Snowflake Inc. (SNOW) Earnings Call Transcript & Summary

Good afternoon, and welcome to day 4 of the Morgan Stanley TMT Conference 2021 Virtual Edition. I'm Keith Weiss. I run the U.S. software equity Research group here at Morgan Stanley. I'm very pleased to have with us this afternoon, Mike Scarpelli, CFO of Snowflake. Mike, thank you so much for joining us this afternoon.

Thanks for having me today.

Excellent. Before we kick off, I do have to read a brief disclaimer for important Morgan Stanley research disclosures, please look to our website at www.morganstanley.com/research disclosures. And as a programming note, any of the investors listening to the call. If you have any questions, you could put those questions into the webcast. This shows up on my screen, and I'll try to weave it into the conversation as we go along.

So Mike, thank you for joining us. I think it's been 2 years since you've been at the Morgan Stanley TMT conference, and this is the first year as CFO of Snowflake. So first question and just kind of a high-level question. You and Frank have been successful in scaling multiple companies in the past. You had a great role at ServiceNow. You guys did a great job at ServiceNow. As you were looking at the Snowflake opportunity, what is it that you and Frank saw that kind of was the pattern matching recognition, if you will, saying, "Hey, listen, this is an opportunity that we can make a go at that, that we can do really well, and we can improve this business and make it even better than what we found." What was that at Snowflake that kind of hit that pattern for you guys?

Yes. I can't -- I'm not going to speak for Frank, but I know the way that I thought about it was really very similar to Data Domain and ServiceNow. When I looked at those companies, it was a disruptive new technology that was taking advantage of a shift that was happening. In the case of Data Domain, it was getting from the old tape world to a more modern, at the time, storage space system, in discs that was usual access and things. In ServiceNow, it was the case of taking your legacy on-prem system and putting it into the cloud. But with Snowflake, it was really about the -- there really hadn't been any innovation in the whole data warehousing market. You have these legacy on-prem systems with a massive explosion of data that customers weren't able to use. And by putting it into a Snowflake system, you open that data up to everyone in the organization. It's just a massive market opportunity. That was what was really appealing to me. And then when I started hearing more about the longer-term aspirations, this is going back 2 years ago, the longer-term aspirations around data sharing and the network effect with Snowflake, and becoming a de facto standard platform that everyone would be using if they want. That's the way they're going to consume their data through Snowflake. That was really appealing to me. But I have to tell you if Frank wasn't going to do it, I probably would have never come to this opportunity at Snowflake. It was also to work with Frank again. And I just really enjoy working with Frank. We're very aligned in the way we think. And we just have such a great working relationship.

Yes. And frankly, you guys have done a great job in the past, too. And thus far, Snowflake has been tremendous success. So that partnership really -- works for everyone involved. In talking about the -- it must be hard to dissect kind of macro versus micro at Snowflake, just because the growth rates are so tremendous. As we come out of 2020 and the disruptions that we've seen in the past year, there's a lot of talk amongst all the vendors at the conference thus far about an accelerating pace of change, people looking to shift to the cloud faster, for people who are trying to drive their digital transformation initiatives faster. And all those transformation initiatives rely on a good set of underlying data, having clarity in that underlying data. Given sort of how fast Snowflake is growing, have you seen any of those impacts in your business? Or like it was growing so fast already, it's hard to see those impacts?

No. You definitely see the impact of that, but it hasn't been reflected necessarily in our revenue. Clearly, we just -- as you know, we just had our conference call yesterday for earnings, and Frank talked about the fact that in 2021 -- or our fiscal year in 2021 has ended, we were in various stages of 75 migrations of legacy on-prem systems to Snowflake. These are large accounts. It takes a long time to do these migrations. And so from a revenue standpoint, I think a lot of that is going to show up next year. It's going to start to show up next year on revenue and in future years. With that said, a lot of new use cases rather than build them on-prem, almost every company is realizing we have to be in the cloud, especially if we're going to have more a distributed workplace, especially if we want to give access to the data to everyone in the company. Snowflake is now pushing over 800 million queries a day, a run on Snowflake. That is a massive number when you look at it, and that's more than doubled from a year ago, from where we were at. So clearly, customers are moving to the cloud, and their workloads are growing. You're also seeing, I think we're storing over 300 petabytes of data now as well, too. That's compressed data. So clearly, we are operating in a very, very large footprint around the world. And we just see that growing. And what I would say is there is not a company that we go into that does not say, they don't want to move to the cloud. It's just a matter of when.

