Data Guard. They don't have anything remotely like this. They don't have good query processing. We took workloads off of Amazon Redshift and ran them on Oracle, and Amazon Redshift took 15 times longer. I know you didn't believe me when I said this about Oracle Exadata, Oracle Exadata is 50 times faster or 10 times faster and all that. We published the benchmarks. They're on our website. Go look at them. They come from real companies. You take a Amazon Redshift workload and you run it on Oracle, it's 15 times faster. You said, "Who cares?" Well, when you're charging by the second, you care. 15 times faster means we're 15 times faster, Amazon's 15 times more expensive, because they're about the same price per second. Actually, we're more per second, but not 15 times more.
The problem isn't that they don't have the machine learning half of the equation, they don't have the database half of the equation. Not many people use Amazon Redshift or Amazon Aurora anyway. That's not our concern, Amazon Redshift and Amazon Aurora. Our concern is people taking Oracle workloads to Amazon. That's our worry. If you say, do we think people are going to move an Oracle workload to Amazon Redshift or Amazon Aurora? No. Don't get me wrong. A new developer who wants everything for free the first year, and Amazon's done a very good job attracting new developers. They're starting over and can they use some of these technologies and work around some of the limitations? Sure. I got it. That's not our primary concern. Our primary concern is preserving our majority market share in the database business.
In fact, increasing our market share at a higher rate than we're currently increasing it at. Our database business continues to grow. We think it should be growing much faster. We're more than half of the market. We'd like to have what I used to call gate share. Gate share is nice. In the era of the internet, it's possible. You see network effects and the possibility of getting very, very large market shares. Our concern was people taking Oracle workloads to Amazon, because Oracle is a very sophisticated database. Our cloud, we've made a bunch of investments in our cloud, added a lot of AI to our cloud, added a lot of features to our cloud, where we now run our database five times faster in our cloud than Amazon can run it. It's not 15 times, it's not Amazon Redshift. Oracle's a lot better than Amazon Redshift.
We run our database or the Oracle Database five times faster. They can't compete with that. We patch when the database is running. They can't compete with that. We upgrade the database while it's running. They can't compete with that. We have a 99.995% availability guarantee. They can't compete with that. We will sign a contract. You can take any Oracle workload, you can take any Amazon Redshift workload, and when we come out with it in June, any Amazon Aurora workload, we don't even look at it. Bring your bill from Amazon, and we will cut that bill in half, guaranteed. We'll put it in a contract to all comers. They can't compete because their systems are primitive and slow. You ask yourself, when did Amazon build Amazon Redshift anyway? How many people does Amazon have working on Amazon Redshift? Where did it come from?
Did the Amazon engineering team build Amazon Redshift? No. The Amazon engineering team, are they investing huge amounts in Amazon Aurora? How did they get this great data Oracle killer? From Oracle. We own MySQL. We are not worried about MySQL. Again, there are three pretty good databases on the planet Earth, relational databases on the planet Earth. One belongs to us, one belongs to IBM, and one belongs to Microsoft. Amazon doesn't have any. Being good at AI ain't going to help them solve this problem. Yes, sir.
Hey, Larry. Alex Zukin from Piper Jaffray. I wanted to ask a question. Safra made a comment that Oracle is finally ready to take, on the application side, mission-critical production workloads to the cloud, and that's a big deal. I wanted to ask you, as you look at your application customers, over what time period and what % do you expect to take you up on that value prop?
It's really very interesting. We've seen huge companies take their HCM workloads. You have to mention the application specifically, which application you're talking. We've seen very, very large companies. We compete with Workday. You can argue they're a little in front of us, we're a little in front of them. We think we beat them a lot more than they beat us, but what have you. They're a formidable competitor. We compete with them very aggressively. We both have significant market share in the HCM business, and lots of big companies are putting HCM in the cloud. That's been going on for a couple of years now. HCM is a rather simple product compared with ERP. Financial, supply chain, manufacturing, procurement. This is an order of magnitude more complex set of processes than HCM.
