All right. Good afternoon, everybody. My name is Tyler Radke. Thanks for joining day two of the Citi Tech Conference. I co-head the software sector here. I know we got a super packed room. I believe most seats are taken, but if you have one open, feel free to raise your hand and folks can continue to filter in. We're happy to kick off the afternoon sessions with Microsoft. Really excited to have this discussion around everything Azure data management. With me today, I have Arun Ulag, who's the Corporate Vice President of Azure Data. Arun, I thought just to kick things off, for folks who haven't met you, everyone here knows who Microsoft is, but you've had a really interesting career, I think at Microsoft.
More than 20 years running, security, SQL Server, now Azure Data Platform. Give us an overview of your journey at Microsoft, and what are the areas you're focused on?
Awesome. Thank you so much, Tyler. Thank you for having me, and thank you so much for showing up. So I started my career as an entrepreneur in Silicon Valley, building a telecom network management start-up. Did that for six years, was an engineer selling to engineers. I sold it, went to business school, and I came to Microsoft because I figured it'll teach me a few things, and I'll go back to Silicon Valley and do another start-up. I really loved it here, and it's been 20-plus years. The last decade has been entirely in the data space. So I started running Power BI, which is our business intelligence platform from its early days. That's been really exciting. You know, Power BI is very, very widely used today.
A few years ago, I took ownership of big data analytics, and that led to the launch of Microsoft Fabric last year, and then very recently, I started taking on our operational database as well. So I run all of the data services for Microsoft, but engineering, product management, cloud operations. So that's what I do.
Awesome, awesome. And look, there's a ton of products within your purview, and particularly within Azure data, databases, data integration products. You talked about Power BI, some of the business intelligence products. Give us the vision for how all these products come together. What are sort of the elements of unification that you're working towards?
Yeah, absolutely. Before I talk about our product strategy, I'd love to just share a few thoughts about what we're seeing from customers.
Sure.
You know, every customer recognizes that AI is transformative for their business. They also recognize that, you know, they need their data estate needs to be in shape to be able to leverage all of the power of AI. Because the best AI models, if you put garbage in, most likely you're gonna get garbage out, right? So it's become critically important for customers to get their data estate ready for AI. Unfortunately, it's a lot harder than it needs to be, you know. I like to say there's good news and bad news. The good news is there's been a lot of innovation in the data and AI space. The bad news is there's been a lot of innovation in the data and AI space because it's created so much fragmentation and complexity, you know.
There's literally hundreds and hundreds of products and technologies out there, and the burden falls on customers. They have to figure out which products to use, which ones work together, how are they priced and licensed, and put them together to get to business value, right? And so when I talk to customers, the message I often hear is this, you know: "Please simplify. I want to be the chief information officer, not the chief integration officer. Help me unleash AI on my data estate," and that's exactly what we're doing here at Microsoft, so if you look at our approach, I would say it's three things. One is we have a very comprehensive portfolio of data products and AI products, right?
Everything from the world's leading operational databases with Azure SQL and Cosmos DB and PostgreSQL and MySQL, to big data analytic solutions, to data integration, et cetera. So we have a very comprehensive portfolio, that, you know, that we've invested in over a long period of time. The second is we're bringing all of these capabilities on the data side together with Fabric. And what we're doing with Fabric is we're really integrating this together so that customers get a unified product delivered as software as a service. You know, this is very different from the industry today, because most of the data components are available as platform as a service or PaaS components.
And if I were to use an analogy, you know, if you're using PaaS services, you go to a dealership and you want to buy a car, they give you a whole bunch of parts, right? And it's your job to go assemble the parts together and assemble the car before you can start driving. What we're doing with Fabric is all of it's available as SaaS, which means that the developer doesn't have to put anything together. They don't have to wire things together. They don't have to worry about infrastructure. They just focus on business value, right? And that really resonates with customers. The third thing that's unique about Microsoft is really our focus on business users, right? Because all of the data, you know, really is in the service of business.