Yes. Got it. We definitely hear that as well. And it's really the gearing, how fast can you drive innovation, how fast can you change or drive change in organizations, particularly large enterprises that have a lot of regulatory and compliance.

Yes.

I want to shift gears and dive into the data cloud vision that you and Frank talked to a lot, because part of it is it really speaks to the fundamental difference of Snowflake and being built for the cloud versus the older legacy technologies. And I would say even more so than perhaps something like the SaaS moves that we talked about prior, where moving into SaaS-based environment led to a much faster pace of innovation. There's just a fundamental difference in sort of what you can do on Snowflake and the economics of doing sort of the queries and holding data on Snowflake versus the legacy systems when you had to pre-buy all of the compute and all the storage, and you have to presuppose everything you're going to ask of it. So part of the data cloud vision is leveraging the fact -- and correct me if I'm wrong, it's just leveraging the fact that you can have everybody query the data. You don't have to get everyone all lined up. You can have -- as you -- you could ask a question that you don't know you're going to drive a lot of value for them because you don't have to buy the capital ahead of time. You don't have to buy them years ahead of time. Can you talk about sort of the chicken and the egg there, like which comes first is like the capabilities in the solution or the vision of the data cloud?

So what I would say, first and foremost, in every new customer engagement, we start out with -- it used to be, we talked about data warehousing, data warehousing. Now when we go into a customer, we're talking about the data cloud and the vision of the data cloud and being able to share data, breaking down the silos of data within a company. So it's data available to everyone in the company. But the reality is, a customer-first has to get their data into the cloud. And so they first have to go through their migration journey. And then once they've gone through their migration journey and moved to the cloud, they can then transform the way they're doing work and we can really drive real business outcomes. And that's when a customer can start to consume data through other partners or they can share their data with other people and monetize it. I will say on the monetization today -- the monetization by a data vendor today is done outside of Snowflake. The consumption happens in Snowflake. We are building the functionality to be able to allow our data-sharing partners to monetize directly in Snowflake, where we will take care of the billings and sending the money to them as that's being paid. But there's a lot of stuff on the back end that we've been working on, quite frankly, for over a year. And it will be ready this year, to have that infrastructure in place to do that, where we can act more like an app store. But don't think of this as an app store like an Apple where you're getting 30% of revenue. This is going to be very small. It's going to be more like a marketplace in the context of like how AWS and Microsoft have it, where we'll take a small percentage to enable customers to monetize their data through Snowflake.

Got it. So can we take into the Snowflake Marketplace? And so this is enabling partners. And I think you mentioned 380 partners are now on that marketplace. You saw a 48% increase in sort of the data consumption on those partners. Can you describe a little bit about like what are the partners doing there? Give us a little bit of a sense of kind of the use case. How are people using the Marketplace? What does the end customer look like? And how big do you guys think that could get?

Well, the easiest example is like FactSet, because most of you guys get data through FactSet, you're going to have the ability to get your FactSet data directly accessed in Snowflake, where you're not having to store that data. You just access the data, you're given your provision to see and you can run your queries directly on the storage that FactSet has. Same thing with S&P -- is doing that. Those are 2 public customers, but we have many more and similar to BlackRock with Aladdin that we just announced in a joint press release this week. They're going to work on delivering more of the whole Aladdin platform through Snowflake which will make it so that anyone who wants to get Aladdin through BlackRock, it's going to be a much better experience if you're a Snowflake customer to consume that data. We're pretty excited about that. But once again, we're still in -- for that to be a meaningful piece of our revenue, I think that's going to take another year or so before that becomes a meaningful piece of revenue. But it is clearly something that helps win customers today. We're seeing the impact of that. But it just takes customers' time before they're ready to start consuming that way. And then many other interesting data providers that we're in discussions with that want to be in our marketplace, but we also have customers who realize they have very valuable data, and they're actually coming to us because they're not in the business of selling data, how can you help us monetize this data. We think we have something of real value, we just don't know how to monetize it. And we're working with a bunch of those customers right now. And that's still in the early innings as well, too.