Right now, we are the world leader in ERP in the cloud with no one remotely close. I think in the cloud and ERP, our closest competitor is Workday. I would almost describe their product as not really being out yet. They have someplace between almost none, depending on how you count, to a few hundred customers. We have more than 5,000 Fusion financials ERP customers. Then on NetSuite, which is the lower end of the market, we have another, what, 13,000 customers. We think Oracle moving to Fusion financials is a very big deal because And there are other big companies that are also in the middle of migrating to Fusion financials, if you will, Fusion ERP. We think that is a gigantic market because the incumbent doesn't have anything. SAP has just bowed out of that.
I'm not trying to be cute, but they don't have a cloud, really. They have this thing called S/4HANA for the cloud, which really is their old on-premise system that they host. They can try to fool some of the people some of the time, it's not going to work. Every line of code of Fusion Financials was rewritten for the cloud. Every line of Fusion HCM and Fusion Sales and Fusion Service, we had to start over and rebuild them on cloud middleware, the cloud versions of our database, all of those things. It was a long process for us. I thought it would take us five years. It took us eight. SAP has yet to start. They had a project that they canceled. The project aimed at NetSuite, called Business ByDesign. They ran it didn't work, they canceled it. They're gone.
I'll almost be curious to ask you guys, who do you think in the SaaS world, in enterprise applications, manufacturing, supply chain, ERP, financials. Let's just keep going. Marketing, sales, service, all of those things. We have all those products. We're the only company on the planet Earth that has all those products. Who do you think is going to win? Five years now, is there going to be a dominant enterprise SaaS company? If there is one, who do you think it's going to be? Seriously, who? Salesforce? They're the only ones that still have a larger SaaS business than us. They're the last guy, we're growing more than twice as fast as they are. That's not their problem. They don't have an ERP product. They're not playing in most of the market.
They have nothing in manufacturing, nothing in supply chain, nothing in ERP, nothing in HCM. They have a powerful position in sales automation. It's in the name of their company. They have a service offering. We're their primary competitor, we're about even against them in service. They're really not a big player in marketing. They play in marketing. Adobe's a bigger player in marketing. We're their biggest competitor. We have a big play in marketing. We're in all of these. Who is going to be number one in SaaS? We have no competition in ERP. ERP is more than half of the total enterprise market. The last generation winner was an ERP company called SAP, the second-place finisher was an ERP company called Oracle. Then you had Siebel and some PeopleSoft and some others. I'm serious. I'm just curious.
Who do you think is going to beat us in enterprise SaaS? It's a gigantic business with phenomenal margins. It's much better than the on-premise business, the cloud business. You're adding much more value, and you really can count the users. I think we're the big winner in SaaS. By the way, this will be the third year in a row, this year, that we've sold more new ARR than salesforce.com. Yeah, they started 15 years before we did. We haven't passed them quite yet, but it's very close. Last three years, we've been the biggest seller of SaaS applications on Earth. Last two years, this will be the third year. I'm not talking about % growth or anything. Just dollar for dollar, we sell more SaaS than they do. This will be the third year in a row.
They don't have a lot of products. They don't have a lot of presence outside the U.S. Who's going to beat us? Who's trying? I can only find one company, Workday. Our ERP is, I don't know, years in front of them. Their real strong suit is HCM. We're beating them badly in HCM in the middle market right now, and let me explain why. There's no such a thing as an HCM middle market. HCM and ERP go together in the mid-market. You don't make a separate HCM decision and a separate ERP decision. Whenever it's an ERP HCM bake-off, they don't have anything much in ERP. We win all of them. Can Workday survive just selling some number of high-end HCM systems? Let's say they're even with us in high-end HCM system. We get half, they get half.
We get all the mid-market, and we get all the high-end ERP and all the mid-market ER. How do they continue to invest? They have lots of problems. They need a platform. People, if you really want to be the ERP business, people write extensions, they write special reports, they write extensions to ERP. They add to the SaaS. You need to provide those tools in your cloud to let your customer add to your ERP suite. We have a platform. They don't have a platform. What are they going to do? Security has become a bigger and bigger deal. Securing the data, having an autonomous system that makes sure that your cloud is protected, that your computers are defending your data against their computers. What is Workday going to do? Or any of these really small companies going to do in the era of cloud?