Otherwise, what's the point of data sitting beautifully in a data lake or a database? And, you know, one of the things that we're doing is we're taking all of the data, and making it available to where business users are, whether that's Excel, whether that's Teams, that's Power BI, that's SharePoint, you know, all of the Office products, and making the data available in a secure and governed way. So, you know, this really resonates with customers because it gives them an ability to move quickly, democratize access, drive down costs, and really get down to business value.
I love how you led off with the customer example, because I think that's what a lot of folks in the room are sort of interested in. And I guess I'd love your perspective on sort of comparing this rush to modernize your data estate now relative to, you know, other data themes. Like ten, 15 years ago, it was all about big data, then it was about cloud. Like, what is different this time about those types of conversations with customers? And obviously, what's sort of different on the product side, too, in terms of Fabric and how you're approaching it?
Yeah. I, I, I'd say, hey, you know, today, AI is top of mind for customers. You know, it's not news to anybody here, right? Gen AI is front and center. And, you know, customers see the potential. They're trying to translate that into business outcomes, okay? And what it means for the, for us is, you know, we see a lot of opportunity to help customers accelerate their AI journey, by bringing the data estate together. When we think about, our strategy with Fabric, we really think about, you know, very similar to what we did, you know, if I were to go back a few decades, you know, there was a time in which, you know, we had Word competing with WordPerfect, right?
Then we had Excel competing with Lotus 1-2-3, and then we saw the opportunity was with productivity, right? And then we put Office together as a single integrated product. And then even today, decades later, Office has a very substantial market presence. And we see the same opportunity with data, right? We see really the opportunity to bring everything together into a single unified stack so developers can focus on business value. When we think about, you know, Fabric, we think about maybe four promises, right? The first is really a complete data platform so that customers get all of the capabilities they need, whether it's data integration or data engineering or data warehousing or business intelligence or data lakes, it's all available in a single unified product.
The second opportunity we see is really embracing what we refer to as lake centric and open. You know, embracing open source data formats. You know, one of the things that really gets in the way of customers is all the data silos that exist in organizations. There's so many different data silos, and every data product out there has their own proprietary data formats. So customers see their data locked into so many different places, which really makes it hard to get the business value. So what we did in Fabric is really embrace open source data formats at the heart of Fabric, right? So anytime you bring data into Fabric, by default, you build a data warehouse. By default, the data is sitting in Apache Parquet, and Delta Lake, and now we support Iceberg as well.
By default, it's open source data formats, and you know, customers really love that because it means that they don't have, they're not worried about being locked into Microsoft. They don't have to worry about shuttling data over and over again to just get it into a particular format for a particular analytics or data engine, so the second promise we thought of with Fabric is really a lake centric and open architecture. The third one is really about empowering every business user, so we built all the data and Fabric into every part of the Office experience, right?
So it's built into Power BI, it's built into Teams, it's built into Excel, it's built into PowerPoint, it's built into SharePoint, with all the governance flowing through, all the security permissions flowing through, because you want to put this data in the service of business, right? And that's something that is really, really exciting for customers. And the fourth promise is all of the AI capabilities that we built into Fabric. Every part of the Fabric experience has a built-in Copilot, as you would expect from Microsoft these days. But it also accelerates the entire data journey, but it also helps you build these new GenAI applications much more quicker than you otherwise would have. So that's kinda how I think about the, you know, the
Right
- areas we're investing in.
Got it. Got it. It certainly seems like a pretty compelling consolidation message for customers. I know that's a big theme that you're... we're hearing from IT executives. I guess I'm wondering how you sort of are positioning Fabric in the context of also some really strong partnerships with ISVs. You know, out at Microsoft Build, we were out there in Seattle and you highlighted companies like Snowflake, Databricks, MongoDB. How do you sort of think about where to partner versus, you know, pushing your own cloud-native services?
Yeah.
How does that work for customers if they're trying to consolidate?
Yeah, it's a really good question. I'd say, first, Microsoft is a hyperscaler, right? And as a hyperscaler, we have to do two things simultaneously. The first is making sure Azure is a great platform for ISVs, you know? So it's awesome to have industry-leading ISVs, like Databricks, like Snowflake, like Mongo, make huge bets on Azure, right? Because that makes Azure exciting for customers. That means that they can bet on Azure, knowing that they have choice of industry-leading ISVs. And as these ISVs bet on Azure, they also make Azure better. It allows us to improve our services, you know, provide more capabilities, et cetera. So that is number one for us. Number two is that, you know, as a hyperscaler, we also have a great set of first-party offerings that compete with many of these ISV offerings.