Got it. A couple of questions on that. And you touched on this earlier, but -- so a partner like FactSet host their data in Snowflake. They enable the sharing, vis-a-vis the marketplace, their customers come in, run queries against the data and whatnot, drive more consumption. And you get paid by FactSet. They pay you for the consumption of people utilizing their data.

No. It's not quite that way. So FactSet has their instance and their storage and anything they're doing directly, but when 1 other -- when a customer actually accesses that data. They're doing it through their Snowflake accounts. And their consumption is occurring directly in their own individual Snowflake account. So we know exactly who's doing what. And we share that reporting with the data provider.

Got it. Got it. So the change that you were talking about earlier is potentially compressing that into one of where you can become a monetization avenue for?

No. So the difference is right now, is that FactSet outside of Snowflake, we'll transact with that customer on the contracting and the payment for access to that data. The access of the data happens to Snowflake. What we envision is where there doesn't need to be any offline transacting. It can all be done through Snowflake, and Snowflake's Marketplace with the payments and everything going through Snowflake.

Got it. Got it. On the BlackRock deal. It's pretty interesting that they are branding the solution, Aladdin powered by Snowflake, right?

Correct.

And that's a pretty big win for Snowflake in terms of kind of market recognition, that this is a powerful enough technology and enough of a differentiator and seen as innovation that BlackRock themselves sees that as a marketing benefit to having your name attached to their solution.

Absolutely. And there's more and more companies every day that are building their solutions on Snowflake. There's a number of ISVs out there. And part of our strategy on the ISVs because we know it's going to drive consumption. We're giving them better pricing as long as they agree to that Powered by Snowflake marketing that we have. And because at the end of the day, those guys will be the ones that will drive a lot of our consumption growth.

Got it. One of the, I think, difficult things to achieve in software is for any company to come up with like a durable competitive advantage. Software innovates so fast, it's so fungible. It's pretty easy for, not easy. But you see future capabilities catch up pretty quickly. One of the areas that you do see real sort of durable competitive advantages is when companies can establish network effects, and true network effects amongst their customers and amongst their data sets. And it seems like, to me, the data cloud vision broadly, and it is marketplace, specifically is a great area to drive those network effects. Can you talk to us about how you foresee network effects driving from the marketplace? And any early indications of that taking place?

Well, 20% of our customers are sharing data in some form, today. We think there's no reason why in the future, all of our customers can't share data. And once you have that network effect of people sharing data, it makes Snowflake extremely sticky. We become the de facto standard. Think of any company with its suppliers where they force their suppliers to be on Snowflake to share data, whether it's inventory levels or anything, and we're seeing that start to happen now. I was just on the phone the other day with the customer, and they're demanding their new vendors be on Snowflake because that's how they want to consume everything. And we're seeing this is happening real time. And so I think that is probably the most competitive differentiator for Snowflake over any of the other technology out there that we compete with.

Got it. If we think about the broader set of solutions around that core data warehousing and data cloud capability, the data science side of the equation, the Snowpipes, Snowsight for the data visualization. As investors look at that, should we think of those as additional monetization opportunities? Or is that just you guys helping the customers pull more data, do more with their data within the core product, and still the fundamental monetization avenue will be storage and compute against that data set?

All of these things drive compute consumption, and that's how -- where our revenue is coming from. For instance, Snowsight. That's our visualization tool that is available to every customer. It's only for their Snowflake data. It's optimized for Snowflake and customers have it. It's for free. In essence it's like Snowflake. Same thing with Snowpipes. Everyone has the ability to use Snowpipes. It's to help get data into Snowflake easier. Why? Because it's going to drive consumption. But we are very much an open platform. You can choose the BI tool you want to use, whether you want to use Tableau, you want to use Looker, you want to use Power BI, you want to use Snowsight. That's fine. You choose your thoughts about same thing with the various AI companies out there. We'll work with any of them. You, customer, however you want to do things, as long as it gets data into Snowflake, we're okay with that. And we are very much focused on our Snowflake ventures around looking at investing in companies that are going to drive consumption on Snowflake. That's why we announced we did an investment with DataRobot, Hunter. We did one with Knoema. Knoema has become a great data-sharing partner, and that is -- their data sets are being consumed by many, many customers today. Same thing with Laceworks. We announced our investment in Laceworks. Why? Those guys are built on Snowflake. As they grow, their consumption of Snowflake grows, and that's good for us. So we're very, very focused on anything that will drive consumption at Snowflake. And I really don't see our business model changing. I love the simplicity. We really sell 1 product with 3 different SKUs, and the revenue comes from really 2 sources, 3 and potentially a 4. You have compute, which is the biggest, then storage, data transfer, which is really just a pass-through, and that's very, very small. That's when people want to move from 1 cloud to another. And then eventually, 1 day, whatever fees we get out of our marketplace. I don't see that as being a material amount, but it's all going to drive consumption, and that's where we make our money.