I think we're going to win in SaaS. I think we're already winning in SaaS. I think we're so close to being the largest SaaS company. Every year, we sell more than anybody else. With the autonomous database now, we've taken a huge leap forward in the database application. Did anyone see the article? I was actually shocked when I read it, but it made complete sense to me. Anyone see where Satya Nadella was asked, if he could have any technology on the planet Earth, what would he take? Yeah. Take our database. It's called the information age. It isn't called the web search age. It isn't called the cloud age, not called any of that. We right now have the most sophisticated applications, technologies for managing information. Most of the world's precious information is already stored in Oracle Database.
Do you think it's going to move to Aurora? Let me tell you who we competed against. IBM, the relational model came out of IBM. IBM in those days, IBM today, not what it used to be. IBM in those days was a pretty powerful company with an incredible track record of things. They invented the disk drive. They invented core memory. AT&T did the little thing called the transistor. Those were formidable companies, AT&T and IBM back in the day. They were the dominant database company, and we killed them. We killed them. Then along came Microsoft, the most valuable company of the day, and a software company, not a bookstore. It's not funny. I love my Kindle. It's not funny. I love Alexa. Alexa, play something. Machine learning. There you go at its best. It kind of knows what I like. All right.
We killed Microsoft in the database business. All these technologies over the years. We've been working on this problem for a long time. We have a lot of good people. Oracle's fault-tolerant database. We lose a server, we keep running. People said, "Gee," Larry Ellison said, "That if Company X had used Oracle technology, they wouldn't be testifying before Congress today." It's true. It's true. Our stuff is secure. It's reliable. It's robust. We've been working on it forever. There are all of these companies' products that were going to kill us. The most ridiculous are Amazon Redshift and Aurora. Those are really ridiculous. We've already beaten them in the past. They're just renamed old stuff. Old open source stuff that's been. By the way, open source was going to kill us, too. All these things were going to kill us.
Object databases were going to kill us, blah, blah. Didn't happen. What Amazon came up with was this fabulous way to deliver computer services, lots of them, in a compute utility. Brilliant. Great idea. We're on board. We think it's a fantastic idea. Absolutely a great idea. Great way to deliver applications. We think we're going to win there in SaaS, and a great way to deliver data services. You still have to have a database, and they don't have one. If they don't have one, then they have to figure out a way, okay, let's get the Oracle customers to come to Amazon rather than going to Oracle. Oracle's late to the cloud. Oracle's late to the cloud. Okay. All right. You want to say we're late to the cloud? Fine. We'll guarantee ours is half price. We'll guarantee ours is 100 times more reliable.
That's what we do. 30 minutes of downtime a year, 100 times better reliability than Amazon, half the price. We never go down. We never, ever go down. Guaranteed half the price. The reason we can guarantee half the price, because it's probably 20% the price. They're probably five times more expensive. Put in a contract. I think we're going to be just fine. We're the number one database on the planet Earth. We've been the number one database company for a very long time, and Amazon doesn't even have a database. How are they going to beat us? Okay. Yes, sir.
Thanks. Hi, Larry. Michael Turits from Raymond James. You rolled out security monitoring and analytics. I want to talk about security. Many people are discussing what happens to security in a cloud world, what happens to that whole universe of firewalls and endpoints, et cetera, when workloads move to the cloud. Tell us how you see all of that changing and how Oracle participates.
Oracle is unique in the cloud world that we offer our cloud in two locations. We offer our cloud in our own data centers, and we offer our cloud behind your firewall. Now, let me be clear. It really is our cloud. Those are our computers that we own. You don't own them. You don't buy them. We own them. That's our software on those computers. You don't buy it. We own it. BYOL is what we get. Sure. Okay. We manage it. That is literally a version of the Oracle Cloud that we plunk behind your firewall for security reasons. We're the only ones that do that. We do it for lots and lots of reasons. There are some industries that are highly regulated, where the regulators prefer that kind of configuration. That's fine. Makes no difference to us.
