Good afternoon, everyone. Thank you for coming to Snowflake Summit 2022 and our Investor Day. It's good to see everyone in person after a few years, and we hope you enjoyed all of the product announcements this morning at the keynote. Before we dive in, I'd like to note that we will make forward-looking statements during today's presentation, including relating to our long-term operating model, market size and growth as future product releases. These statements are subject to risks and uncertainties, which are further detailed in our Safe Harbor provisions. In addition, we will present both GAAP and non-GAAP financial measures. Non-GAAP financial measures are presented in addition to and not as a substitute for GAAP measures. A reconciliation of historical non-GAAP measures is provided in the appendix of today's presentation. Now, let's dive into the good stuff.
Today's program is really focused on our market opportunity and how we are developing a new market. This market seems to be misunderstood, you know, for a number of reasons, and so we wanna help you understand how we are viewing it. There's a few different strategies that we're using to develop this new market, and we're gonna dive into those today. First, you'll hear from our Chairman and Chief Executive Officer, Frank Slootman. He's gonna talk about mission alignment. You heard him this morning at our keynote talk about the importance of aligning our customer's mission with our mission. Next, you'll hear from our Senior Vice President of Product, Christian Kleinerman. He's gonna talk about how we're creating this new opportunity through workload enablement and through product innovation.
Workload enablement is really focused on bringing on as many data types and workloads to the Snowflake platform to address all of our customers' needs. Next, our Chief Revenue Officer, Chris Degnan, will speak with Thomas Mazzaferro, the Chief Data and Innovation Officer at Western Union, to talk about how we are aligning with their mission in the financial services industry. Next, you'll hear from our Chief Financial Officer, Mike Scarpelli. He's really gonna talk about how we're executing against this opportunity and then close with the puts and takes of the long-term model that you heard on our last earnings call. Lastly, we'll open it up to executive Q&A with Frank, Mike, and Christian. With that, I would like to welcome to the stage our Chairman and Chief Executive Officer, Frank Slootman.
Well, welcome. Always appreciate the opportunity and to have this kind of conversation. As well as, you know, seeing you in person, I'm only used to seeing you on screen.
You know, when we got together and tried to figure out what we thought we should talk about that would be productive for you in terms of understanding, you know, and as well as you know appreciating the business and you know what I specifically was gonna talk about, I sort of wanna you know talk about what's sort of top of mind for me, you know, sort of at the most strategic level you know how we think about the business. Because the interactions that I have on a day-to-day basis now and really over the last year, year and a half are very different from what they were in 2019, 2020. I'm gonna try and characterize that a little bit for you.
It's been actually amazing that you know just going to the conference you know having conversations with people and just hearing the most amazing stories of the things that people have done and what they have accomplished the business impact that it has and you know maybe I'll tell you a few stories. So as most of you know you know we came to market in whatever end of 2014, 2015 really connecting with the opportunity to modernize legacy data warehousing workloads. I hate to talk about data warehousing because it's a historical concept. You know it's sort of a definition of things that is really not that current anymore.
I mean, we're really into a mega market around, you know, data operations, data management, and it's the boundaries or the categories or whatever you wanna call it, you know, they're not so distinct anymore because, you know, it's really, you know, one giant thing. How many of you watched the keynote this morning? Just a show of hands. Yeah, quite a few of you. You know, you've seen some announcements that we've made there, you know, around Hybrid Tables. You know, we're gonna blow away, you know, all the distinctions that historically, you know, existed, you know, anymore. It is with data warehousing.
It irritates me to no end when I see in the press, you know, we're getting called a data warehousing company because it just hasn't been true for a long time. It's really something that's of our own doing, because when we first came to market, the easiest thing that we could do to connect with opportunity is go into large enterprises, you know, talk about the systems that they had, the problems that they had, whether, you know, we would benchmark them side- by- side.
Because of, you know, all the things that we could do with elastic provisioning and so on, I mean, we would jump the ingest rates, you know, by an order of magnitude or two orders of magnitude and just blow people's minds in terms of the absolute contrast that existed between these on-premise systems, you know, versus what could be accomplished, you know, with Snowflake, you know, running on the public cloud. We loved it, and we're gonna continue to do this till the end of time, right? That is sorta, as I said this morning, that is not the end of the story. It was sort of, you know, the end of the beginning. We're now, you know, moving beyond that.
Certainly my conversations, you know, I used to talk about, you know, database migrations, cloud migrations, all the problems. Wasn't terribly exciting, but that was just the nature of our business. Those were the conversation that we've had. Now, it's a rare week that, you know, I end up talking about that because our conversations have moved on, they have evolved, and people wanna talk about stuff, you know, that is along these lines, mission alignment. You know, people are calling us, and they wanna talk about their stuff. They don't wanna talk about, you know, our stuff. You know, historically, our people were always gushing with, you know, what great architectural advances we've made and what all the benefits were. People's eyes would roll over because that's our world, it's not their world, right?
They wanna talk about their world to us, and how do we, you know, bring that together. You know, quick anecdote. I can't drop any names because I just heard it. This was a CIO who was here, speaking to her this morning. This is a very large diversified industrial company that everybody would recognize the name from. They have you know, huge amount of you know, backlog in terms of industrial you know, products you know, going through their supply chain to customers. They were able to you know, reprice you know, work in process backlog you know, through Snowflake. That in other words had like 56,000 you know, line items that they had to reprice.
Just the amount of data operations that they had to run to be able to achieve that accomplishment in a short period of time, you know, ended up saving them some $800 million, right? Which sounds like real money to me. It's that kind of stuff, right, that you become absolutely utterly, you know, mission-critical. You know, years ago, we had a chat with a large insurance company that was also on stage this morning, so you can figure out who that is. That was my first huge wake-up call, you know, when somebody said, "I have no interest in all these architectural conversations. You know, we're an insurance company.
You know, every month and every quarter when we do the analysis, you know, we have, you know, glaring differences between one zip code and another. You know, we're losing money here and making money there, and we're not understanding it. We don't know what we can do about it, whether we should raise our prices or redefine our policies. You know, insurance companies are data businesses to begin with, right? They're actually quite sophisticated, you know, around data already. But it's like, look, they're looking at a map like this, and it's like, you know, there's so much anomalous behavior, understanding it, reacting to it. You know, that really is what they need to do. The conversation is shifting to their core business, you know, away sort of from our core business.
You know, what we knew how to do well. It gets, you know, much more profound. You know, I mentioned the example of Novartis, you know, this morning. You know, they are just hell-bent on accelerating the time- to- market when they do drug development, not just because the drugs themselves are incredibly important, you know, that they reach the ultimate beneficiary, you know, but also because they get, you know, much more time to monetize the patents, right? If you get an extra one, two, three years to monetize the patents, you redefine the economics of an entire industry, right? The problem is, in clinical drug development is that they're highly collaborative in nature. In other words, they have, you know, academia involved, they have consultants, they have other pharmas.
They have all these parties that have to collaborate, so data sharing, you know, needs to be there. Because if we're gonna start, you know, flip-flopping files around to people, I mean, it becomes this unmanageable situation. They also need to be able to run, you know, concurrent processes, you know, against data. In other words, you know, you're running parallel as opposed to in sequence, which is what pushes out the timelines. They're very serious, you know, about driving themselves down from 12 years, which is their average, you know, to nine years, right? The potential impact on whole industries, you know, of these kind of operations is incredibly important.
You know, I had a chat with the chief science officer at J&J not too long ago, and this is an absolute pharma, you know, guy, life sciences type of guy, a doctor actually. He said, "Look, you know, the impact of data science on healthcare will be far greater than the impact of life science on healthcare in the next, you know, whatever, 10 years." Which is quite a statement coming from people, you know, whose whole, you know, career has been on the life sciences side. In part that is because there's this enormous amount of, you know, low-hanging fruit. You know, obviously, this is just unbelievable. Right now we're living through the supply chain upheaval that we're experiencing between Walmart, between obviously Target was caught flat-footed.
I'm actually on the board of a tiny little startup supply chain management company, mostly because I'm really interested, you know, because supply chain management is something that has never been platformed, right? You know, we have Workday and ServiceNow and SAP and all these things. But supply chain management, we're pretty much an email, you know, spreadsheet operation, and it just sucks. It is the absolute worst, right? If you live in an environment where, you know, things are an inch to the left, inch to the right, you know, you can kinda anecdotally observe, you know, what the hell is going on. Yeah, you'll be wrong, and it's costing you money, but it's not gonna kill your company. You know, now it's different.
You know, first the pandemic, you know, caused enormous supply chain upheaval. I remember talking to hospital supply chain companies like J&J, and they said, "You know, we send surgical kits, you know, to hospitals, and then you have, you know, an explosion of COVID, and they're not doing these procedures anymore. So all that product is going left, right, and center, and it's all in the wrong place." The problem is maybe you can get it right, but then the next day it's wrong again, right? Because the circumstances have changed. You can no longer get by by just sort of anecdotally observing what's going on. I call it sort of steering the ship by its wake, right? Because they're basically observing recent history, and then they try to extrapolate that as if that's, you know, reality.
You just can't do that anymore. The spiking inflation does the same thing, right? People are shifting their buying patterns, right? They stop buying the discretionary stuff and shifting, you know, their spending to what they have to have. All of a sudden, you know, wrong products, you know, wrong place, wrong time, you know, all that kind of stuff. To start to data drive and to sort of parse reality and, you know, through data is going to become just standard, right? It's just really hard as an economy and these industries to get there really quickly because it's not just data platforms, you know, like Snowflake that are incredibly capable. It's also the people that can really, you know, master data sciences and really drive these kind of outcomes.
Everything you saw this morning, you know, this is our, you know, strenuous effort, you know, to drive technology into the marketplace to enable people, you know, to do that, and not just hardcore data scientists, but also people, you know, with more moderate, you know, development skill set to be effective. You know, the whole acquisition of Streamlit, you know, if you saw Adrien's presentation is. I mean, they're trying to make, you know, machine learning consumable by non-machine- learning type of people. That's the whole point of it, right? Otherwise, you can do a heck of a lot of, you know, training models and all this kind of thing, but nobody can consume the output, you know, of that kind of work or even understand it, right? This in our mind is just incredibly important.
Now, ironically, Kraft Heinz is a great customer of ours, and they're actually, you know, as a CPG, you know, they're quite sophisticated, you know, in terms of, you know, hitting the correct inventory levels in terms of not too much, not too little. They've really gone to much higher levels of sophistication around managing supply chain. It's pretty cool. You know, I have many hospital examples of people that are trying to save lives. You know, I had one hospital meeting with a CEO there, and actually Christian was there as well, you know, and I thought I had a good story for him because, you know, Anthem is a really big customer of ours.