Yes. Like it -- it's very reminiscent of like the earlier days of like an Amazon or Google, tremendous amount of investment being made in product innovation, but that product and innovation is all about driving more volumes to the core because the core is so good and it's such a great monetization vehicle.

Correct.

Got it. I want to talk about sort of the market opportunity in front of Snowflake. And the way I'd see it, you can kind of -- it's almost like a layer cake, if you will. Today, the core market opportunity, you could look at sort of the data warehousing opportunity. You guys talked about the competitive displacements of some of the legacy vendors like Atera Data. If you look at IDC, $16 billion market, of which only 25% is in the cloud and the vast majority is going to the cloud. That's kind of the here and now. You guys talked about unstructured data, right? Expanding into unstructured data. You gave us a little teaser, wait for the second half, and we should hear some more announcements on that. Why is it -- besides just a broader set of data to go after, what is it important to go after that unstructured data side of the equation?

Customers are asking for it. And that unstructured data is growing, and customers want to be able to gain insights from that and analyze it. And it's a very difficult thing to do. And it just opens up so much more market opportunity for us. I do have some concerns on it, that it will have an impact on storage because unstructured data doesn't quite compress the same. But clearly, it expands our market. But this is a massive market we're going after in a growing market with data exploding the way it is. And I don't necessarily believe the market is as small as what IDC has said. But the other thing to note, a lot of people think of we're just going after these big on-prem legacy systems, it's not that. There's also a lot of workloads that were -- customers couldn't afford, those big legacy on-prem systems, and so they put them in transactional databases like SQL and other things. And there's a lot of those workloads that are moving over to Snowflake. And there's going to be more of those moving over to Snowflake as we continue to improve our performance. As we improve performance, more and more of those workloads can come to Snowflake.

Got it. And I mean is there a road map to Snowflake? It's not like is -- today as an analytical database, is there a road map so like also being a transactional database?

I would say we have customers asking us, why can't we do more transaction-like processing for them, so they just have one. And so we do have a road map to continue to improve performance meaningfully. I think Benoit has an aspiration that he put out there, too. Benoit, our founder, to improve it by 10x. And I think that's aspirational. And we do have people looking at OLTP-like architectures that we're working on as well, too, but that's a long ways away.

Got it. You brought up an interesting point about sort of the IDC categories. I remember, ServiceNow, if you go back there, when ServiceNow, when we first were doing the IPO on ServiceNow, the same pushback was like, "Oh, it's small market ITS, sounds like only $4 billion." And now ServiceNow is bigger than the ITS, what would have been the -- I think it was actually small like $2.5 billion and ServiceNow is now bigger than what it would have been. And there's a dynamic of when these solutions like ServiceNow come into the market, when you push it to the cloud, because you expand the scope of what you could do with the solution, it really expands the market opportunity. And you talked about this a little bit. On the low end of that, it lets more people in the door because of the subscription model and the consumption model. On the high end, do you find your customers sort of expanding into use cases already that they wouldn't even have imagined on the high end of it because of the dynamics of what Snowflake can do and the price points you can do with that?