We're not unique in that. Other hyperscalers do it, too. But we do have really, really compelling offerings in market. And it's great for customers because they have really good choices, you know? And that's what we're doing with Fabric, that's what we're doing with Azure SQL Hyperscale, with Cosmos DB, with all of our, you know, our first-party services. The third thing is that regardless of whether it's third party or first party, we owe it to our customers to make all of this stuff work together. That's why, you know, at Build, I had Christian Kleinerman from Snowflake on stage with me, talking about how they're making a bet on Fabric as a native data store for Snowflake, right?
Mm-hmm.
We had Ali Ghodsi from Databricks on stage with me in November, talking about the partnership we have with Databricks and how it's integrated into Fabric, right, so we really do think about, you know, all three things. Azure is a great platform for ISVs, making sure we have great first-party offerings, and we have strong interoperability between our Azure services and the third-party services.
Right. Right. And I think with Fabric, it's been about a year and a half since the announcement out at Build in May of last year. What's sort of the customer feedback and reception been? Any adoption statistics you can share with us?
Yeah. The customer reception to Fabric has been really been absolutely incredible. Satya announced our last earnings call that at this point, we had more than 14,000 customers, paying organizations, pretty much, you know, from industry leaders in pretty much every geography. Companies like Accenture or, you know, Kroger, or, you know, Rockwell Automation or ZEISS, like, they're pretty widespread. So we're seeing adoption everywhere. We also, you know, had our first Fabric conference in March, which at that point, Fabric had been generally available for four months. We were surprised when 4,500 people showed up, right? Like, we didn't expect that kind of a reception.
We were one of them, actually.
Thank you. So we were pleasantly surprised, and you know, we also talked to a lot of customers about why they're so excited about Fabric, and you know, there's lots of reasons, but three reasons come, you know, show up. The first is just time to value. Right? Instead of putting all of these complicated things together, Fabric gives you all of the capabilities end-to-end, so you can build solutions much more quickly than you could ever do before. A great example is One New Zealand. One New Zealand is the largest telco in New Zealand, and they were able to build a solution end-to-end and get it in production in just three weeks with Fabric, right? This thing would have taken quarters or semesters or much longer in the past.
So the first thing we're hearing is just time to value, right? Really matters to business customers. The second thing we're hearing is costs, right? Now, let me go a little bit deeper here because I think it's worth understanding. If you look at the data and AI infrastructure, as you string multiple of these PaaS services or platform-as-a-service together to create a solution, each of them have different compute meters, right? So you're forced to create lots of pools of isolated compute, and they're not all running all the time. So there's a lot of wastage in the data and AI infrastructure. Now, what we're doing with Fabric is, all of the compute in Fabric, whether it's data engineering or data warehousing or Power BI or whatever the compute is, all compute is virtualized and serverless, right? So you just buy one thing, Fabric Compute.
So the same compute overnight could do a lot of data integration, maybe some data engineering, maybe some AI. And in the morning, as your people walk into the office, the compute flows to Power BI for business users and to SQL, right? This means that customers can consolidate a lot of wasted compute into Fabric and get much higher utilization, right? One example, a large consumer company in the U.K., they were spending, you know, about, publicly traded. They were spending about $165 million in their data infrastructure over five years. Most of it running in competitive clouds and competitive products. And they were able to move their entire stack from what they had right now into Fabric, right?
The cost, their spend goes from $165 million over five years to $45 million, right? They're able to put $120 million back into the business, right? Really massive savings for customers and obviously huge opportunity for Microsoft because our spend went up substantially as a result of the customer saving money. Number two, I would say is cost savings. Number three is really GenAI, right? That comes in two parts. One is because Fabric has Copilot built in end-to-end, so it really accelerates time to value. It's not just Copilot for a single link in the chain, it's Copilot across the data journey. Whether it's Data Factory or data warehousing or BI, you know, Fabric helps accelerate the entire time to solution using Copilot, and that customers really love.