Anthem says, "Hey, we're paying those guys $1.5 billion a year, you know, as a payer, you know. It's a monstrously painful process because the data is slow to get here, and the data is dirty, and the data is incorrect, and the data is missing. I'm sure they would be interested, you know, if we were both on Snowflake. My God, that would just, you know, speed up the process enormously." The hospital CEO wasn't interested, you know. Kinda weird, but hey, he wasn't interested. He said, "Look, we're a hospital. We're into saving lives. We're into, you know, quality of life. We're into, you know, all these, you know, very, very high-minded, you know, things." They're into cardiovascular disease. They're the biggest institution in the United States that deals with cardiovascular disease.
He says, "Look, we have generations of data, right? Clinical and genomic and diagnostic and demographic. I mean, massive amount of data. It's like, what we wanna do is, you know, use the data resources that we have and create predictive outcomes. In other words, if we can figure out who is going to get cardiovascular disease at what time with what probability, right, then we have, you know, the beginnings, you know, of a strategy to go and deal with that, right?
Because then we can also, through data, you know, develop the certainty around protocols that will work for that particular person, you know, given their unique makeup and so on, and know the probabilities that we can actually, you know, either stave off disease altogether, you know, or treat them early or treat them later or whatever it is. That's what we mean by mission alignment. People think very differently, you know, around data now than they ever have. You know, we talked about cybersecurity this morning. You know, cybersecurity is fundamentally a data problem. Supply chain management is fundamentally a data problem. As you get into these conversations, you know, you start to realize these are all data problems, okay?
I think it's really important, you know, for you to have a little bit of appreciation for the fact that this is not your grandfather's data warehousing company anymore, okay? I mean, the scale and size of this, I mean, I've had people say, you know, "If I can solve this problem, I have $2 billion to spend." You wanna try and size the market? I have no idea. I know that if we can do it, the market will be there, right? I mean, there is just an insatiable appetite at this point, you know, for achieving and pursuing, you know, these kinds of initiatives. We're at the beginning, you know, of, you know, Obviously, we have a ton of partners here.
I hope you get a chance, if you're in the conference, to walk the partner pavilions, because a lot of our partners are very, very heavily invested, you know, in trying to insert themselves, you know, right, in the middle of the problem that sits between data and envisioned, you know, outcomes. That's really the opportunity that I think, you know, we're gonna talk about this more and more and more so that we give you an opportunity to sort of, you know, also sort of include that kind of thinking in when you look at companies like Snowflake and you advise, you know, investors to buy side on what does it all mean, right? Because this is a very, very rapidly developing, you know, thing.
Now, everything that we talked about this morning, and Christian is gonna rehash some of it here in a minute, or even less 30 seconds, it doesn't happen unless we enable these workloads. We can have all the data in the world, right? If the workload enablement and the tooling is not there, if we cannot run at performance, you know, with the economics, the ease of use, all these things, and we can't serve all these different constituencies, you know, people that are data engineers, people that are programmers, everybody in between, then it's just not gonna happen. People are just gonna take data out of Snowflake and, you know, put it somewhere else. This tool, that tool, every other tool.
That's why you've seen such an incredible heavy focus from Snowflake literally from day one, you know, on doing exactly this. You know, today, in terms of Hybrid Tables, Native Application Framework, we're trying to unleash the whole world of software development, you know, on Snowflake, right? That is. You know, I think Dave Vellante from theCUBE, I don't know whether you read his stuff or not, he referred to us as a supercloud. I don't use that word because, you know, I get in trouble with the other people, or I think I could get in trouble with other people. We try not to be too presumptuous about what it is. It's much bigger than the Data Cloud, right?
Because a Data Cloud is essentially a database in the cloud that use, you know, standard database drivers, but most of the work happens outside of that cloud. We're bringing all the work into the cloud. Christian did a great job this morning really articulating why that's so important. Why don't I invite you up here, and you can, you know, sort of, you know, maybe make that argument again for this group, because it's really important because that's the crux of, you know, what we're really trying to do here. I need him to give you this one, so all right. Thank you.
Okay. Hi, everyone. Great to see you all in person. Some of us have met also on virtual calls, but it's cool to be back in person. If you didn't see it from this morning, I really love the interactive dialogue and feedback, so look forward to our dialogue. As Frank said, I will do a little bit of recap of the larger announcements from this morning, but also put it in context, and I try to anticipate some of the questions that may be top of mind for you. Of course, we have Q&A for whatever I do not cover. Now, we'd like to start with a recap of what is the Data Cloud for us.
Frank just talked about it, but for us, the thesis is that delivering amazing technology for customers is insufficient. We believe that better decisions, better outcomes, better services, better applications are gonna happen if companies can collaborate with data, exchange data. You see it. Actually, the financial services industry is one that knows it well. You take all this alternative data, and you improve investment thesis. That is applicable to pretty much any business. Super interesting is pretty much any business also has interesting data. Complete side effect of whatever is their primary business activity, but the friction has been too high to collaborate, monetize, distribute, and that is what's behind the Data Cloud for us. Of course, I said awesome technology is not sufficient, but you do need the awesome technology.
This is as of the announcements from this morning, our revised platform diagram. We will pride and emphasize as much as we can that Snowflake is a single product with a single engine, and it's very intentional. Each time that you hear everyone else saying, "And here we have 20 new products," that's opportunity for us 'cause who's gonna integrate those 20 products? We do a lot of that integration work ourselves so that our customers don't have to do it. Obviously, the economics work out better. We spend time do it once as opposed to 6,000 customers doing it 6,000 times. The Snowgrid layer was covered in the keynote by Benoit this morning. Very important to us.
Everyone these days claims, "I'm multi-cloud." What that means is I took some VM, and I put it to run on all three clouds, and I'm good. Or four clouds or five however many you wanna count. For us, multi-cloud and cross-cloud, it's truly— We did a lot of engineering to make sure that Snowflake is Snowflake is Snowflake on all three clouds. If you ever wanted to go deep into how different the clouds are, there's a lot of nuance so that you can deliver a consistent product experience, and we did that with Snowgrid. It's that abstraction layer that not only normalizes the experience for our customers, but interconnects them. We have the three foundational workloads: data warehousing, data lake, and Unistore, and we'll talk more about Unistore.
What Frank was alluding to, we will continue to enable additional use cases and workloads because that is how we prevent data from being copied outside of the governance perimeter. Probably the question I hear most often, especially from folks like yourselves trying to understand the business and trends is like, "Well, but everyone already caught up to Snowflake. Are you guys done? How are you gonna continue going?" I created this taxonomy to assess how we think about it, and feel free to. We can engage in dialogue and disagree on the different assessments. But we see a real progression on the technologies generations, and we think that what we're doing with the Data Cloud is at the forefront. Started with a single engine.
Like, the more everyone else in the industry adds additional products, additional ways to do the same thing, the more they are diluting their own efforts and the more they're complexifying customers' lives. The topic of all data, I would say yes, everyone check mark. That sort of was solved. That's easy. Governance is a complicated topic, especially if you focus on business continuity. Some of the items we were talking about this morning at the keynote. Pipeline replication and pipeline continuity, those are really, really hard problems. If you really believe that at some point any one cloud product can have a global outage, then how do you do cross-cloud business continuity? We do that. We enable that. From what we said this morning, we're doing even more to simplify those use cases.
The notion of cross-cloud, very important to us, is gonna come up when we talk about native applications, because it is really difficult, as I just alluded, to implement one cloud, let alone three or four or five clouds. Topic of self-managed. All of us in the database industry are— It's a small world. We sort of are all friends, or we all know each other. I hear consistently from our top competitors, also friends, they're like, "The level of ease of use that Snowflake has delivered is unprecedented and really difficult to catch up to." Of course, we're not standing still. We continue to invest. Programmability, that's sort of where everyone has been.
If you look at, zoom out, the competitive landscape is a lot of people can program to all data, but there are all these other dimensions, and we think that marketplace monetization is a key differentiator that enables the Data Cloud. When I say that companies can have interesting data as byproduct of their primary business activity, if you tell them, "Do you wanna create a new business or a new revenue stream?" They start thinking through, "Oh, I need to price it and distribute it and sell it, and." It very quickly becomes a distraction. That's how strategic the marketplace and monetization efforts are.
On top of all of this, something that we put a lot of energy on, but I think it's also highly, highly differentiated, and in many conversations it's not apparent, is the fact that we deliver an opinionated view of the technology stack. We do not give customers every single choice they want, and trust me, they ask a lot, but we know that we can deliver a better experience and better abstractions if we control those. As an example, I talked this morning about large memory instances. All that we're giving is regular memory or large. We're not gonna go surface the 300 VM SKUs that each of the cloud providers has, 'cause once you wanna make it portable, the problem gets complicated.
We love the fact that not only we continue to advance the platform, but we do it transparently for our customers. The framework in how we think about, in general, our initiatives, and in particular the announcements from today or throughout the conference, is three categories where we are just doubling down on what we have. How do we make what we have even more differentiated? We're also expanding our market opportunity through the three that you see, native apps, cybersecurity, and Unistore, and we'll talk more about it. At a very high level, assuming most of you heard the keynote this morning, we continue to invest heavily, heavily, heavily in the core foundation of Snowflake. If any of you has a technical inclination, there's another keynote called Platform Enhancements, and that goes deep in the weeds.
If today felt weeds, no, that's not the real weeds. There's maybe another 20 different enhancements, all of them very differentiating. I talked about cost governance is not lost on us, yeah, the occasional customer conversation is like, "Snowflake is expensive." The message that we're trying to convey this morning is like, of course, if you leave the water running at your home for hours and hours and then, hey, water is expensive, right? We wanna go and help people and customers understand and get alerts and visibility into that type of expense. Governance, we can talk a lot. Replication, I just alluded, very important for us, not only from business continuity, but replication is also a key foundation for how we enable data providers and soon application providers to deliver cross-cloud solutions.
Data access, which is where we can go talk about the specific. We announced the partnership with Dell and Pure. What was not clear in those announcements, or we didn't make it clear as explicit as we've made it this morning, is we are providing access to any S3- compliant store. I can tell you I have three, I think there's a fourth partner ready to implement additional compatibility integration with Snowflake, which just says the reach of the data cloud is only expanding, and that is just compute for Snowflake and governance for our customers. We announce Iceberg Tables this morning. The hit ratio was way lower than expected, so here maybe it's even lower. Apache Iceberg? One, two, three. Okay.
Without going too deep in the weeds, this format maybe is easier to say it. There were three open table formats that have been in the market. Delta tables from Databricks, Apache Hudi, and Apache Iceberg. We evaluated all three to figure out how do we deliver open file formats and open table formats to our customers. I'm happy to walk you through how, with an objective evaluation, Iceberg is far ahead of the alternatives. Iceberg also has the backing of Apple and Netflix, and Amazon is leaning on it. I think you're gonna see it emerge as a de facto standard. Obviously, our announcement today is important for our customers, but for the industry is a very massive, massive endorsement of an alternative. With this, what we're giving customers, our customers, is choice.
