Snowflake Inc. (NYSE:SNOW) Piper Sandler Growth Frontiers Conference September 13, 2022 10:00 AM ET
Company Participants
Mike Scarpelli – CFO
Conference Call Participants
Unidentified Company Representative
Good morning. Super excited here to be in Nashville, and super excited to have Snowflake Computing present here. Getting things off this morning, we have Mike Scarpelli, CFO of Snowflake. Mike, welcome to Nashville.
Mike Scarpelli
Thank you, [Brent]. And by the way, we used to be called Snowflake Computing, before we went public, we dropped the computing. Snowflake Inc.
Unidentified Company Representative
Shows that probably — [deal] with you guys way too much prior to the IPO…
Mike Scarpelli
Thank you for having me.
Question-and-Answer Session
Q – Unidentified Analyst
Absolutely. It’s great to see everybody in-person again. It’s pretty special to after three years to see a lot of familiar faces in the audience here. Over the next 25 minutes, so it’ll be a fireside chat. Three things that I’d love to cover and hopefully this audience comes away with a better appreciation for Snowflake Inc. One, secret sauce. What makes the technology, the product, the pricing so unique, if you are able to go out and compete and win against the likes of Amazon and Microsoft that could be products. Two, world automation 2.0 is one of the themes we talked about this morning and a big theme that we are going to be focused on here today. And then three, the growth equation.
Mike Scarpelli
[Technical Difficulty] the auto provisioning that we do to spin out new clusters where customers don’t need to do anything. Unlike some of the other products out there, it takes more highly skilled people to do that. The second, we’re really the pioneer in figuring out how to split or separate the compute from a storage, and then with the global service layer on top that allowed the — to do multi cluster, so you could access the same data to run analytics at the same time. As the speed at which we figured out how to do that and that concurrency is a huge thing that was very unique with Snowflake. Now, others are catching up to that. And no surprise that we think we’re still many years ahead. And ever since we went public, probably the closest to us is Google with BigQuery, but we still continued to win. I just — a lot of people talked about Databricks and Databricks is really more on the data science and data engineering side, we were never there. Actually, we brought them into a number of our accounts that coexist in over a thousand accounts.
But with Snowpark and Python support for Snowpark, we go after Spark [jobs] down. We are seeing Spark migrations now to Snowflake, that’s something that we never dealt with before. But I would say that those are — that’s the part about the product. But our strategy, the other thing that’s unique about Snowflake is the way we do data sharing. And data sharing is one of the key things that is landing as many accounts that when we do data sharing, the data is never transferred, we just give people access to the data, and they can run their queries directly against that data. The person who owns the data controls all the governance for who has access. And that’s a really, really important thing. People talk about data sharing other companies but they don’t do, they actually transfer data, not the same.
The other thing about Snowflake that’s unique, because we principally compete with Google, Azure and AWS. We’re multicloud. We can run on any one of those clouds. And that experience, from a customer perspective, they had — unless you — if you never told them, they would not know which cloud they’re running on, it is the exact same experience for them. And that’s a really important thing. And that’s part of our strategy too and what the real end game of Snowflake is native app development. And when you can offer a platform for native app development that it will automatically work that app in any cloud. We get to understand is when you build an application to an AWS, you have to port that application to GCP or Azure, that’s a lot of work, it doesn’t just work and that’s really important. So I’d say that’s the most important piece from a product standpoint, there’s a lot of other things too…
Unidentified Analyst
Yes, from a product standpoint, you did say that there are people trying to catch up, maybe on that separation of compute and storage and some of the…
Mike Scarpelli
Everyone pretty much can do that now…
Unidentified Analyst
Will do that now. Data sharing seems to be the one that does stand out when we talk to customers. Why is it so hard for Google to replicate that, or Amazon or…
Mike Scarpelli
You really have to understand, we were built to do this, that was always the vision of Benoit and Thierry, the Founders. Remember, BigQuery was not built for the cloud that was actually built on an on prem system for running things within Google. And then the reposition, even Redshift was not built for the cloud. Synapse is a bunch of different technologies that they’ve tried to cobble together and repurpose for the cloud, but it was never originally built for the cloud, their architecture is very different.
Unidentified Analyst
And you said, Google BigQuery went to the time the IPO was kind of the company that was the closest to where you’re at. They’ve lost a lot of engineering talent. They’ve all kind of gone off and then joined Firebolt. What have you seen…
Mike Scarpelli
We have a lot of new Google talent.