The second part of it is really getting data ready for GenAI applications, right? Because Fabric on OneLake is built into Azure AI, you know, Azure AI Studio. So you're building these GenAI applications, and you need to get the data estate ready, you know, Fabric makes it possible for you, and it's just built into the AI stack as well.
Right.
So these are the things that we're hearing from customers about why they're so excited.
Right. And as you think about the cost savings, I mean, those are some pretty impressive numbers. Is that primarily, you know, cost savings from other, you know, hyperscalers and other competitive products, maybe some of those which you partner with? Or how do you sort of think about where that budget for Fabric is coming from?
I'd say, you know, again, it varies customer by customer, and, you know, and customers experience different levels of savings. So, you know, what I shared was one example.
Yeah.
In that particular case, the customer was on a competitive cloud, right?
Mm.
Using a whole bunch of competitive products, some of those products available on Azure as well. And for them, you know, they had a change. Like they had a new Chief Data Officer come in, and he looked at this and said: "Wait a minute, we're spending all this money on a data infrastructure." The second is not just the money they're spending on licensing, it's also the heavy investment in engineering to make all these different products work together, right?
Mm.
A significant chunk of his capacity was just, you know, on wiring things together, configuring things, monitoring things, making sure one product worked with another, et cetera, which doesn't accrue to business value. And the third thing that they heard from the CFO is that, look, if you want to make all of these AI or BI investments, well, you're not going to get a much bigger check. So you got to free up funds to be able to invest in these areas. And for him, you know, he found out about Fabric at the right time, which helped, you know, helped to make the choice. So that's, that's what happened in that, that particular case. But we generally see, you know, cost pressures exist in customers. They want to get more value. They want to accelerate time to value.
They don't want to always spend more, and we have some really compelling options for them.
Yeah, yeah. And it sounds like great scale for, you know, a product that's only about a year and a half old and some months.
That is true, and I would say that, look, most of the components of Fabric are not new, right?
Right. Right.
Like, we've been in data warehousing for decades. You know, we have been in real-time intelligence for a long time. We have been in Power BI for a long time. So as much as Fabric is new, it's built on a robust set of technologies.
Mm
... that have wide market adoption, that have matured over a long period of time, that are used at scale.
Right.
Yeah.
I'd love to hear from you just the significance of the engineering organization that you have working on Fabric, and then also what you're most excited about, maybe what customers are most excited about in terms of what's coming on the roadmap.
Yeah, it's a very big investment for Microsoft, you know, so it's a very big bet. You hear Satya talking about it, you heard Scott talking about it. So Fabric and our data stack is a big bet for us, and it's not new. We've been investing in the data space for decades now. So, you know, Microsoft has always been a big player, starting from when SQL Server launched, I guess, 30 years ago, right?
Yeah.
So, it's a very big bet. Now, when you asked about some of the innovation that I'm most excited about, okay? And I wanted to call out two things. One is OneLake, which is part of Fabric, okay? And, you know, if you look at most customer architectures, the data lake is a core part of a customer's architecture. It's the easiest place to store your data, right? But data lakes are pretty messy and complicated today. If you look at what a data lake is, it's nothing more than a bunch of storage accounts provisioned by developers, right? Which means customers have to write a lot of code, a lot of infrastructure, to make data lakes usable, not just for IT, but for business, right?
What we saw is an opportunity to simplify the entire data lake infrastructure. Use an analogy, like if you use Office, for example, you just let OneDrive deal with your documents, right?
Mm.
You don't have to worry about which storage accounts, where it lives, OneDrive, OneDrive just takes care of it. So that's why we introduced OneLake. In fact, we had to. We picked the name so that the analogy to OneDrive will be clear. But OneLake basically makes data lakes very, very easy to work. It not only provides storage out of the box, but it also allows you to virtualize storage and connect to storage data wherever it lives. You might have data living in other parts of Azure. You might have data living in on-premises systems. You might have data sitting in AWS or GCP, right? OneLake allows you to simply shortcut the data wherever it lives and provide business value.