If you like Snowflake's file and table format, that it still performs better, compresses better, so the economics are better. If you want the alternative, we let you have the alternative. In my mind or in our mind, this is the end of this dialogue on like, "Oh, but I'm locked in on Snowflake." By the way, the truth of the matter is lock-in is not about data. Lock-in is about applications. The data moves fast between open and close and medium source, whatever you want. It's always the application that is not easy. Pillar number two, how do we bring the development to the data? We announced streaming pipelines. We do not think that there needs to be a distinction between end-of-day reporting and last-minute reporting. Like, no, all reporting should be fast. Massive announcements this morning.
This get— You all may also have questions on streaming platforms and how we think about it. Yeah, we're getting there. We're going deep in that type of use case. Python, if I were to say there's one large actionable today announcement for our customers was Python. Our private previews, the philosophy of those is it should be a small preview, the word private. So we get 10, 20 customers to evaluate, get feedback, and improve. I think we have 100+ , close to 200 customers in the private preview because the demand was over the top, and we have hundreds of customers in the queue that we just couldn't enable more in private. So I'm dying to see all of my Python usage charts tomorrow morning. The latent demand that we have is through the roof.
In machine learning, we've been hearing it for a long time. Oh, Snowflake should be doing more. How do we enable more? We don't just want to do what everyone has done here. Like, "Hey, here's a tool that abstracts some framework." No, our thesis is very clear. We want to enable machine learning to run close to the data, again, because of governance. It took us three or two and a half years to secure the Python runtime. Now that we have it, whoever wants to follow has a steep journey in front of them. Frank alluded to Streamlit. I think Adrien did a great job this morning. I think now all of you may have a sense in your head of how is it that it fits our thesis?
Because what we're doing is we're simplifying the development of experiences and applications, and we didn't have, frankly, the front end. How do you see it? How do you do it? And some of the experiences that I'm seeing with Streamlit are overly cool. When I said at the end, I want people to think of how are you gonna wow your business users, this is a big opportunity, and we're integrating as a first-class concept at Snowflake. Pillar number three is around productizing development. This is where all of the collaboration in the marketplace fits into place. And the most meaningful announcement this morning is this topic of a native application.
Of all the competitive chatter that I get to hear all the time, the one that is probably funniest of all is when I hear, "Hey, everyone caught up on data sharing." I think everyone missed what was our true vision, which is what we laid out today, and we've been working on it for a couple of years. When someone tells me, "Hey, they caught up in data sharing because there's an easier way to move files around," the mental analogy is something like I'm building cars, and someone tells me that they are 75% of the way competing with me because they have a tricycle. Literally, that's how far off file-sharing protocols are relative to what we're enabling.
We showed at the keynote an entire app with a user experience, cool interactivity, Python code, all of that's the unit of sharing. You can say, "Well, we could have called it a share." No. At the end of the day, I don't know what's an app. That is what we call a native app, and that's what we think is gonna be a whole new generation of software being built on Snowflake that runs on all three clouds. Customers can monetize those applications out- of- band. Fair game. We were fine. If they monetize through us, we take a fee for that. We announced the Native Apps Accelerator program, and it's going quite well. This is private preview for the providers.
The other admission I would say is we have more interest from people wanting to leverage this vision than we can support use cases. Right now, I think we can support two out of every 10 use cases that come our way. That means we have work on the roadmap, but I'm not gonna share it today, not at the conference. Next year, we know how we enable seven out of those eight missing use cases. Cybersecurity, I think Frank said it very well. Cybersecurity is a data problem. Almost everything is a data problem.
The interesting insight for us was every company that is taking an aspect of cybersecurity, because cybersecurity is not like, "Oh, I did cyber." No, there's threat intelligence or how do you protect your laptops or how do you do email security. Each one of those apps is copying today subsets of the same data. What we're doing is turning it on its head. How about a common data substrate and all these apps built on Snowflake? We've seen great momentum on Snowflake being leveraged for cybersecurity at large companies that you see here on this slide. In many ways, us formalizing cybersecurity as a workload is because we saw customers were starting to do it and like, "Why don't you adopt this, talk about this?" Here we are. The last large announcement from the day, this is a multi-year investment.
It's Unistore. This is pretty much the holy grail of the database world, which is how do you create a transactional store with very, very fast performance, and that data is also seamlessly, readily available for analytics? We need to be careful because we partner very well with a number of ETL partners, but any of these apps does not need ETL because the data is seamlessly available for analytics. The preview started end of March. We have the companies that you see on this slide. It's early on. I would say the preview for Hybrid Tables Unistore will still run long through the end of the year, maybe into the beginning of next. We're already seeing validation of the performance, the latency, the ease of use of the use cases, and we couldn't be more excited about it.
I end where I started. This is our platform. At Investor Day last year, I showed this visualization where every dot here is an actual Snowflake customer. Like, this is true. This is not a diagram. This is a data visualization. Every link is a stable sharing relationship, and this continues to grow very healthily, and you'll hear more on how we see the impact in our business and the loyalty that customers that are doing data sharing represents for us. I'll end today where I started the keynote, which is trying to frame the significance of what we're saying today is not trivial. This is not a way to try to be revisionist and say, "Well, every four years something cool happens, so this time probably had to be cool."
No. We know that it is the combination of many years of investment on our end that finally got to a stage where we can talk about it, share technology with customers, and all of us, Frank, myself, Greg, Benoit, Thierry, are incredibly excited to see what people do with the platform, 'cause I don't think we can predict the types of things people are gonna do with Snowflake. With that, I think we're gonna do a stage setup, and then Chris is gonna come and talk to Tom. Thank you.
Hey, everyone. My name's Chris Degnan. I'm the Chief Revenue Officer of Snowflake. I was the first sales rep at Snowflake, so I've been at Snowflake for almost nine years now. I'm super excited to have Tom from Western Union up here today, and we're just gonna talk to you a little bit about what they're doing with Snowflake. Tom, how are you?
Good. How are you?
Good. Thanks. Maybe you could just give us a brief, you know, introduction to Western Union and your role at the company.
Sure. I'm the Chief Data and Innovation Officer. I run everything from our data platforms to our data engineers to our martech and adtech space, our entire banking platform globally, as well as a lot of our payment network as well. Western Union itself, right, is a business of moving money between 200 countries in the world. Right? What's moving money? Actually moving data around the world. Snowflake helps us enable that across all of our markets globally.
Great. Thank you. In your role, how do you assess, you know, technology decisions? What are the decision criteria? How do you actually make your decisions?
I've had the pleasure of working at JPMorgan Chase, HSBC, and now here. At any financial institution, it really is one of three levers. You're either gonna drive cost savings, you're gonna drive revenue, or you're going to reduce your RWA or fraud risk. That's it. That's how financial companies have been run for hundreds of years around the world, right? For us, we started with the first lever, right? We were able to take out costs and optimize our data footprint by leveraging Snowflake to do that holistically.
Awesome. Why did you look at Snowflake to begin with, and when you did, who else did you look at?
Sure. We went through a pretty large analysis. I've had the opportunity of doing this in two other companies, as I mentioned, right? Through an evolution of my career. We looked at it. We said, "Okay, what's out there today? We wanna be cloud. We wanna be cloud native. How do we think about enabling that?" For us, we went and looked at both BigQuery, Redshift, CDP from Cloudera, as well as Snowflake. We went and did the analysis. From our standpoint, it really was a no-brainer. We ended up choosing Snowflake. As part of that, we were able to really not just accelerate, but really start driving value and overall cost savings.
One thing that we were able to do, which I think is really quite amazing, is we migrated 34 data warehouses in 18 months across 20 PB of data around the world. I don't know of any other platform that can do that fast.
That's awesome. I mean, these are great stories to hear. Since actually migrating 34 data warehouses and 20 PB of data, have you seen any additional benefits?
Quite a few actually. As I mentioned, for the start of this, we really chose that first lever, cost savings, to get the initiative started, get our buy-in through our executive team and the overall company. Through this journey and through the evolution of Snowflake, the great products that your team has been releasing into the partnership, we also have driven a lot of other benefits as well. We're happy to announce that our fraud is the lowest fraud in the company ever, in the history of the company, driven by the models, the analytics we built to build on Snowflake, right, and then plug into our applications. We now have over 150 applications globally running off of Snowflake, running our business.
We now have enabled Snowflake to actually be our settlement system with Visa for our new bank we've launched in Europe. We now settle with Visa. The transactions across their entire network with our network every single day on Snowflake. Quite amazing, actually.
Awesome. After you selected us as your platform, what was the first migration? What was your first targeted migration? How long did it take?
When we started, we said, "Okay, we have a big journey ahead of us, right? Let's get things moving in the right direction." It takes about three months to really get the first workloads up and running and to do the parallel testing, and then turn off the old. That was the start of a larger snowball that we basically drove into Snowflake. I mentioned before, going from one workload in three months to 34 warehouses in 18 months, we had a lot of parallel work streams going. Once you learn and train the company on how the right processing, the right op model, the right framework to do that migration, it really is a pretty repeatable process going forward.
In that vein, what were some of the actual business use cases that you migrated initially?
Sure. We started really with, I'll call it our BAU analytics to start, insights, reporting, dashboards. We said, "Okay, we have applications that run in our warehouses. Let's move those over." We went and moved the overall applications over as well, plugged them in, made sure we had no business downtime, provided a better service and a better experience for our customers. Then we started looking at, okay, well, what's our mission-critical applications? For us, it's pricing. We need to do pricing in 20,000 corridors across 200 countries, right? Enable that analytics to drive the right pricing at the right time across all those markets. Really having that engine in the back to calculate and to revamp what that pricing should be across those countries based upon market conditions was key for us.
That really is a huge value add that we basically bring to Snowflake, and then largely enhance as well going forward.
Great. So when you did the migrations, what was a major benefit that you saw of the migration?
When we think about migrating, you know, I mentioned before we had that first lever, right, around cost savings, but we actually had two principles. Right? One is we had to make sure that we put this on was resilient, drove cost efficiencies, operational efficiencies with our staff and our teams as well. On top of that, we need to make sure that all the applications, the thousands, tens of thousands of jobs, the hundreds of thousands, right, of impacts we have around the world every day, that they had to perform at the same level or better. What we saw is actually over 90% of all the jobs, the workloads, the applications, performed 2x-4x faster, right? Moving to Snowflake.
It wasn't just a cost efficiency, it wasn't just an operational efficiency with the aligning all the people in the company to one warehouse and one platform. On top of that, we actually largely improved our business, our customer experience as well by leveraging Snowflake and that overall power we have to process faster.
I'm sure in the sales process, we talked to you about data sharing. Was that in your initial plans for using Snowflake?