While I look at one of our largest customers and what they're doing around Snowflake is not they're actually using us, a big chunk of their consumption through a security data lake, which I never envisioned them doing. So there's many different use cases that our customers are pushing at or going into on Snowflake. And many times, they're finding these on their own. I still think we're in the very early innings of that with the very large customers. But clearly, the IDC and others is really a look back at how people were doing things. And you really need to look forward and just the technology that has enabled access to everyone in the company is exploding the market in which we operate. In the concurrency, where in the past, you could never have the number of people accessing the exact same data at the same time. Your -- and because of the speed, what you might have been running a query once a day or once a week, you can now literally run every day or multiple times a day, which was never available before, so you can make real-time business decisions. And that's the reason we got a lot of people asking, well, but the cost is as expensive for customers to do. I guess what -- are you just looking at a cost? Are you looking at value? You have to look at what is the value of the customers getting out of this. If the customer didn't see value, they wouldn't be doing it.

Got it. I'm looking at the questions that the investors are tossing in, and every single one of them is on the competitive environment. So we'll hit a couple of those head on. So one question is, what's the data cloud moat that makes it irreplaceable when compared to Big Query or RedShift or Synapse or Microsoft, the other hyperscaler, their versions of the data -- cloud-based data warehouse.

Well, I guess, first of all, they weren't built to be data-sharing platforms. We were built to do this from day 1. And we are really the only one that was built for the cloud. All of those products were on-prem products that have been repurposed for the cloud, even Big Query, first off. Second, we're the only 1 that's multi-cloud. None of those guys are multi-cloud, even Google talked about their omni-channel, but that has never -- show me a customer who's actually doing that. It has failed miserably from everything that I've heard. And so that is really the competitive moat that we have. None of these guys have figured out how to do this.

Got it. You mentioned the security use case. There's a question here about competition with Splunk. Splunk is a data vendor who's sort of positioned into security and uses that data for the security use case. They -- one thing that Splunk has gone into more recently is streaming data, and being able to analyze real-time data stream versus just data at risk. Is that something that Snowflake has ambitions of going after, of looking at real-time data flows or being able to analyze data in motion versus just that rest?

Well, I will say, not talking about Splunk itself, but that whole real-time data, customers want that. And that is something we will be doing more of in connection with more of a security. It's customers pulling into that. We didn't -- we're not going to customers and selling them on a security data lake. Customers are realizing they can do certain things. I'm not saying we're replacing Splunk. I think we augment some of the things that Splunk do, and our customers choose to do that. So yes, I think we can do more of that, and we will be doing more of that because that's more valuable to customers.

Got it. Shifting to more of the cost side of the equation. In the -- you came onboard Snowflake, and I think you saw a very good ability to sort of optimize on the cost side of the equation to drive towards cash flow breakeven. And you gave a target right out of the gate, and you're exceeding that target, and we'll be coming to cash flow breakeven this year. In calendar '20, it seems like a lot of that cost efficiency or a big chunk of it came from being able to renegotiate some of the agreements with the hyperscalers, get better pricing because of the volumes you're driving from them. Can you talk to us about sort of how much of that stone has been squeezed, if you will? And then on a go-forward basis, where do we expect to see further efficiencies, particularly on the gross margin line on a go-forward basis?

So clearly, our renegotiation with cloud vendors, mainly being AWS and Azure was a meaningful impact, and you started seeing that at the end of Q3, and you saw that for the full impact on Q4. I fully expect as we move further along in those contracts, and it's not probably going to be a year, 2 years out, we will renegotiate those contracts as long as we see that we're tracking ahead of our consumption forecast. Because at the end of the day, they're only going to agree to a contract amendment. If we're willing to commit more and extend longer. So I fully anticipate we'll continually get into a cadence of doing that. By the way, like what our customers do to us every couple of years, I would do the same thing. But on top of that, we have a lot of different projects undergoing within our cloud engineering group to take costs out of our operations to support our customers. There's a lot of optimization going on around the way we do some of our monitoring. And as we get more scale within our data, we have a lot of deployments. We have about 20 deployments around the world. A number of those are not nearly at the scale we need to be at. Because remember, when we sign up a new deployment, think of it, we have to we have a -- think of it like the dial tone on a telephone. Someone told me, you have to pay to get that dial tone, and then there's a usage. So there's a flat amount of money we have to pay for every deployment. Until you have a critical mass of customers, your margin is very low in those data centers. We have a number of those around the world. So as we put more customers, that will improve it there. And then just with scale, we have really focused on building out our support organization as we move up into larger enterprise customers. We have prebuilt our support organization in advance of some of those customers coming so as we add more customers, we'll get more scale there. I feel very confident we can get to the mid-70s in gross margin in the longer term. I would say it's even kind of medium to long term in the next kind of 3 to 5 years. Beyond that, there's a lot of unknowns out there. And I said this when we were going public, I don't see us ever being like a SaaS company where we have those 85% product gross margins. The reason being, we have a lot of storage. As we move more into unstructured data that is going to take up more storage as people use us more for backup as well, too, with some of their data with the data lake, that costs more storage. But the offset to that as we get more into data sharing, that really just drives consumption, not storage, which will offset that. So let us get to the mid-70s, and then we'll see where we can go from there.