Customers really love that because, you know, we don't begin every customer conversation with migrate everything to whatever my platform is first. We talk about delivering business value, but leaving the data where it is. So not only provides storage out of the box, but it also, you know, virtualizes storage and allows you to get to business value very, very quickly. So that's been a huge innovation for customers. Customers are really excited about it. The second area which we launched a public preview, it's not yet generally available. It will be in a few months, is what we refer to as Real-Time Intelligence. Now, as the world has gotten increasingly digitized, and there's a lot of data out there that is coming out in real time, okay?
It might be, you know, trucks sending telemetry, it might be NFC sensors in your supply chain, it might be, you know, applications providing telemetry, logs getting written. And generally, you know, real-time data is very hard to work with. It comes in very large volumes, it's semi-structured, the schemas drift over time. So as a result of it being very hard to work with, people don't use a lot of real-time data today. When you look around and work with businesses, you'll find that most of them are working with data that is in batch mode. You'll be lucky if it is from yesterday. Often it's a week old, it's a month old, it's a quarter old, right?
And back to my car analogy, it's like driving, you know, working with batch data is like driving by looking at the rearview mirror, right? But all this data is out there. So with Real-Time Intelligence and Fabric, we make it drop-dead simple. You don't have to do complicated things like spin up Kafka clusters or write Spark Streaming jobs. Fabric takes care of all of that for you, so you're really focused on business value. And just like OneLake, you know, we can work with events from everywhere, IoT events from your sensors, events from Azure, but events from AWS and GCP as well. You know, we make it really, really simple to work with it.
And I see huge opportunity to help customers work with data in real time, and it also substantially increases the value for customers and the TAM that we address.
Great. Great. I'd love to zoom out and just sort of get your perspective, since you've been in the space a long time, like, where we are in this data modernization and gen AI journey, right? You talked about how obviously you need good data to flow into LLMs and build copilots and everything. If we use the analogy of a baseball game, I know that gets, that's overused, but where are we in terms of that maturity curve of that data modernization? And then conversely, as you start to think about gen AI use cases, POCs, and everything, where do you think we are in that stage of adoption?
Yeah, I think on the data modernization side, I think we're well on our way, you know?
Yeah.
A fair bit of migrations have already happened, but customers still have a substantial part of their data estate still sitting on-prem. So we still see a lot of opportunity ahead to help customers migrate. And AI and gen AI is now providing more incentive for customers to accelerate their migrations, you know, to the cloud. You know, in addition to migrations, we also see some additional opportunities. The first is a lot of customers are just interested if they're not ready to migrate today, they still want to connect their data estate to Azure, to the cloud. So for example, we have Data Factory and Azure and Fabric.
It allows customers to connect their on-prem estate into Fabric, into Azure, and get value without even before they do their migration, right? We have hundreds of thousands of these gateways already deployed by customers. The other thing that we're seeing is a lot of interest in Azure Arc, right? Now, Azure Arc is our multi-cloud, you know, on-prem management solution, where you can get Azure value, where your bits already run on premises or multi-cloud, without necessarily before you migrate to Azure, right? I think Satya announced in our last earnings call that we have over 36,000 customers using Azure Arc, which grew 90% year over year. So we're really seeing a lot of interest in customers connecting to Azure even before the migration.
The last thing I'd say is that, hey, these Gen AI projects, you know, there's a lot of them going on. While I would say a year ago, there was a lot of tinkering and doing something with it just to learn, now customers are becoming a little bit more thoughtful about where the ROI is, what is the time to value, how does it change their business process? And some of these things, you know, matter. And this is where I would say, out-of-the-box solutions like GitHub Copilot or M365 Copilot, you can say, hey, you can get to business value without doing a lot of work, right? So those things are attractive to customers, and they're being a little bit more thoughtful about, you know, prioritizing the AI investments based upon business value-
Mm-hmm.
... than maybe a year ago, there, where there was a lot of experimentation going on.
Right. That's helpful color, and obviously GitHub and Copilot is kinda off.
Yeah
-and this is for you a big, big success story. I'd love your perspective, just as you think about some of Microsoft's kind of traditional data products, be it on the data warehouse side with Synapse, maybe it's called Fabric now, I don't wanna-
It's Fabric.