It wasn't in the initial plans, you know, but what we saw was a lot of added value over time. As part of our business model, we operate around the world, and we have partners around the world. We call them agents. There are 600,000 retail location agents around the world that allows a customer to come in and to provide cash and send money around the world or to receive cash, send money around the world. What we have done is leveraging the data share, the marketplace, we're able to actually now connect our agents and our partners with us. Now we can share the information, settle faster, cleaner, better security, right? It's all within the same encryption key. On top of that, it really improved our operations.
Now when we share that information, we can share it almost in real time with our partners, and they've now built their processes and their operations off of that same landing data store within Snowflake to then run their business. In the past, we would have to actually go and physically take the data out of the warehouses, right? Go and bundle it and send it as a SFTP transfer to our agents and to our partners. Obviously, that has, you know, concerns about data privacy, concerns around the movement of data outside of our ecosystem. Then on top of that, the partners then have to go take that information, somehow upload it into their systems, and then run additional processes over that. That takes a long time. This way, it's pretty seamless across both our side of the platform and their side as well around the world.
Sounds like we saved some time there.
For sure.
Inevitably, as we've been successful across Western Union, you're getting probably more requests to use us. How do you justify more expenses on Snowflake?
We don't look at it as expenses. We look at it as opportunities to drive customer acquisition, opportunities to drive additional revenue and new products. One thing that we're very proud about is in Q1 of this year, we announced launching two banks in Europe, Germany and Romania. We built that in less than 12 months, completely cloud native. One of the key fundamental capabilities of launching it that fast was partnering with Snowflake, and really being able to drive not just data analytics, but actually business processes and applications on Snowflake that allowed us to really build not one bank, but two banks, Germany and Romania, in less than 12 months and launch it. Now we're actually in friends and family for both Italy and Poland as well, just three months later.
Really, the opportunity is not just around data and analytics, it's around new products, new services, and speed- to- market.
Okay, great. One of the things that people do come at us is like, "Hey, we're costing more." A lot of times, our professional services team will come to you and will say, "Hey, we can help optimize." Have you used our professional services organization, and if so, how'd it go?
We did in Q4. I actually was quite impressed. My rep from Snowflake came and said, "Hey, we can save you some money. Let us help you optimize it." I'm like, "Well, wait a minute here. Like, do you want more money?" It was quite an interesting discussion to have. I said, "Okay, you know, I'm open to it. Let's figure out how we can optimize and make things better, and in turn, then I can take those funds and then help to drive things faster and better and be able to then drive new product development from there." In Q4, we did the assessment of our platform. In Q1, we then implemented those changes to optimize.
We saw about a 15%-20% improvement, right, in our overall costs, but that was in March. Now we're in June. Three months later, interestingly enough, we actually are about 20% over our pre-levels of optimizing, not because we did things poorly, but rather we now have leveraged those funds and that money, and we leveraged them to actually go and build other products that drives more revenue for us and customers going forward. It's actually a great opportunity for us to not just take the investment in Snowflake, but actually find ways to take that partnership and really then drive additional products and services going forward in that partnership.
Was there pressure from the finance team to put it back to the bottom line?
We did. That was part of the deal. We said, "Okay, you know, we'll make sure it hit the bottom line." But as part of that, you know, we then had the ability and the drive from our top of our house, our CEO, where we needed to build new products. We needed to go to market. I'm like, "Well, great. We have this opportunity. We just found efficiencies. Let's leverage that money, and let's go accelerate." He's like, "I agree. Let's go." Even though we did optimize, and we did, you know, have those three months of savings, as mentioned earlier, we are now actually above those levels, but because we built additional workloads and processes and applications now on Snowflake.
Awesome. Well, Tom, you've been a great partner to us. You know, what do you see the next phase of growth for both Western Union and Snowflake?
Sure. We are on a journey with you, where I mentioned before we just launched two banks in Germany and Romania with two other ones in friends and family. We are continuing to launch additional banks in Europe and then looking to expand that beyond Europe as well. We see a huge opportunity there, not just around the financial sector, but also then adding incremental consumer ecosystem-based products, right? To then provide those services to our customers. We actually have a very interesting customer base where there are migrants, right? Where they have lived in two, three, four countries in their lives, and they actually want to participate in their previous countries, right? Where they send money back home, supporting family or a loved one. Really being able to offer those services is quite key for us.
As part of this, and with some of the announcements mentioned earlier, how we can accelerate some of the new product build, the applications, but also being able to go quickly, right, and be able to launch quickly. Once again, I don't know of any other company in the world that's launched not just two banks in 12 months, but then two more three months later. Really quite amazing the partnership that we've built here together and the ability to not just accelerate with our analytics and our insights, but also the ability to drive forward new business processes, new applications. Those applications aren't just for internal purposes, they're for new products and services to our customers, and driving revenue and improved customer experiences to them.
You certainly are pretty tied into our roadmap, and you heard Christian talk about some new workloads. You know, one of the workloads that you've heard us talk about is operational workloads that we're pursuing. Are there workloads that you'd consider moving to us in the future?
I've been one of the people asking for it. The short answer is yes. We actually do a lot of real-time workloads within our company. I'll give you a simple example. You mentioned pricing a minute ago. Today, we do not have our pricing call on Snowflake. We actually have that today on Cassandra, where we do over 1,500 pricing decisions every minute, every day around the world. Right? We have the models and the analytics on Snowflake, but not the actual real-time processing. What's really interesting for us is we see, and mentioned earlier, is having to move data from Snowflake to another partner, why? Right?
If we can just put it all in one place and allow us to do that, it's just a lot easier for us to manage our business and then to make changes quickly depending on the market's conditions and what's happening there in every country in the world. For us, the short answer is yes. I'll continue to push, you know, ask for help from the team. From our standpoint, it actually drives a lot of synergies and operational efficiencies about how we can manage our business better by having it together.
Thanks. The finance team listening, hey, please don't give us more quota for what I'll have said. Again, Tom, thank you for doing this. This is awesome, and we look forward to growing the partnership.
Thanks so much.
Thank you.
Thank you, Tom. Thank you, Chris. Now, joining us virtually, our CFO, Mike Scarpelli, will talk about executing against the opportunity.
First of all, Tom, thank you very much. I don't know if you're still there, but that was great because a lot of what you said we see in our business, and that's what's driving so much of our business. I'm gonna talk a little bit more with some other customer examples what Tom was just talking about, especially around optimizations. I really wanna give a highlight for 2022. 2022 was a great year. We passed a number of milestones, $1 billion in product revenue, growing more than 100% year-over-year. We onboarded approximately 1,500 net new Snowflake employees. For the first time as a public company, we reported non-GAAP adjusted free cash flow.
From everything on our side, it was a great year. Next slide, Jimmy. I wanna talk a little bit about the growing markets. You know, we are really— the database management spend systems are growing every year. Today's focus is really the road ahead for Snowflake, and we're gonna detail how Snowflake is executing against the current market opportunity. Frank talked about how customers are demanding more from their data. Data's increasingly important in driving business decisions.
I think you heard that a little bit with Tom as well too, and serves as a key competitive differentiator for our customers. You can see this dynamic reflected in the numbers. According to Gartner, database management spend will outpace the broader software market, growing from 13% of software spend in calendar 2021 to 17% in calendar 2026. Next slide, Jimmy, please. It's not only our market growing, but we're also benefiting from a shift in how organizations harness their data. Workloads are moving to the cloud. You can see here that cloud, as a percent of the database market, is growing, expected to grow to 71% in calendar year 2025, and OLAP is gonna grow to 56%.
The workloads are moving to the cloud, and more work is happening in OLAP databases, and that's what people are predicting. This market is moving towards Snowflake's core competencies. Next slide, Jimmy. There's a significant market opportunity in front of us, and we're not standing still either. You've heard from Christian that workload enablement is key to unlocking market opportunity. With the addition of Unistore and cybersecurity workloads, today, we can address more within our customers' data infrastructure. We believe our cloud data platform TAM totals 28— or $248 billion, today, and is comprised of these eight workloads that you see on the slide here. Next slide, Jimmy. The Data Cloud represents an even larger market opportunity. You heard Frank say we don't know how big this is, but we know it is a massive market opportunity.
The Data Cloud encompasses all workload execution. It also includes our opportunity within Marketplace and data sharing. We are paving a new path with this market that grows beyond existing data management spend. We are not able to size the market, as I said, but we know it is key to unlocking new customer wins. That is a key differentiator for us when we're talking to potential new customers. Next slide, Jimmy. What are the growth drivers? Let's take a closer look at how we're tackling this market with some of these specifics. Our primary focus is on quality customers. You've heard me that in the past on earnings calls. It's not all about the absolute number of customers, it's the number of quality customers.
We really define a customer— How we define a quality customer is really a customer who has the potential to generate greater than 12— or $1 million of trailing 12-month product revenue. Of our current $1 million customers, only 45% of those are Global 2000. 55% are enterprise, and 1% are our corporate accounts. Outside of the Global 2000 customers, over $1 million product revenue are often early Data Cloud adopters. 64% have at least one stable edge or are a Powered by Snowflake customer. Next slide, Jimmy. We are targeting the largest organizations in the world. You can see here how the increase in our Global 2000 customers is going.
You know, we did this shift from the Fortune 500 last year, and we were also giving you the Global 2000 to just the Global 2000, because really, the Global 2000 is more global. It is only public companies. It does exclude a lot of the large private companies. This is really just a proxy. We focus on the largest enterprise in the world, whether you wanna call them Global 2000, Fortune 500. We're tracking this. You can see that we have 560 of these Global 2000 as of Q1. We landed quite a few last quarter, and we're gonna continue to focus on these large organizations in the world. Next, page.
You know, one of the biggest questions we get from people, and I thought we'd show you this data because we've never showed it to you before, is who are you guys replacing? Many people just think of us as a Teradata replacement. That is far from the truth. This is really looking at the first deal we've done with customers. Of the migrations that we've done, 44% have been cloud migrations, 56% have been on-prem migrations. Now, when you look at those on-prem migrations, 12% have really started with a data warehouse, 73% are a mix of OLTP or OLAP, think of SQL Server and other things, and then 15% is big data. You can see our growth is really driven by these lands that we're doing. It's not just about a Teradata replacement.
Yes, those big data warehouses tend to be some of the bigger deals we do, but we do many, many, many migrations of on-prem software. Next. Customers land with small initial contracts. That has really not changed that much. The median contract value for customers won in fiscal year 2020, they're now spending more than $1 million in product revenue, was $180,000. The first contract is a way to buy capacity to get up and running on Snowflake. That's the way most customers look at it. They haven't made that big commitment to Snowflake yet. It's not representative of our expectation or the customer's plan for their annual consumption. Next page. Once they're up and running, customers expand, and you see this in our net revenue retention.