By the way, our labor still accesses that portion of our cloud. It just looks like another Oracle Data Center, if you will, to us. This is another thing. Everyone say, "God, Amazon's way ahead in data centers. Microsoft's way ahead in data centers. Will we ever catch up in data centers?" Oh, yeah. We'll have way more data centers than they have because some of our largest customers are electing to say, "Why don't you put a data center in behind our firewall on our floor? You own the hardware. It really is just an extension of your cloud." We call that Cloud@Customer. That's very different than, people talk about all these, I take my existing HP or IBM mainframe and you make that part of the cloud. That's not what we're talking about. This is not a fake cloud behind your firewall.
This is the same exact software, the same exact hardware that we have in our public cloud. It's not a private cloud. We put that Cloud@Customer. We think that will allow us to have large installations at banks, at governments, at telecommunications companies that otherwise would be hesitant to put. Some people would argue a phone company is just a big computer. Some people would argue a bank is just a big computer. That's their primary business. We are going to build lots and lots of data centers on top of our customers' floors, and that's proved a very popular option amongst some of our largest customers. That's one thing. The other is who should be better? Again, in my speech about cyber defense, that's what we're playing, a cyber defense. Who's playing offense?
We know about cyber criminals, we're talking about state actors. We're talking about nation-states going after your data. If you're a medium-large company and you're in a cyber war with the People's Republic of China, who do you think is going to win? I think a lot of people are going to feel a degree of comfort that they've outsourced a lot of their cybersecurity to Oracle and the Oracle Cloud as opposed to trying to do it themselves. That may not be true at AT&T. Maybe AT&T says, "No, we're on it. We're good at this. We'll do this." Some of our largest customers, the JPMorgan Chases, the AT&Ts who pay a lot of attention to this, maybe can afford to engage and be a participant in these cyber wars, most people can't.
I'm going to say most companies are going to have much better security, much better reliability by relying on somebody else to protect their data. We're very good at this. Again, going back in history, Oracle's very first customer, the Central Intelligence Agency. Oracle's second customer, the National Security Agency. Oracle's third customer, the Defense Intelligence Agency. We paid a lot attention to building security into our database from day one. We think we've done a pretty good job. Is it perfect? Nothing's perfect. Is it unbreakable? Nothing's unbreakable, but it's closer than anybody else. Again, we think the vast majority of our customers will be more comfortable and actually feel safer if someone else is protecting that data against that as opposed to them. Our very largest customers, we give them the option.
You can run some of your stuff in the public cloud and some of the stuff at Cloud@Customer, or you can put everything in Cloud@Customer. We really don't care. We're happy to locate the computers anyplace. You want them behind your firewall, we'll put them behind your firewall. It's still part of the Oracle Cloud. We own it. We upgrade the software. We manage it. You pay by the drink. We think having those options, which Amazon doesn't have and Microsoft doesn't have, gives us a big advantage, especially with the largest customers in the world. Governments, especially governments outside the United States, are all looking for that kind of thing. Cloud@Customer.
Thanks very much. Hey, Larry. It's Brad Zelnick with Credit Suisse. Larry, over the course of four decades, Oracle has defied the natural forces of obsolescence in enterprise technology. You've had what once were significant competitors, either out of business or completely irrelevant. Meanwhile, today, you sustain your relevance. My question is, why do you think that is? Why do you think it persists over the coming decade? Because as we sat here this morning, personally, I can't speak for the whole room, I was blown away seeing Thomas' presentation, listening to Steve, really appreciating all of the innovation. Last I checked, you're a pretty large shareholder.
What gives you the comfort that there's not a generational effect that it's just, if we look around, you look at the average age who's walking the show floor, our customers, even if you have the best technology, because we've seen the best technology lose in this game in the past. Even with the best innovation, how do we get comfortable that Oracle sustains its relevance?