But just as you think about the role of, say, a traditional data warehouse or a transactional SQL-based databases, as we start thinking about emerging gen AI architecture, LLMs, a lot of JSON documents, right? Vector databases.
Yeah.
What role do those play? I mean, do you still think there's a huge growth curve ahead for data warehouses and traditional databases, or do you think we're sort of entering this massive tectonic shift?
... Yeah, it's a good question. When I look at the application patterns, when I think about database, you know, transactional operational databases or data warehouses, I kind of feel like they're evolving more quickly today, Tyler, than they ever were evolving before. And the reason is that, very quickly, you know, pretty much every AI application- every application is becoming an AI application, right? And guess what? They're using operational databases or data warehouses or other parts of the data stack. So they're evolving more quickly today than they ever were evolving before. I see there are, you know, four areas in which the evolution is happening, right? The first one is not new. It's just, you know, reliability, security, transactional performance.
Those things still matter because these applications have to be reliable, they have to be secure, and Microsoft has a bit of an advantage there. Like, if you look at SQL, you know, it has a lot of security capabilities. You know, we have customers using very large transactional volumes with very strong ACID guarantees. If you look at Cosmos DB, we provide single millisecond performance. You know, it's used by ChatGPT, the most, you know, the fastest growing consumer app in history. So I do think there's a lot of value that we continue to deliver in terms of just you know, security, reliability, the core. So that's number one. Number two, when you think about these AI applications, we see a huge opportunity to bring the AI capabilities into the database itself, right?
So you'll see, you know, you talked about vector databases. You know, it's required for LLMs. The RAG pattern, the retrieval augmented generation, is how people often build AI applications. Well, vector databases are now available as part of the database itself. It's built into Cosmos, it's built into Postgres, right? So you don't have to take the data out. It works with the database. It makes it easier to build these applications. In Fabric, we launched something called AI Skills that allow you to interrogate your data lake or your data warehouse with just natural language and get those results in an application, right? So we really see the AI capabilities now being infused into the data stack, which is, again, a huge leap forward for customers.
The third thing we're seeing is really the fact that developers are getting more and more impatient, right? They want quick time to value.
They've never been that patient.
They're less patient today than they were, let's say, because they don't want to waste their time. They don't want to spin up clusters, they don't worry about infrastructure, they don't worry about networking, storage. Like, these things should just work, right? So we're seeing a lot of interest in SaaS-based platform, software as a service. If you look at Power BI, we've been SaaS from day one, like, when we started launch like nine years ago. We have 350,000 organizations on the planet that use Power BI, you know, by far the world's leading BI service. If you look at Fabric, it's all SaaS, right? So we're seeing a lot more receptivity for developers for SaaS-based platforms because it accelerates time to value. It helps them get to those business outcomes much more quickly.
Right.
The last thing I'd say is really the integrated data estate. Because if you look at an enterprise customer, they're not happy with all these silos of data, all these proprietary data formats, all these walls that exist between different applications, different data warehouse, different data lakes. They want all of this thing to come together. They want all of this to provide business value, regardless of which cloud it lives on. And that's why we've been thinking about OneLake and really embracing multi-cloud, and helping customers accelerate to business value. So, you know, to your question, we see databases and data warehouses evolving much more quickly. These four areas are perhaps the biggest areas where we're seeing a lot of innovation.
Right. Right. So we talked a lot about how, you know, a lot of positive receptivity to Fabric, and we're well along this data modernization journey. I'm curious on sort of the constraints, right? What are the biggest roadblocks that you see for customers in terms of not being able to go down this data modernization journey? And then also on the GenAI front, we talked a little bit-
Yeah
... about prioritization, but would love your perspective on those constraints.
Yeah. I mean, customers are doing this extensively, and it's, you know, if I think about constraints, I'd probably call out three areas. The first is just complexity. You know, building GenAI applications is not easy. These are new application platforms. You know, not a lot of developers are familiar with them. Lots of services to string together, learning how these things work, lots of work on that front. So complexity is something that customers are wrestling with. The second thing is really cost, and we talked about that a little bit before, but these things have to... You know, there's not massive new budgets that for overall IT spend, to go invest in all of these projects. If this thing goes up a lot, something else has to adjust, right?