It was 177% in Q4 of 2022. This increase reflects the exceptional growth in some of our largest accounts, and a lot of what Tom was saying supports this net revenue retention. It is pretty typical of our customers that they grow similar to this. We expect our net revenue retention will decline over time, but will remain world-class for the foreseeable future. It is really hard to forecast long-term what net revenue retention will be, and as I had said before, we do expect it to stay above 160% this year. Let's look at some of the fundamental drivers that sustain this metric. This is an example of a telecom customer. Their net revenue retention benefits from customers consolidating on Snowflake.
Customers typically land with a single workload, but plan to migrate a much larger data estate. The sample customer first migrated their legacy data warehouse to us, then their consumption expanded as they moved work from additional legacy solutions to Snowflake. This is pretty typical of what we see with our customers. This customer, by the way, also went through a couple optimizations with us. That's why you see those dips similar to what we talked about with Tom. Next. The other way is expansion through new use cases. We have multiple vectors for expansion beyond consolidating existing workloads. Customers also expand with new use cases. With the addition of data engineering workload, one of our healthcare customers increased consumption five-fold. Snowflake replaced an open source solution to provide faster data processing at a lower cost.
Next, Jimmy. We see this pattern play out again and again. One of our financial services customers grew with the addition of a data science workload. This organization now relies on Snowflake to process 1,500 model features for its loan risk assessments. Next. Today, our customers over $1 million in product revenue spend on average $3.7 million on a trailing 12-month basis. Next. We are in the early innings of our expansion opportunity, and I want to stress early. For our most recent quarter, our largest customers, the Global 2000, consumed on average $1 million on a trailing 12-month basis. As you can see here, our average million-dollar-plus customers, $3.7 million, yet our Global 2000 is only a million dollars. Why?
Because we're still very much in the early innings with these, and I will say the larger the organization, sometimes the slower you move. These customers are all in the very early innings of their journey with Snowflake. Next. Okay, I know you guys are all interested about the profitability drivers for our business, and I wanna stress we're committed to addressing these growth drivers while demonstrating improving operating efficiencies. You know, I wanna stress since day one when I joined the company, and Frank did as well, it's not growth at all cost. Yes, growth is the number one thing, but it's being responsible growth, and only spending where you think you can get a true payback for that. So I wanna walk you through some of the dynamics propelling this operating leverage.
You've heard us talk about before, the price- per- credit. Well, the price- per- credit is really what drives our product gross margin. We expanded our non-GAAP product margin from 63% in fiscal year 2020 to 74% in fiscal year 2022. The price- per- credit really fuels this margin expansion. As a reminder, customers sign a contract to purchase a set number of credits at a specific price. They're contractually obligated to spend that money. As customers consume higher priced credits, we recognize more gross profit. Next, Jimmy. You heard us talk about before the consumption mix driving this. You can see here our price- per- credit has increased with the adoption of higher-priced product additions and discounting discipline. Enterprise and business-critical additions carry the highest margin profile.
As we have started to focus on the largest customers in the world, these guys are the ones that want business-critical and enterprise. That has a much higher profit margin for us. I think you're gonna continue to see that upward trend in the higher additions for Snowflake. Next. Moving to operating income. As we mature as an organization, large customer relationships provide an important source of leverage. Increased scale lends additional yield to R&D and G&A investments, and our mature accounts are less sales intensive. In the most recent quarter, we added 22 customers over $1 million product revenue, and 10 customers over $5 million in product revenue. We believe we are just getting started in scaling these accounts and reaching our full margin potential. Next slide, Jimmy.
Finally, adjusted free cash flow is supported by operating leverage and strong collections. Current adjusted free cash flow is benefiting from significant bookings growth. Over time, the disparity between operating margin and adjusted free cash flow margin will normalize, but you can see the large deals how we've booked them by quarters, and you're always gonna see Q4 is gonna be our biggest quarter with big deals. Next. Going forward, I'd now like to shift and talk about the investments that underpin our profitable growth. Next slide, Jimmy. You've heard us talking about our vertical sales force. We're tackling this market with this vertical approach. We think this is really important.
In North America, our sales force is organized around seven vertical opportunities, financial services being our biggest, or healthcare and life sciences, retail and CPG, advertising, media and entertainment, technology, public sector, and education. This industry focus supports mission alignment with customers and is critical to unlocking our full market opportunity. It's really important with this alignment that our reps understand not just the technology, but how our technology can drive value in those specific industries based upon those industry needs. Next. You know, we're investing in our sales headcount. You can see here the growth that we've done from 2020 to 2022. You can see we're still investing very heavily in North America, but EMEA and APJ, we're seeing really big investments relative to the size, but obviously they're small.
We still see a lot of opportunity within our biggest markets, where we're the most concentrated, where we can add more reps, and we see the yield on these salespeople that we're adding. Next. We'll continue to invest in R&D, which furthers our aim of workload enablement and directly impacts our expansion opportunity. Next. We're expanding geographically. We're advancing our geographic footprint with new regional deployments. We're now at the end of Q1, we have 32 deployments, up from 23 in fiscal 2021, and we will continue to add more deployments around the world based upon the needs of what our customers want.
A lot of these deployments that we end up opening, a lot have to do with data sovereignty issues with some of the countries we're going into, or the types of customers we're dealing with in certain countries, where the data must stay within those countries. We're getting really good at opening these new deployments and bringing our costs down to open these new deployments. We will continue to do those. Next. Finally, you hear a lot about Snowflake Ventures, but what we're really doing is we're really cultivating an ecosystem around the Data Cloud. We do a lot of investments in these companies, and as Frank talked about at the all-hands or at the keynote today, it's not purely a financial return we're looking when we invest in these companies.
We're really looking at building a tighter alignment to make the integrations great with Snowflake so that our customers have a great experience. Snowflake Ventures is really helping accelerate that for our product teams. Next. What are our priorities? Our guiding principles have not changed and will continue to be. It's really driving growth, showing improved operating leverage year-over-year, and maintaining dilution of approximately 2% for the company. Those have been our priorities since the day I joined, and will continue to be our priorities as long as I'm here. Next. Long-term, what does this all mean for our long-term model? Let's dive into some of the dynamics that are unique to our business. Next, Jimmy.
First, there's a lag between when we land a customer and when a customer, when we recognize consumption. On average, it takes customers eight months before they're consuming Snowflake at the run rate implied by their first contract. Remember, most of these customers start small, so we don't get significant revenue out of these customers in year one. It's usually in year two, year three, and beyond, and you see that in the net revenue retention for us. Next slide. Performance improvements. At the end of Q4, we talked a lot about this, and I just wanna reiterate this as well too, and as Christian outlined. Product innovations increase our competitive differentiation and expand our market opportunity. We launched Warehouse Scheduling Service V2 earlier this year, which improves performance for low- latency, high- concurrency workloads.
We're able to address some transactional workloads with Unistore now, it's really because of the foundation that Warehouse Scheduling Service built. Performance improvements will always be part of our product strategy, and therefore will always be part of our financial model. They always have been. We've always forecasted a 5% revenue headwind every year from these performance improvements and hardware improvements. We think most of the improvements going forward are gonna be software, not hardware. There will be some hardware, but really we think it's software. We estimate that each year, as I said, it's 5%. That hasn't changed, and we think that's gonna continue. I really wanna stress, this is an estimate. Until you've actually developed the software, you really don't know.
Some years it could be higher, some years it could be lower. For long term, we estimate it's 5% a year. Really, those improvements are passed on 100% to our customers, because every year they can do more with the same credit that they had purchased last year. This is good. Why it's good, because as we become cheaper for our customers, they move more to us. You heard Tom from Western Union talk about that. We did optimizations, they saved money, and then he reinvested that into doing more with Snowflake. We see this time and time again with our customers, and so we're gonna continue to do this. Next. I wanna show you a customer optimization here.
Not only are performance improvements part of our product strategy, but they're also part of our financial strategy. In these examples, customer optimization serve as a good proxy to understand the impact of performance improvements in any given account. One of our retail customers engaged our professional service team to optimize their existing Snowflake environment. Optimizations entail an immediate lull in consumption, and you heard that from Tom with their consumption drop. They also introduce efficiency gains and unlock existing budgets, which customers can elect to apply towards new use cases, and we usually see them do that. This retail customer did exactly that, consuming ahead of their prior run rate within two quarters of the optimization work. Their consumption was net neutral on an annual basis and ahead of prior forecasts in subsequent years. Next page, Jimmy.
This telecom customer engaged our professional service team multiple times. After their second optimization, the telecom's consumption rebounded within six quarters. We saw $7 million revenue headwind in the first year, but extracted more value over the long term. Next slide. We're reiterating our fiscal year 2029 revenue target of at least $10 billion in product revenue. To help you track our progress towards this goal, we reissued our customer framework. In 2029, we expect 1,400 customers over $1 million product revenue, spending on average of $5.5 million on an annual basis, totaling 77% of our overall product revenue. Next slide, Jimmy. In 2029, we will still be growing 30% year-over-year. Our margin target reflects the sustained investment necessary to reach $10 billion in under seven years. They are not terminal targets. We're not gonna stop there.
In 2029, we expect 20% operating margin with R&D at 15% of revenue, sales and marketing at 35% of revenue, and G&A at 7% of revenue. We expect at least 25% adjusted free cash flow margin. Next. We're gonna go to Q&A.
We get to see Mike on the screen. Hey, Mike.
Now I can see you. I couldn't see anything earlier.
Hey, guys. I've had the m ic over here. It's Brad Zelnick with Deutsche Bank. Thanks so much for a great presentation. All the disclosure, especially Mike, in your presentation. I got a number of questions to ask, but maybe if I could just start with the Data Cloud opportunity, which since the IPO, you've always had a picture up there which makes it seem like it's much greater than the data platform opportunity that we know today. And I appreciate you can't quantify it, but can you maybe just help us mere mortals to wrap our heads around, you know, how you think about it and how you're monetizing data sharing and data marketplace today, and what it takes to ultimately get us to realize that opportunity?
Yeah. You know, we've been getting this question, Brad, since the beginning of time and, you know, we give you a lot of anecdotal information like, you know, the interview with Tom, because there's just no way to do a, you know, Gartner-group- style analysis around it. Even if there was, you know, I wouldn't put a lot of faith in it. I mean, what we're trying to do is that the runway, you know, that we have in this business is enormous, right? I mean, even if we give, you know, whatever $10 billion worth of guidance, putting it in the context of the overall opportunity, it's still very tiny.
I mean, I'm actually embarrassed about that number, considering, you know, how big the opportunity is and the number in there, right? That's also when I talk about the relationships with the public cloud vendors, we're such a nit, you know, relative to, you know, how big their businesses are. You know, obviously we're trying to change that, you know, as hard and as fast as we can. You know, there is such a thing as snowball effect. Snowball has nothing to do with Snowflake. I know Warren Buffett coined that term, you know, snowball effect, which is, you know, size begets more size, you know, faster and faster, because it just rolls up.