Got it. And in terms of the -- like the growth versus margin framer, from my perspective, like from an external perspective, it seemed like you had a very strong -- a strong desire to get to free cash flow breakeven. Now that sort of you're basically there. You had positive free cash flows in this quarter, and you're guiding it to -- for FY '22. How should we think about the philosophy about growth versus margins on a go-forward basis? There's a huge market opportunity. You have a competitive lead. There's new big markets that you're trying to get into. The -- is breakeven kind of where you stay in the near term to address those opportunities? Or is it going to be growth and profitability, both driving higher?

Yes. What I would say is we are, first and foremost, a growth company, but it's not growth at all costs. And believe me, when I want to say, we are spending money as fast as we can to grow this business, but we're not going to spend it on stupid things. And COVID has helped last year with T&E being reduced dramatically. But that doesn't mean we increase so, and we used do a lot of field marketing events. What -- we pivoted and everything went to digital marketing, and we spent a lot more on digital marketing to support that pipeline generation. And we'll continue to do it. And what the yield that we were seeing out of that, I don't think we're going to be going back to doing the number of field events that we used to be doing. Most of it's going to be digital. Same thing, too, with -- we just had our sales kickoff. Our sales kickoff was a virtual sales kickoff. A lot of our salespeople were very disappointed before the event that we were doing that, even our sales leadership. But then after the event, the feedback was, everyone loved it because of the learning that everyone got out of that event. And we were also able to repurpose everything because we recorded -- a lot of the stuff was prerecorded, that we can now use that as we onboard new salespeople throughout the year, we can repurpose all of that collateral. And so much so that next year, we've just decided we're never going to be doing a sales kickoff in-person again. They're all going to be done virtual like that. Our users conference. I'm not going to say we're never going to do another user's conference in-person. But because of the lead time for planning, this year is going to be done virtually as well too, and the feedback from our customers was they really enjoyed that virtual event last year. So it's -- stay tuned. And I do envision that we will start traveling again the second half of the year. But it's going to be customer-specific travel. There's not going to be any travel where someone from Atlanta flies to the San Mateo office to do meetings. Do them over Zoom. There's absolutely no reason to get on a plane. And I'm putting controls in place that people won't be booking travel to do internal meetings like that. It's just not necessary. By the way, likewise, I never want to meet another vendor in person. I'm happy to do it on Zoom. It's a much better experience from my perspective. Because I can end the meeting very quickly, there's no lingering.

Speaking of that in the meeting, a little bit over time. I want to leave with one kind of closing question. You guys are doing an Analyst Day in June. So there's going to be more sort of excitement to come in, more product announcements to come on that, that you guys have previewed. Outside of the stuff that you haven't told us yet, what part of the story do you think investors don't quite get, yet? If you have to put your finger on something that is less appreciated about this Snowflake story that you're excited about that we should be focusing on more in calendar '21?

Yes. I don't know if it's stuff they don't get. But I think probably the least understood is the power of the data cloud and the whole network effect in data sharing and what that can do for us long term. And I think a lot of people also still do not appreciate how hard it is to do these large legacy database migrations to the cloud. There is a lot of work involved in doing them. And that's one of the reasons why our GSIs are excited because it's big dollars for them. The likes of Accenture, Deloitte, Infosys, Cognizant, they're all over this right now. And they're great partners.

Outstanding. Mike, thank you so much for joining us. This has been a great conversation, as always. Congratulations on all the success in Snowflake. Looking forward to how the story evolves even further in the year ahead, and looking forward to maybe having you in person at our conference next year.

Thanks, Keith.

Outstanding.