Well, I'll start with, let's look at the SaaS business. I'll argue we invented it. It's not like we played catch up in SaaS. I started NetSuite. It was my idea. Salesforce then, it's my idea. Okay. Whatever. We didn't miss that one. People say I missed the cloud. I feel slightly different about that. I thought I invented the first cloud company. I thought NetSuite was the first cloud company I thought I invented. Missing it and inventing it are quite different. That was 20 years ago. It was a long time ago. It's just coming. Everyone would agree, obviously right now, everyone thinks that applications on the internet are better than application on-premise because it's 20-year-old technology.
I was just with Elon Musk and saying now everyone pretty much agrees, including the Chinese who are pushing their car manufacturers, electric cars are the way to go. I think this whole new wave of information utilities. Talk about the network computer, the idea of hiding, all the complexity should be in the network, and you should have a simple appliance to access your data and access your applications, and carry it around with you. Even if it's on a desktop, it should be a simple appliance. You shouldn't have a complicated device to get at all this stuff. That was computing catching up with all of the other networks on Earth. The electric power network's ungodly complicated. There are nuclear power plants and thermal plants and solar plants and all different kinds of solar plants.
There's the grid, one-way, two-way grids, the fail-safes in the grids. The water networks, capturing the water, purifying the water. Ultimately, a simple device at the end, a faucet and a plug to tap into these enormously complicated networks. Lo and behold, finally, computing notices that's the right way to build networks. With the internet, all of the complexity was pushed in the back, and we have simple appliances. These network computers, whether they're smartphones or tablets or what have you, are now the way to go. Even your PC has become primarily a device for running a web browser. Yeah, you have some local tools. You have a word processor and a presentation and spreadsheets. By and large, most of the complexity is not on your desktop. All your data, your files are probably stored in a Microsoft cloud, unless you're a gambler.
The world's IT networks have become like all of the other networks. With the complexity in the network, the capital investment in the network, and consumers of data and consumers of applications paying the utility for those services. I think that's the final state. I don't think there's another state. That's been going on since the internet. Nothing has changed about that as far as I can tell for the last 20 years or so. The relational model, which we've been working on. We had the first commercial relational database. We're constantly looking at. There are lots of these new technologies. People said Hadoop's going to replace relational, not Oracle specifically, relational. They even had something that was really right in our face called NoSQL databases. Like NoSQL. NoSQL allowed. NoSQL.
They changed the word NoSQL from the word meaning no, not any SQL. They said, "Oh, no, you understand. That means not only SQL." It means yes SQL, but other stuff too. Oh, that's what NoSQL meant. Okay. They changed the name. There's no database technology I know of, not Hadoop, not Apache Struts, that have the power and the resilience and the ability to migrate schemas. I won't go into all of it. There's a reason why today there's nothing that's taken its place. There's nothing really that's taken its place. Our version of that relational database is just much better than IBM's or Microsoft's or whatever's available in open source, what have you. There's no obvious successor technology.
When more than a decade ago, we decided as a company that we had to rewrite all of our applications for the cloud, if you will, for SaaS. At the same time, we said we had to put a bunch of new features into our database, like multi-tenancy features into the database. Like the latest autonomy features with machine learning into the database. It was really just absorbing these next generations of technology into the same underlying idea for modeling data and storing data. We have successfully. We do store objects. When object storage became very popular, we say, "Okay, no problem." We'll solve the object problem for you inside of the relational model because you want all of your data together. It's not like all of your data is going to be modeled like objects.
Some is going to be modeled like objects, some is going to be modeled like tables, sometimes called relations. That's where relational model comes from. What about these NoSQL databases, being able to have these key-value pairs for very fast streaming? Well, guess what? We put that into the Oracle Database too, and we got a key value NoSQL database in Oracle that's faster and more reliable and more secure than the open-source stuff that's around. The idea is to have one unified. I don't think you want to have 20 different databases that you're managing in your shop. I think there's only one asset that's really important inside of an IT, and it's not the servers, it's not the network, the asset is the data. You better make sure it's secure and reliable. If you have 20 different databases, good luck.