You know, I can only tell you that we won't have to resort to, you know, acquisitions, market expansions, I mean, things that you've seen Salesforce do over and over and over to try to keep their growth alive. You know, our challenge is not keeping our growth alive. Our challenge is just enabling, you know, our capabilities, you know, fast enough, you know, that, you know, we're there when a customer need us. You know, like, you know, you saw, you know, Tom talking about he has another, you know, transactional database engine hanging off of his architecture because, you know, we can't do what he, you know, needs us to do. But we will. You know, so that's our moat, you know. It is.
We are literally, you know, building our platform as hard and as fast as we can. The growth will absolutely happen, you know, as long as we do that. Even, you know, we're a very paranoid, you know, bunch, so we're building really hard and fast. Even on the basis of what we already have and what we already do, you know, we think we can sustain enormous growth because we're way ahead of the marketplace here. I mean, we saw this morning people don't even know what Iceberg is. You guys don't even know what Iceberg is, you know? I mean, it's like there's a lot of things that, you know, people are gonna have to catch up to, you know, to where we are. So, you know, that's not a very satisfying answer to you.
If I would, you know, if I were in your shoes, you know, I would really look at us in the context of the broader market opportunity as defined by the public cloud vendors, as well as, you know, some of the definitions that we're giving you, because that's sort of the force field. We are the second largest spend in more and more, you know, of our customers. So we are sort of a proxy to that world, and that's how a lot of customers think about us as well. They say like, "Hey, we spent X on Amazon, we spent Y on Snowflake," and that relationship is a thing. It's real, you know. A lot more questions. I'll let Jimmy decide, because otherwise I'm like Joe Biden and picking the wrong guy, you know.
Hi. I think I've got the mic. Can folks hear me? Testing. All right. Good.
Yeah, check. That was good, you know.
This question's about the growth in the customer counts. I see the growth in the customer base is great, but the growth in net new customers is kinda flattish, and the like, and yet we see like good growth in the overall, looks like G2K customers. Just help us understand sort of how you're optimizing acquiring new customers versus and where you're seeking to acquire them and where you're pushing for expansion. I just don't quite get it.
Well, I can say something about it. Mike will probably have, you know, more to say about it. You know, we don't really instrument from a sales incentive standpoint, you know, whether you're selling, you know, new customers or existing customers. We actually think that we will differentiate that to make sure that we're more balanced in that regard. You know, salespeople go for the line of least resistance. They're gonna try to reach our quota, you know, on the basis of the opportunity that is, you know, most productive for them to pursue. Oftentimes, you know, those are, you know, account expansions. Growing accounts is an incredibly productive selling motion, you know, for us.
You know, you see the sales organization wanting to go there, you know, versus like, "Hey, I'm gonna do a $100,000, you know, cap one , you know, type of deal," because that doesn't put any meat on the bones. That said, you know, our corporate sales organization, which is sort of the low end of the tier of our selling motions, I mean, they are the cap one generator. You know, they deliver most of the cap one , you know, accounts. You know, they're not the majority of our sales organization, obviously. We don't have specific, you know, incentives or metrics around accounts.
I mean, we look at it, but, you know, we fall, you know, well short of saying like, "Hey, you know, it's all about the number of accounts." Not all accounts are created equal, you know. Some of them will do a $100,000 deal and that's all they'll ever do. You know, that's not very productive for us. You know, we really veer towards the ones that we feel that have the potential, you know, to grow, and, we have to be differentiating on that basis. Anything you wanna add to that, Mike?
I'll just add, as I said earlier, we're really focused on quality customers versus quantity of customers. We want those ones that have massive amounts of data that we know can grow to be those big customers. Are you having a hard time hearing me, Frank?
[crosstalk] Okay, good. Okay. Yes. Hopefully that answered your question.
Yeah. I said more or less the same thing. Mike, you just put another point on it. So...
Hey, guys. Alex Zukin here, Wolfe Research, right up front. Hey, Frank. Hey, Mike. Maybe I'll ask the recession question 'cause I know it's on everyone's mind. Then I'll ask a product question as well, just to get myself out of the doghouse. With respect to the recession question, we're actually out there surveying your customers right now, and half, you know, over half of them are saying they're not planning to decrease any spend. Most of them are not seeing an impact to their forward spend, and most of them are actually planning to increase spend. I wanna ask just why, right?
Like what— When you are talking to your cohort of customers at the executive level, they're all reading the same headlines, yet they're not choosing, at least according to our data, to take down their planned spend. Mike, you talked about optimizations and the model you showed, you know, has the actual spend going above your original spend post-optimization. How much of that is you know, have you reflected any assumptions for macroeconomic weakness in those models, and how do we expect to feel it? The product question is around Unistore, and is it more of, you know, the idea that Snowflake becomes a transactional database, or is it more trying to maybe optimize the need for other tools like ETL that you mentioned?
You wanna start with the product question, Bill?
Sure. On the product side, it's Snowflake becoming a transactional database. OLTP is the more buzz-worthy description. If you ask what's unique and why we have a chance of succeeding in this business, is because there is no need to synchronize the data between the transactional store in Snowflake and the analytics store in Snowflake. It's the combination of both. Yeah, we're very much going on the transactional side. If you ask me, are we gonna go into the low microsecond use case that Oracle is great at? It's gonna take a very long time, if ever, but that's a very tiny slice of the market. The big opportunity is the 10 ms, and that's what we're going after.
Just in, you know, in terms of your macro question. I sort of retired that word from my vocabulary because I'm sick of hearing it. I feel like, you know, we're at risk of media inducing, you know, a recession because people are gonna hit the brakes just because they have high anxiety, not because they have any real data points. I actually have seen that, you know, people are just doing stuff because they're nervous, not because they have, you know, real data that's actually happening to them. I sort of see a similar thing that we saw during the early days of the pandemic. You know, during the pandemic, you know, the tide was running out on hospitality and aviation and the industries that were just getting crushed.
Yeah, they're gonna not spend as much, you know, clearly, right? But at the same time, you know, because of the dislocation that was happening during the pandemic, there was also an enormous need for what we do. I now see the same thing again, you know, in regards to inflation, for example. You know, spiking inflation is driving an enormous need, you know, for Snowflake because you can only parse reality and, you know, figure out what you have to do, figure out what is going on, you know, through data rather than, you know, sort of steering the ship by its wake, which is what people have traditionally done. Again, I see, you know, pluses and minuses, you know, in that regard.
There's also, you know, in the world of crypto, and we're into this crypto winter. There's a term that I've heard, you know, on CNBC. Do you think crypto companies—I mean, they're cutting their limbs off right now, you know, to save money. Do you think that affects us? Of course, it affects us, you know. That's not the whole world. Is that macro? Yeah. Well, I think it's more micro than macro, I guess. In other words, it's not a categorical answer to your question. You know, on the whole, dislocation tends to favor, you know, people like Snowflake because, you know, we really help, you know, discipline, you know, a way forward on that. I can't tell you how supply chain problems that people are.
We are finally getting to a point where supply chain is going to become, you know, fully instrumented. It has to, because the way we've been doing it historically is just, is near insanity, right? Now with these dislocations that are happening, I think it's going to accelerate because it just simply has to. Pluses and minuses. You know, what I've said over and over, and I've said it publicly as well, because CNBC tries to get me to say, you know, that everything that's not perfect must be macro type of a thing, is we just don't get that hysterical reaction, you know, from our day-to-day interactions with customers. You know, they're normal. They're stable. They're doing multi-year deals. I mean, you heard Tom.
I mean, I realize it's anecdotal, you know, it's not a sample of 1,000 customers. That's the norm, you know, not what you're hearing in the media, you know. I hope that's helpful. Yeah.
Thank you. Mark Murphy with JP Morgan. Just wondering if you think that you might be through the kind of holiday related consumption dip that you had seen, you know, maybe five-six months ago, and sort of out of the woods on that based on the way you described the month of May. Or do you think you're kind of more in a kind of a wait- and- see mode on that heading into Juneteenth and, you know, 4th of July here?
Well, I'll let Mike talk about that. You know. Did you hear that, Mike?
Yeah. What I would say is, we forecast all these, if I'm understanding your question correctly, we do forecast based upon what we see historically, what customers do during holidays, and I'm not concerned about any of that. I think our forecast is appropriate for the quarter, and it will be similar to maybe a little bit more we have forecast in the prior year around those holidays.
Hey, Frank, Christian and Mike, Kash Rangan of Goldman Sachs. Congratulations on a terrific Analyst Day. One question for you, Frank. The categories that you're playing into are increasingly becoming larger and larger. It's a $248 billion TAM. Who are you envisioning selling to? Initially, it was replacing a data warehouse, but, like, you're getting into OLTP, you're getting into machine learning, you're getting into running crypto workloads. Who's the buyers? When you become a $10 billion revenue company, closest parallels are SAP, Oracle became ERP companies, and they sold to a certain head of operations. Who do you see yourself selling to that'll be the key sponsor for Snowflake as you become a larger company?
Yeah, you know, it's an interesting question because, you know, historical categorizations, classifications, they mean, you know, very little. It's becoming. It's blurring into, you know, one mega market. That's why I think getting to the Western Union way of describing, you know, you have technology leaders that are really, they're driving the business. I mean, they're building products, they're generating revenue, they're saving money. You know, data is becoming sort of the core currency, you know, of enterprises and institutions. So in other words, separating the world into IT and business, I mean, IT and business are now becoming more or less the same thing. They're still, you know, the classic IT organization and they're very risk relationship and cost-oriented.
You have IT like, you know, Tom at Western Union, that are incredibly business-oriented, even though his title is IT-oriented. You can tell by the way he talked that he is, you know, a business-oriented, you know, technologist. That is going to become more and more so, you know, what's happening. You've heard the term digital transformation for years, and you're sick of hearing it, right? That is really what that is, right? We are really going through a process, you know, where we are becoming digitally transformed, and digital is really leading enterprises. You know, I think, you know, we're biased, and it might be a self-serving comment, but, you know, so many problems are becoming data problems, right? And having data solutions, you know? That favors us enormously.
That's why I sort of took pains to describe, you know, the severity, both in downside and upside, you know, of what happens if you don't become seriously competent, you know, in data strategy and data operations. This is becoming a very central discipline in institutions and enterprises. It's hard to sort of, you know, project this is exactly what it's going to be because we haven't seen it before. We're on a journey, you know? We're now loose, okay? We're no longer sort of sitting back where we were. You can feel it, right? The type of conversation that you're having. We're on our way.
How long that all takes to fully sort of, you know, resettle into a mode that we can all recognize and talk about, I don't know. You know, personally, you know, I don't really care, you know, about the answer because we know what we have to do to enable it. That's what we are. We are enablers of digital transformation in the broadest sense, you know.