So customers are looking for cost savings and making sure they can fund these initiatives, and the third thing we see is really responsible AI, right? Because these LLMs are new, they behave in different ways, often unexpected, right, but you want to make sure that the LLM that's providing customer service for your customers is respecting your brand promise, is safe, is not hallucinating, is not, you know, providing output that, you know, that offends your customer, right?
Mm.
And that's where we're making a lot of investments on the responsible AI side, to do this in a way that helps customers. So, these are the biggest areas where I find that, you know, it does create friction for customers to overcome, to be able to go on their AI and GenAI journey.
Got it. That makes sense. I'm curious, so as we sort of connect all the dots together, we talked a lot about Fabric and sort of the whole Azure, you know, data platform, which I think is really unique versus anyone out there. How does this sort of all add up to your, you know, Microsoft's overall GenAI strategy in your competitive position? Can you sort of bring it all together and just talk about how this ties in with the goals around-
Yeah
... GenAI?
Yeah, I think overall, I think Microsoft is in a very, very strong leadership position. And when I think about our competitive advantage, I'd probably call out four areas, right? Some of which we've talked about before, but I'll just recap it for you. The first is that, you know, we have a very complete data and AI stack, you know. And the data and AI stack that we have is built in-house. It's not an acquired collection of pieces that somehow, you know, don't work together, right? So, because they're built in-house over a period of time, they've had a chance to mature, they're engineered to work together, and we're bringing it all together with Fabric. So a really complete data stack that we're bringing together with Fabric. The second is really a GenAI leadership, our AI and GenAI leadership.
And, you know, it's very extensive, right? Because we have a comprehensive set of GenAI models, everything from the leading frontier LLMs to small language models, to thousands of models from third parties, all being available in Azure with responsible AI. You know, all of those things are a very extensive set of customer assets that customers really, really appreciate. But it's not just about the models we offer. We also have very strong first-party services that use these models, things like GitHub Copilot or M365 Copilot or Dynamics Copilot, you know, which gives us an opportunity to add value to customers, but it also gives us an opportunity to learn, right?
Mm-hmm.
And inform what kind of tooling we need to provide for others to do the same thing. So I do think that we have very significant strengths on the, you know, the AI and the GenAI side. The third thing is really our advantages with developers and developer tools, right? The most popular tools on the planet are the ones that Microsoft runs, whether it's GitHub or, VS Code or Visual Studio. We have well over a hundred million developers building on the Microsoft tools.
Right.
Many of them are fans. They love the tools that they get to work with, and we have a really big opportunity to connect this tooling to our Azure infrastructure, make it easier for them to discover these Azure services and consume them, leverage the copilot experiences to guide them appropriately, so lots of advantages with developers. The last one is really an advantage with business users because eventually, as we talked about, all of this data is in the service of business.
Mm-hmm.
And most of the business users on planet Earth are using Excel, they're using Teams, they're using Office, they're using PowerPoint, they're using SharePoint, they're using Power BI, and we have the opportunity to use all of this data to inform all of these business users.
Yeah.
So these, I would think about, are, you know, sustainable competitive advantages-
Yeah
... as the world is transitioning to the era of AI.
Mm-hmm. Right. And as we think about that positioning as, you know, one of the best-positioned hyperscaler for the world of AI, certainly the numbers from Microsoft have been very impressive, with Azure growth north of 30% and commercial bookings, you know, up in the high teens last quarter. One of the things we've been hearing is just how sort of the AI positioning or GenAI positioning is actually just contributing to more share gains within the core sets of services. I'm curious, is that something you're seeing, and how significant is sort of the Azure data platform as part of these commercial booking strengths that we've seen?
Yeah, you're absolutely right. We are. You know, as people build these GenAI applications, they're not just using the LLM in isolation, right? You know, they're using Cosmos DB, for example, to store the chat history, right?
Mm-hmm.
They might use Azure AI Search to be able to provide back to search on top of the database. They might unify the data in OneLake across clouds to be able to feed it to the LLM, right? We're seeing a lot of growth associated with the AI workloads because it's not just about the AI workload. I think we shared in our last earnings call that we have over 60,000 customers using the Azure AI services. We also shared that about half of those. The growth of Azure data services among the customers using our AI services grew 50% year over year, right?