You have a separate NoSQL database, a separate object store, a separate this, a separate that. You got all the different Hadoop databases. I mean, God bless, if you want to do that, go ahead and do that. I think that's a very expensive strategy. You have to have people who then are experts in all of these things. It was kind of funny to see Apache get blamed for a data loss. The patch was available. The team just didn't patch all the instances of Struts, that made the system vulnerable for data theft. I think as cyber warfare increases, there are more actors going after your data. I think having a highly robust, highly secure way to store that data, it becomes more and more important. We have a big lead over everybody.
I think we proved that again by just running these benchmarks against Amazon. By the way, I think not only are we much cheaper. The whole point of running those benchmarks is not to say that we're faster, more secure, blah. It's because everyone would say, "Yeah, probably. Who cares? Amazon's really cheap." You go there to save money. No, we're much, much cheaper than they are. Security, got to be willing to pay less. 99.995% uptime. Got to be willing to pay less. Having all of those features and all of those capabilities are the best way to deal with obsolescence that I know of. Yes, sir.
Hi, Larry. This is Phil Winslow from Wells Fargo. Obviously, you talked about it from sort of a platform side, but I really want to focus on the applications front and sort of how AI and ML can get integrated in there. When I think back to the launch of Fusion, you guys talked about integrated business intelligence and visualization of data. When you think about AI, sort of taking that a step further, sort of automating the analytics next phase to action. How do you think this plays out, and then how does Oracle monetize this? Because the applications are better, is it more share gain? Is it the churn is reduced?
I think it's more just share gain. I think I'd answer that right. For example, in our competition with I'll let you go Did I cut you off? Is there more to the question? Okay. I think it all becomes share gain. I don't think it suddenly means we can double our price. We're interested in gaining share, and that's our primary focus. Of course, we like making money too. We think EPS should go up a little. We think when you gain share, that happens quite naturally. One of the things we're doing in HCM, for example, is we compete with Workday. Workday needs a platform, Workday needs security. Workday also needs to have machine learning. Just going to get harder and harder for them to do that because what we want. We hire a lot of people. We've hired 5,000.
We're building this huge hub in Austin, Texas, right on the Colorado River. We're building another huge hub in Santa Monica, where rather than having branch offices everywhere, we're selling out of these hubs, providing support out of these hubs, providing implementation services out of these hubs. By having the hubs, we can train our employees every day. We do a lot of recruiting. A lot of recruiting directly out of colleges, and we have for a very long time. We would love for our HR system to tell us when they look at the background of different people, of all the people who are applying for jobs, who should we put at the top of the queue? What are the backgrounds that have been most successful? A lot of people say, "Gee, athletes make great sales.
If you participated in college sports, you have a great competitive spirit, and that's ideal for selling." Is that really true? I've heard that. Is that true? It's got to be in the data, right? We got all this data, all these people we've recruited. We know their backgrounds, we know what works, and I would love to know the answer to that. Is that a good thing? I don't know. I think machine learning is going to help us enormously, to prioritize the different resumes that we look at and kind of discover the things that have made people successful. I think it will help us prioritize who we make offers to. I think it will also help us figure out the people inside of Oracle who've been successful. What are their common characteristics? Forget their background. What groups did they work in?
How many days of training did they have? Were they located in a hub, a regional office? Were they close to their manager, not close to their manager? There are a whole bunch of other things we can discover about, okay, we've hired you. Now, some people are successful after they've been hired, some people are not. Might have nothing to do with their backgrounds. It might have to do with how we manage them, how we train them, where we put them. We'd like to know about our onboarding process. Starting with the onboarding process, the mentoring process, the management process, all the different processes we have in place for helping to make our employees successful. Which are the ones that are really adding a lot of value, and which are the ones that actually make no difference, and which are the ones that hurt?
I think that's one example, but it's an important example. That's one example of how machine learning can fundamentally change HCM. There are a whole bunch of other, picking suppliers. There are a whole bunch of other examples throughout our application suite where we're using machine learning, and I think it's going to make our applications more valuable, make them more competitive versus the competition, allow us to gain share. When you gain share in this business, obviously our development, our engineering costs are kind of a fixed cost, which then get amortized across our community of users. The more users we have, the more money we make. Share is important. Yes, sir.