If I may add. When you hear us talk about technology, and I emphasized a lot this morning, and you'll hear it throughout the conference, that we're trying to eliminate the trade-off between governance and programmability. If you translate it back to the question on personas, in some ways it's central IT, central chief information officers, and the lines of businesses. As we succeed on this trade-off, I think we're gonna find advocacy across buyers.
It's actually benefiting us a great deal that business is becoming so strong on technology. Because, you know, IT people are not always our friend because they're sort of afraid of their own shadow, and they wanna surrender themselves to a big cloud vendor because that's career-wise, that's safe. Those people don't favor us. As the business is becoming more decisive and more driving and more dominant, that favors us enormously. That's also the reason, you know, why we verticalized the company over the last couple of years. It's exactly for that reason. Our power is gonna come from the business. It absolutely will, okay? Again, Amazon and Microsoft will do well with IT, and so on.
You know, we're aiming for, you know, demonstrable business impact and, you know, that's why it was good to have Tom up here to talk to you because there's nothing like hearing it in those terms, you know?
Thank you. Brent Bracelin over at Piper Sandler. Frank, for you, it is interesting hearing a little more about on-prem. It's a part of the language we haven't heard at Snowflake for a while. How important are some of the extensions via Iceberg or even Hybrid Tables to address some of the on-prem workloads, even if it's read-only, one. Then two, Mike, great to see you. You're up on screen. Very large picture of you there. I'll be careful how I word this one. It's really just around the monetization strategy. I saw 41% of the mix is now business critical. Is that the primary lever to drive higher growth in the say price, pricing SKU by mix shift or are there other factors outside of that mix shift to business critical that could drive higher growth and price- per- credit and/or margins over time? Thanks.
Do you wanna answer that question first, Mike, and then I'll address the first one.
Sure. The movement in our customers to business critical is one of the factors that's driving our product margin to go up. The other, obviously discounting as well. You can control that, but I think we have good discounting right now. The third thing that will drive it in the future, and we're not in any discussions with anyone right now, but we will get into, is getting better pricing out of the cloud vendors. As you know, in almost two years ago, we did Microsoft and AWS. We've yet to renegotiate our GCP. I will tell you, GCP is on average 45%+ more expensive for us to run there than in AWS or Azure, so big difference there.
I do think we can get better pricing out of AWS and Azure, but obviously we need to make a bigger commitment for them, and I expect something in the next year. We will do that, and it will impact more next year's product margin.
On your first question, you know, I had a conversation with Dell almost a year ago, and that conversation didn't go very well in the beginning because they are the ultimate on-premise company, and I guess we're the ultimate cloud company because we don't run on-premise at all. Like, what the hell are we gonna talk about? You know, Michael's talking about edge computing, and he's very exciting about all that stuff. You know, we decided to have our CTOs, product leaders get together, and you know, Christian was in the middle of it constantly.
We had ups and downs, you know, where we just, y ou know, because it was so difficult for them to talk to a pure cloud play like Snowflake, like no good is gonna come of this, you know, kind of thing. In the end, you know, they started to, you know, understand that, you know, a lot of these analytical processes are going to be, you know, dominated from the cloud, and we have to be able to reference, you know, data that will never go to the cloud or cannot go to the cloud, you know, for the time being. Even if it references single attribute, you know, that is on-premise, you know, as part of our overall analytical processing, that could create incredible insight and impact, right?
The fact that they're virtualizing it, you know, as Amazon S3 storage means that, you know, I don't wanna minimize this because Christian will correct me, but it minimizes the amount of work that we have to do because it just looks like another file on Amazon storage, right? That's really nice. It's an incredibly clean way of doing these things. The other thing I like about this is now an enterprise architecture. It's not a cloud architecture, right? So we're bringing the world of on-premise, you know, under the umbrella now, you know, where it can participate in these incredibly mission-critical, you know, processes. You know, we're not very, very far along yet in use cases. This is very much we're in the enablement stage, you know, of doing these things. I'm, you know, I'm optimistic, about it, you know?
Yeah. I'll add, we brainstorm a number of integrations, but it was customer-driven, what we pursued, and it resonates. Many of the organizations here represented have lots of data on on-prem storage systems, and we heard we want Data Cloud for on-prem data.
Yeah. There's whole bunch of businesses, like for example, in power generation, the utilities, I mean, they have power plants. You know, they have data. That is not going anywhere, right? They also wanna be able to, you know. The reason it can't go anywhere is because they're processing the data on the edge, you know, where it is physically resident. That's why it can't move, you know? That will always be that way, but it doesn't mean that we cannot reference that data, right? This I think this is all goodness for the industry, you know.
Hello? Hey, Kamil Mielczarek from William Blair, in the back here. Congrats on the great conference. I just have a quick one on profitability. You're hitting your 15% free cash flow margin targets many years ahead of initial plans. Assuming your sales efficiency continues to positively surprise, how do you think about balancing growth and margin long term? Will you continue to let more cash flow through to the bottom line, or is 25% margin now closer to maybe an upper limit, where going forward you may choose to instead increase investments, accelerate growth, and keep margins flat? Just how do you think about that balance?
I said those were not terminal margins. We just said in 2029, I think we can do at least 25%. Obviously, we will spend money where we think we can spend it efficiently and get yield. I feel very confident in the 25%, and we'll see where we go. Now, I wanna stress too, you are going to see fluctuations within a year because there are certain quarters where, like Q1 is always gonna be a very big cash flow quarter. It's really on an annual basis we're focused on. As I said before, it's not growth at all costs, but growth is the number one thing we do look at, but we do it efficiently, and we'll continue to do it efficiently.
You know, just to build on that. You know, I mean, this may sound like fifth-grade logic, right? But you know, big profit comes from big revenue, and big revenue comes from big growth, okay? Everything comes from growth, okay? In other words, you said you're gonna flow cash, more cash flow through. Why? Just to appease some audience. If I can productively invest it in growth, eventually, you know, it will result in bigger cash flow for obvious reasons, right? You know, having your priorities straight around that is really important. But you know, as you've seen, I mean, we've gone from -200 , -$200 million in cash to you know, as much positive as we are now. We're very disciplined.
We're very optimized around all these processes. Make no mistake about it, strategically growing is what's gonna create the separation, you know, our ability to outstrip the competition and is what's going to annihilate, you know, the competition. Growth is not just a financial issue. It is a strategic issue, right? You can never, ever lose sight of that. It's the single most important thing. I mean, as you know, we've been in high-growth businesses for the last 20 years, and it's always been true. It's still true. It will always be true, you know.
Okay, great. Is this on? Karl Keirstead at UBS. Maybe one for Mike and one for Christian. Mike, I'm guessing you feared you'd get into too much trouble if you were to come to Vegas. Is that it?
No. Well, I was in Vegas, and I had to leave.
I wanted to ask you obviously on the last earnings call raised your operating margin and free cash flow margin guidance. I noticed you didn't raise your fiscal 2029 product revenue guidance. It seems like you're on a pretty good path to get there, so you probably had the room to do that. I'm just curious, why not? My guess is your confidence is higher, but maybe, you know, the world's a little bit murky. There was no point doing that, but I'd love to hear a comment. Christian, I'm intrigued by the idea of Snowflake going after the transactional database market. You just mentioned a minute ago, you're not really gonna take on Oracle, but what portion of the transactional database market are you actually going after? Thank you.
I'll just say, Karl, that we are more confident in our longer-term target where we sit today than we were a year ago, and let's save something for next year's Analyst Day.
Karl, on the what portion of the transactional market, think of modern applications. If you go look at what are the applications that need the super ultra-low latency are all legacy that were built in a very specific way. For the newer platforms, like what we're showing in the keynote, you don't need that, and you can get a lot. Think of it as new applications more than migrating existing ones. Of course, that's a bet that says over time, you go after the bulk of it.
Yeah. The only thing, you know, I think it's important just for context, right? I mean, we're hardcore venturing into the cloud application development marketplace, right? Most of what you've seen in cloud applications are really, you know, refurbished, you know, on-premise applications that are hosted in the cloud. The whole cloud applications marketplace in terms of really native cloud, you know, being able to deploy on multiple clouds that can do all these cloud things, that is still a brand-new, you know, marketplace. That's really what's Snowflake is all about. It's not about, you know, competing with the past. You know, it is competing with the future. The cloud application development is gonna be a multiple of mobile application development, right? These are the orders of magnitude that we're talking about here. This is super important stuff in terms of the future of computing, you know.
Hi. Doug Henschen with Constellation Research. I wanted to ask again about the data marketplace. I know you said it's very important, very tough to size the market opportunity, but maybe we can just talk a little bit about behavior. That bloom visualization doesn't quite capture it. You did have a slide that said customers over 1 million, 63% are data sharing. I just wanna know, is that internally data sharing, externally data sharing? And how do companies then mature towards monetization?
I'll clarify the data point is all based on external data sharing. If we look at intra-company data sharing, then the numbers are higher. Because we track is the ability to connect companies, we look at and report primarily external data sharing. In terms of benefits to customers, we think that it's a function of a life cycle of adoption of Snowflake. If you land Snowflake in your first deal, sometimes you just have your Teradata migration or your modernization transformation in mind, and this is not top- of- mind.
You see a correlation that once you start to break down your silos and start to understand the governance and bring the business logic, that's when you say, "I wanna participate in the data economy by being part of the Data Cloud." There may be some correlation there. Actually we see the cord. There is correlation.
Excellent. Thank you, guys, and thank you for hosting the Analyst Day. Two questions, one for Christian, I think one for Mike. For Christian, you described having both OLAP and OLTP in the same platform as a holy grail for database management systems. Completely agree, but it's also the holy grail in that a lot of people have tried it and nobody's succeeded. We've seen this from SAP, we've seen it from Teradata in the past. How are you guys gonna be able to succeed in sort of being good at both of these where other vendors haven't been? That's question number one. A question for Mike. You mentioned in your presentation when talking about the price performance improvements, you expect it to come more from software.
I thought that was interesting 'cause a lot of investors that I talk to are really worried about Graviton 4, Graviton 5, Graviton 6 continuing to sort of improve performance, and is that gonna be an impact on Snowflake? Why does the price performance come more from software on a go-forward basis versus hardware? Thank you.
I'll take the first one. What was it? A transactional database. Sorry, I missed it. Well, why are we gonna succeed when everyone else has failed? I think a lot of the magic is that it's a cloud-hosted service. When SAP or anyone that has done it before tries to do it's a single software product. We're not talking about how we're doing it behind the scenes, and we have no problem saying there are multiple storage engines, and there are multiple things happening behind the scenes. The cloud lets us all abstract it away. The other thing that I'll say, and Benoit mentioned in the keynote, it was a multi-year investment, and only once we had the confidence we are presenting at the conference, to the customer. We're highly confident of what we announced.