So we do see substantial opportunity as customers are building these AI capabilities to not just, you know, provide the LLM, but provide all of the services that help the application, you know, get business value. So-
Right. Right. One of the things you, you mentioned earlier is, is, you know, honing in on that developer focus, and obviously, developers are, are even less patient these days-
Mm-hmm
... in terms of things they demand. As you sort of think about, like, making sure that you have relevance with that next generation of-
Mm-hmm
... developers, I know that's always been a focus for Microsoft, but what are some of the things that you're doing in your organization to make sure that you don't lose that momentum?
Yeah, you know, it's just... When I think about the developer, especially in the data and AI space, it's just such an exciting time. Like, there's so much innovation, there's so much energy, and it does feel, like, very energizing. When I look forward into the biggest opportunities where we're investing, I'd probably look at it in maybe three areas, okay? The first is really AI-powered development.
Mm-hmm.
This is, you know, this is about GitHub Copilot, it's about the Copilot experience at Fabric, the Copilot experience built in the Power Platform. You know, you can really dramatically accelerate time to value. You can train legions of new developers, you can help them write code, build solutions, document it, have the AI guide the developer every step of the way, have the AI guide the, you know, the debugging experience, the documentation experience. So we see tons and tons of opportunity just in terms of the AI-powered development. The second area that I would say I'm really excited about, especially my team is investing quite a bit, is really driving, you know, an AI-powered data estate, right?
Mm-hmm.
We talked about OneLake and how it's abstracting the storage systems, but why wouldn't we bring a lot of the AI capabilities into OneLake itself, right? Why wouldn't we bring that into the storage system and help developers? Just say, it's just built in, you know, there's nothing to string together, there's nothing to connect, it just works. It's a core feature of the data stack of the storage system itself. So we do think there's a lot of opportunity here to AI-empower the data estate, you know, and accelerate time to value for developers. The third thing I really think about is less about developers and more about business users, right? And there, I think about the opportunity to provide AI-powered insights.
If you look at most of the business users on the planet, very few of us can afford to have a business analyst who can analyze all your data and tell you what your data means, right? Most of us cannot, right? But here's where AI can step in: so it can look at the data, it can analyze it, it can find those insights, it can coach the business users to making better decisions and better outcomes, right?
Mm-hmm.
And I see huge opportunity for us to add business value, for all the business users, and there's, you know, hundreds of millions, if not billions, of these people out there, that can make better decisions based on AI. So those are the three that that I'm most excited about: AI-powered development, AI empowering the data estate, and, you know, AI empowering the business user.
Yeah. So, bigger picture question, obviously, GenAI has bursted onto the scene-
Yeah
... over the last couple of years. It's been pretty transformative to watch. As you think about the next five years, what are some of the things that you're excited about? What are some of the things that maybe you think the market is not talking about, whether we're too distracted with GenAI or anything else? What are you most looking forward to?
I think, you know, just I would say the biggest thing that I'm really looking forward to is really democratizing access, you know? The world is very large. We have lots of people in different countries and different organizations with different economic situations, lots of companies with different abilities to invest, and we, you know, we need to bring everybody along. You know, this cannot be just a technology for the top of the top of the pyramid. So I do think that Microsoft, with our mission, with our scale, with our ability to think about democratizing access, driving down costs, getting these, you know, capabilities into production, empowering business users, has a really unique opportunity to really lift the whole planet up. So that's-
Mm-hmm
... for me, that's the part that I feel is the most exciting, on the journey that we're all on together.
Yeah, great. Well, I know we got a couple of minutes left, but I did wanna, you know, turn it back to you if there was anything that you wanted to cover that we didn't address, and, you know, maybe just leave the audience sort of the main takeaways and what the focus is from your end and your end.
No, I think you covered a lot of ground here, Tyler, so I don't know if I have a lot more to share, but I just want to reiterate, it's just such an exciting time. There's just so much innovation happening, so much opportunity with customers, and I'm really, really excited about the opportunity ahead, so thank you.
All right. Thank you very much.
Thank you.