Thank you.
Hey, Larry. It's John DiFucci from Jefferies. Putting aside NetSuite, which you had obviously a hand in how that was architected from the start, and maybe some of the other acquisitions you've made, SaaS acquisitions. It seems that most of the infrastructure for the Oracle Cloud is architected as a single instance for each customer. I'm going into maybe some terminology that I know is very-
Just not true.
Okay.
Just not true. The Oracle Autonomous Database runs on huge Exadata machines that are partitioned among users.
Okay. That-
Time sliced among users. One of the reasons that it's so fast and so cheap at the same time is that we can take a compute resource that one user is using, and literally the moment they stop using it, we reassign that resource to another user. You can at Amazon too. You can have a dedicated instance at Amazon. They sell you dedicated. We will sell you dedicated instances. There are good reasons to have dedicated instances, because you don't want any noisy neighbors. Look, I want to know I get all the resources of this box, and for security reasons, and my own sense of comfort, and the fact that I don't want to compete for IO channels, I want this whole box dedicated to me. We will do that.
We will certainly allow you to do that, and we think that's an advantage. Our bare metal architecture is an advantage. We think it's an advantage not to have any of our software in the control plane running on your computer, which is a security vulnerability. One huge difference between us and Amazon is our control plane software runs on these separate control computers, these Cavium cards that we have. You as a consumer, when you rent a computer from us, you get the whole computer. When you rent a computer from Amazon, you get almost the whole computer except for the code that's running that belongs to Amazon. That is a security vulnerability. We don't have that. Our control plane runs in separate memory on a separate processor, not on the computer that you're renting. We will say, "This computer is entirely yours.
No one else is allowed to put anything on that computer but you, for security reasons. We offer that. Amazon does not offer that. Security is a very big deal for us, and this may seem like a trivial point, but the fact that we went to the effort of putting all of our control software on a separate computer so no one can get into your computer, we think is a very big deal on security. We've talked to a lot of people who are paying a lot of attention to security, that alone has caused them to say, "Oracle, once I can get this from you, this is what I want." Continue.
I think you may have answered it, the control is my computer as a customer of Oracle, but the back end, the database layer with the Oracle Autonomous Database can be a truly multi-tenant architecture across customers.
You have to buy the multi-tenant feature. We put multi-tenancy into the database.
Right.
Most people, if you go back to the early days of NetSuite and salesforce.com, those were the first two multi-tenant application companies on planet Earth. You had no choice because we didn't have a multi-tenant database. You had to put multi-tenancy at the application layer. Not a good place to put it, but you had no choice. When you have no choice, when it's the only place to put it's a good place to put it, because you put it there or you don't have it. We don't think multi-tenancy should ever be in the application. Though there's a reason historically why people put it there, because they had no choice. We think multi-tenancy should be down at the database layer, because we can then provide secure walls between the different consumers. That's not true.
A typical multi-tenant data application, there are lots of security vulnerabilities. Your data is co-mingled in the same address space as other customers' data, I'm not going to go into it. Ethical hacking. There are real problems that we avoid with a multi-tenant database, with our bare metal architecture, to make sure that people can't get malware into the same computer that you're renting from us.
Okay. I guess this is the last point on here. The existing customer base of Oracle Cloud, because the Oracle Autonomous Database, and maybe I'm misunderstanding, is something that is relatively new the way it's described today.
It's even worse than that. It's not fully released until December.
Okay.
It's running, and people can try it, but it will be production released, mass released in December of this year.
Okay. In the current customer base, are they essentially single instance, and you'll see that migrate over time to this greater efficiency of a multi-tenant across customers and the database layer?
At the database layer, Well, no. Oracle has a multi-tenancy feature. You can run multi-tenant. Let me see. I think you're asking multi-customer, which is slightly different than multi-tenant. For security reasons right now, we allow lots of tenants, but if you want maximum security today on Exadata, I'll just pick Exadata as the example. On Exadata, which is different than our bare metal architecture, Let me just talk about Exadata. If you want multi-tenancy today on Exadata, you can have it, but We won't give that to multiple customers because we think that is not secure to our standards. There are a bunch of things we won't allow in our data center. We won't set it up that way because of security issues, and we think we're making the right choice.