Question on where most of the performance improvements are going to come. I didn't say there aren't going to be any from hardware. We do fully expect there will be performance improvements, but not all software benefits from hardware improvements. Christian's more of an engineer, he can talk about this, but we just feel in the future, and I'm giving you long-term, we feel most of the performance improvements in our product are going to be driven by software versus hardware. Not to say there won't be any from hardware, but software is going to drive most of those in the future, we believe.
100% agree with that. It's exciting for what Mike said. Not every hardware improvement improves every software platform. I've heard from many customers that Graviton2 didn't do much. That is sort of like a game of odds, and when does a specific hardware change influence or is material enough for you? Whereas everything we do on our platform, we do it because we know it's going to have an impact. C all it a selection bias that we pick the software improvements that are going to benefit us.
Thank you. Simon Leopold from Raymond James over here. I'm intrigued by the comments you made around supply chain management improvement. I wonder if you can give us some sense of how material this opportunity is, and also discuss what are the hurdles? Because clearly, the organizations had the opportunity to have better systems in their supply chains, and they didn't. You've given us good examples as to how it's broken. I want to get a better understanding of what has to happen for you to help fix it and how material that will be to your business? Thank you.
Yeah. You know, it's a really good question. You know, we could spend you know, quite a lot of time on that. You know, as I started off saying, you know, supply chain management is a data problem. What that means is, you know, we've historically flat out failed, you know, to bring the data together because the data originates in many different places, right? I mean, I remember you know, talking to Starbucks at one point and they said, "You know, when a truck breaks down at 4 A.M. and we can't get the tomato you know, to the ham sandwiches in the store on the corner of Main and Elm in Tacoma, Washington, that's data." You know. That's how complex these problems get.
The way that you start solving the problem is, you know, first of all, through data integration, right? If you say you're Unilever and you're selling ice cream, they don't have too many, you know, suppliers to building ice cream. It's actually a more constrained product set. You could bring the parties together, you know, that together make up the supply chain, you know, for ice cream. You know, you get them all on a multi-tenant database like Snowflake. Now we have a fighting chance of overlaying and blending data, like you, for example, saw in the presentations this morning. We're all looking at a single reality at any point in time, okay?
We're not FTP-ing files because now we have latency issues, and we're always going to be, you know, behind the ball. That's number one. You know, in order to be able to forecast anything, you need to have history. You need to have seen things in data over preferably over a long period of time. You know, one of the reasons that, you know, we've never seen a pandemic before, so there is no data, you know, to model pandemics. That is just absolutely important. Now we have a pandemic, so we are getting data. By the way, you know, we learned, you know, during the pandemic that Starschema data, which is incident data for COVID and fatality data, became a leading indicator because it correlates so well with demand.
We thought that was going to be used by public health people and whatever. No, it was used by half our customer base, right? Because you just can't predict, you know, which data series, you know, are going to be describing relationships and become highly predictive, you know, of things that you need to be able to predict. There are, you know, iterations of this problem that are so incredibly hard that you'll never catch up to it. In other words, there's gradations of difficulty, you know, in this problem. But I'm just looking at the base case, you know, where normal data operations and the history of data can be used for data science, where we're gonna get, you know, a high, you know, accuracy grip, you know, on how to forecast that business. You know, I mentioned hospital supply companies earlier.
You know, if COVID changes day- to- day, you know, you're gonna be behind the ball every single day. You'll never catch up. You'll send the wrong stuff to the wrong places, you know, all the time. That's how hard it is. You know, we have to assume that we're not always going to be in this state of complete dislocation, right? We just-- The more history we're going to develop, and by the way, the same is true for the healthcare examples. You know, the more history you have on the data, you know, you are go-- and the more complete the data becomes and the more accurate the models become, you will get there.
This is something that, you know, I remember when, you know, Mike and I were at ServiceNow. I mean, we were able to automatically route incidents to the correct, you know, resource groups. I mean, we used machine learning for that because we have historical data that could determine, you know, which incident based on its description would need to go where. We got to sort of 95% accuracy. Well, you know, those are very mundane use cases, but they worked and the customers allowed them to eliminate huge numbers of people who were manually assigning and routing these incidents, right? You gotta sort of chew this problem off in layers. You know, let's do the most foundational stuff first, and then we go to harder and harder and harder.
You got to start with the mentality that, "Hey, this is a data problem," right? I mean, supply chain management sort of got caught with its pants down. I mean, they are incredibly inept, because they have historically, you know, not gone down this path. Yeah, we've got a lot of wood to chop, you know? Same thing in healthcare, by the way. I think everybody knows that. I mean, healthcare is just very, very data-rich, but it's in terms of platforming and data discipline, it's still, you know, very early days, you know.
Hey everyone, we have time for two more questions.
Hey, I'm [Adam] [audio distortion] Big picture question towards the end, Mike and Frank, you've both been in kind of high growth organizations, as you said, and I remember in the latter days of ServiceNow, you kind of were like, "Look, I've never been in the organization where, you know, it gets bigger in the multi-billion, it's about process, you know, it gets slightly boring." This one, you get to those numbers much quicker, and it still sounds very exciting, simply the opportunity sounds bigger.
From your perspective, Frank, what's the thing that kind of, and it's a silly, you know, portfolio manager question, what keeps you awake, but where could this, from an organizational perspective, go wrong? What's the stuff that you kind of need to look out that we're not thinking? We are kind of excited about the opportunity, but you still have to build a business to kind of scale for this.
You know, you're right, I mean, businesses get to the point where they become kind of boring because it's about a rinse and repeat, not the most exciting thing, but we are just full on offense here, right?
We're creating and building and enabling, I mean, we're in the middle of, or not even in the middle of our journey, we're still in the very, very early days of our journey because it's unfolding, you know, in front of us. We're up to, you know, whatever, you know, close to, you know, what, 4,000 or 5,000 people, you know, we've not backed off of our hiring plans because, you know, I don't want to say, you know, recessions be damned because it can always get bad enough where you will change your posture. At the same time, you know, there's a huge amount of runway here, right?
You know, Mike and I, you know, I've lived through, you know, enough, you know, episodes of this. We know how to build organizations and how you scale organizations and organizations that hang together. I will tell you, and I think Mike will say the same thing, I mean, we have the best, you know, organization at Snowflake that we've ever had through all our journeys that we've had in separate companies.
I also think that we have by far the best product and technology team that we've ever, you know, had the privilege of working with, so, you know, we're just, you know, really, really focused on, you know, sort of pulling the thread and following it as fast as we know how. We want to give you a sense of that dynamic because, you know, if you think that we're just, you know, moving workloads through the cloud and that's pretty much the end of the opportunity and that's not how, you know, how this is unfolding. I want you to get a sense of that.
You know, I can't quantify it well enough for you. I know that. That's what you guys are for. You can help us do that. You know, when we get better at that, well, we're going to share that with you, you know, so. Anything you want to add, Mike, to that?
I guess the one thing I would add, and [Ramo], you're saying what keeps you up at night, I'll say it in a different way. If you look at what's the biggest risk to our business, you know, I think competition hasn't really changed. It's been the same since I came to the company and probably the biggest risk is talent and being able to attract really good talent to help our growth. And that's predominantly on the engineering side, but also retaining that talent of the company. I would say that's probably the biggest risk, but that should be the biggest risk in most tech companies.
You know, it's the lifeblood of any company, right? We can wax poetic about that. You know, the way I look at risk is, you know, it's not what other people do to us, it's what we do to ourselves. I mean, you look at any company that is not doing as well as they historically did, it's not what was done to them, you know, it's what they did to themselves, right? The things that I worry about, you know, is our culture, you know, the quality of our people. I mean, in other words, are we keeping that incredible energy and focus and momentum, you know, alive as we go through the journey?
For a lot of companies, it's really hard because, you know, the market cap gets big, and everybody starts looking at their personal spreadsheet. All that kinda, you know, all that kinda nonsense, that is what causes companies to lose their mojo. My number one, you know, obsession is the company cannot lose its mojo, ever, okay? Whatever goes into that, you know, is always my concern because the journeys are long. You guys have a quarterly aperture. I don't. I can't afford a quarterly aperture, okay? I have a much longer, you know, view of the world, you know?
Steve Koenig from SMBC way back here. Thanks for getting me in here. You know, I'm probably guilty because of my history of looking at things through an Oracle lens. You guys are obviously much more modern, supporting a lot of new workloads. I see a lot of analogies, though, in terms of stuff that Oracle went through from starting with decision support, going to OLTP. They vanquished all their hardware rivals, Rdb on DEC VAX and Db2 on IBM by being interoperable. You guys are up against the cloud vendors that have their, you know, cloud-specific databases. I'm kinda wondering, you know, you don't have a crystal ball, but what's your view of how this market evolves? You know, Oracle maxed out at 50% market share.
Do you see a market that consolidates around Snowflake, or do you see a multi-cloud world where you're a leading player? Like, where does that go, and how do you adjust your strategy as the market evolves? Thanks very much.
Yeah. You know, we have two founders who spent the majority of their professional lives at Oracle, and really gave rise to Oracle as the enterprise data platform that it became. You know, those guys are, you know, I always refer to them as ex-smokers, you know? Because they always act and think in very, very high contrast to the experiences that they've had because they didn't wanna repeat that coming into Snowflake. Much of Snowflake is about reinvention, you know, of things that were historically done because that's what they want. That's their passion, you know, to rethink even the whole concept of self-management. You know, all these things are really, really important to them.
You know how markets develop. I look at the public cloud vendors, you know. They're I see them sort of going back to the heyday of Microsoft in the early 1990s. They wanna do the integrated vertical stack. "You buy everything from me," you know, type of a play, right? Even though, you know, the more high-minded people, you know, at certainly at Amazon and Microsoft really know that, you know, they're platforms, meaning that there's many, many people on that platform. I don't know whether you've walked our pavilion here, but we have just a massive partner ecosystem. I mean, it looks like Microsoft. You go to our partner pavilion, it is just enormous, right? That is our strategy. We are not them.
We are not creating the single integrated vertical stack, "You gotta buy everything from Snowflake." The reason that we don't is we don't have to. It doesn't matter who drives the workload into Snowflake, whether it's us or somebody else. Economically, it boils down to the same thing. You know, the big sort of strategic fight is, like, are people gonna go vertical stack or are they gonna go horizontal, right? In the end, as you said, you know, in the Oracle era, you know, things went horizontal in the end, not vertical. You know, IBM lost. All the people that tried to go vertical didn't make it, you know? We're certainly betting that that ain't gonna happen here either, you know? So.
Thank you. Thank you, Frank, Christian, and Mike. Thanks everyone for coming. We hope you enjoy the rest of the summit.