With the Autonomous Database now, we think we've solved all of those problems, You can mix customers on the same Exadata machine. Up until now, we have not been able to mix customers on an Exadata. We don't want to. Customer has to commit to kind of a minimum size Exadata machine. They get that machine. It's kind of a minimum fee if you want the Exadata service. That is not true if you want the bare metal service or if you want what Amazon has. A VM, one VM running on a machine that's shared with somebody else. You can have that on our cloud today. The only thing that changes with the Autonomous Database is now you can partition up an Exadata, not just conventional Intel infrastructure.
We have the same ability that Amazon has today to say, "You get a VM on this machine, and that's shared, and one customer gets a VM on this machine, another customer can have a VM on the same machine.
Maybe one more question, Larry.
One more question. Okay. Let me see. At the back there.
Thank you, Larry. This is Keith Weiss from Morgan Stanley. Oracle recently released a proxy with the named executive officer compensation program that caught a lot of investors' attention. Because essentially you guys gave five-year targets, and we've never really seen five-year targets out of Oracle. In particular, when we looked at SaaS revenues, $10 billion in SaaS revenues in five years, that aligns to the trend line that we're seeing in SaaS. That makes a lot of sense. Infrastructure service platform-
This is just one person's opinion. I think we're going to get there way before five years.
The infrastructure service platform as a service getting to $10 billion, that's harder. You're starting from a smaller base. It seemed to imply an inflection. What do you think causes that inflection in infrastructure service platform service? Like you're saying, database isn't going to migrate. It's not going to be a database migration. It's net new business that's going to drive that $10 billion.
Well, I think the Oracle Autonomous Database. How much of our workload? Most of the world's data is in Oracle. $10 billion is not a lot of money for something like that. This is a very different business than licensing software. It's a much bigger business providing all the computers, all the storage, all the network, all the labor. Think about it. We used to say, "Okay, here download this software and run it." You had to hire lots of people, and you had to buy all this stuff and communication stuff and all this. We're going to do that. I don't think $10 billion is a lot. I think we get there. This is early days for us in IaaS and PaaS.
This is the first time we have the entire database team now for a couple of years focused on the cloud. What we came up with was the Oracle Autonomous Database. That's really pretty good. Tiny fraction of the cost of Amazon or anybody else. By the way, tiny fraction of the cost of what it would cost you to run it in your data center. Let me just not pick on Running Oracle in your data center, we run it much cheaper. There's no surprise here. We're using Oracle Exadata technology. We're using our best technology. We have highly specialized labor. We've fully automated the whole thing to get the labor out. Automatic provisioning never goes down. How many of our customers do you think will take us up on that? They save a fortune. It's more secure, it's more reliable, it's faster.
No one else can do it but us. Yes, we are looking for an inflection point. There are two inflection points that I think are important. One is our bare metal offering, where, again, we can just look at where it's really apples to apples with Amazon. This is how fast our network is, this is how fast our storage is, this is how fast our computers are, all of that. Where we think we have a big lead on them, where we're cheaper and faster just on that. We can look at our purpose-built infrastructure like Oracle Exadata and all of the machine learning around it, and that's where we're not twice as good as Amazon, but 10 times as good as Amazon.
We think the combination of those two things, the Oracle Autonomous Database, bare metal infrastructure, is going to allow us to grow this business very rapidly. I really don't understand, unless it doesn't work, why one of our customers, an Oracle customer, would run an Oracle workload in a place that's going to cost them more, be less reliable, et cetera. I don't know why they wouldn't pick us. I expect a huge inflection point, everyone said, "Well, Amazon's so much bigger than we are," they're so much bigger than Microsoft and Google as well.
this is really early days of this, and we think we're so much better at this than they are that at least for Database workloads, for Oracle Database workloads, we're going to get the lion's share of it, and by the way, that's most of the data on Earth, most of the valuable data on Earth.
Thank you.
Okay. Thank you very much.