Unidentified Analyst
We have a lot of Google talent as well.
Mike Scarpelli
Yes, because I know Firebolt, it is a small company that they struggled with scale. Yes.
Unidentified Analyst
But has Google dropped off from a competitive…
Mike Scarpelli
No, absolutely not. Google is very competitive. What I would say is, AWS of the three clouds is probably the friendliest to Snowflake. And you know, I think last year, we, call it, sold about 1.2 billion with them. Microsoft is next and GCP is when we have the, I would say, the — we have a relationship, that’s not very strong relationship. And GCP is the most expensive cloud for us to run and their pricing is not very good with GCP, where AWS gives us great pricing. We’re in the midst of renegotiating a contract with AWS. And I know Microsoft will follow.
Unidentified Analyst
Great to hear…
Mike Scarpelli
It’s funny, everyone was concerned last quarter that how can you guys be growing when Microsoft was seeing weakness, and that’s why I pointed out 80% of our business is in AWS. It was funny they got — Microsoft execs to actually call us to type in a relationship with us.
Unidentified Analyst
Even though pricing [Multiple Speakers]…
Mike Scarpelli
Yes.
Unidentified Analyst
Let’s talk about clean rooms. So I was at the Snowflake summit here earlier this year, and CIOs of Walgreens and Hershey’s up on stage talking about excitement around clean rooms. Never really heard much about clean rooms. Talk to me a little about why there’s excitement, why is there a buzz around clean rooms?
Mike Scarpelli
Data clean rooms is another thing that we have that is extremely exciting. Where clean room does in essence is that you have two parties simplifying it, you have two parties that have data for whatever legal or regulatory, competitive, they can’t show one another the data, but you can put that data together, you can run analytics on it, to come up with outcomes that are you are able to share. As an example, you have a retailer, you have a streaming media company. The streaming media company knows every one of its subscribers. The retailer has all of its loyalty members. They can join that data, the data clean room and figure out exactly what subscribers to target in advertising, who are loyalty members of that retailer. That’s a prime use case for the data clean room. The data is never transferred to you, that are all through data sharing.
Unidentified Analyst
One customer that we talk to told us that the killer app was a double blind query using data sharing for Snowflake. And I was like, okay, now figure out what a double blind query is. But absolutely hearing more and more around those use cases…
Mike Scarpelli
And we do that with partners. We have the technology with the customer partners that have the expertise. And some of these things you have guys like [Indiscernible], you know have those guys — there’s actually, publicly guys that [Indiscernible] with the stack, Trade Desk, sorry I can’t remember the name. There is another big one doing clean room using Snowflake.
Unidentified Analyst
Absolutely. You start talking about stable edges, I want to say, a year ago maybe. Why is that an important metric to track. And as you we think about that metric, clearly, your largest customers are where you’re seeing the most of the stable edges. New Track based on at least one edge. But is there an environment where you have multiple edges, dozens of edges, hundreds of edges. I don’t know. But walk us through stable edges and why that metric matters.
Mike Scarpelli
Yes. So what a stable edge is, is a data sharing relationship between two parties and what a stable edge is, is a persistent data sharing between those two parties. And we look over a six week period time, but they have to consume a minimum number of credits because that really shows it’s being used actively in whatever business they are doing, they’re data sharing for. And that number continues to grow. It’s actually the — where we really see data sharing is in financial services. You look at the likes of a lot of the financial service companies that we have. As you guys know and hear, many of the banks share data, whether it’s through an intermediary, the likes of the Fidelity and others of the world with their private wealth clients or their family offices, a lot of data sharing that happens and transfer data. And that creates stickiness. It’s known that BlackRock, we’ve talked about them within their lab. They’ve been replatforming on Snowflake, and that is starting to take off right now. That is leading to new customers for Snowflake that we were never even trying to sell into.
Unidentified Analyst
Clearlydata sharing, stable edges, big area differentiation continues to be an area where customers are engaging more with you. But there are some new things you guys announced at Snowflake Summit. Last question on the product side, Python, UniStore, Zero Copy Cloning. What are the things that you think is going to resonate and matter the most to customers thinking out in the next year on these new features?
Mike Scarpelli
Well, they’re all important. I would say, Snowpark for Python in the short term that goes GA this quarter, that will be — it’s Q4, I think it was early November. That is going to be meaningful because of the number of large customers we have that are — they are pushing us to get to GA sooner, where a lot of customers won’t go into production until things are in GA. And the reason they won’t go into production until it’s in GA, because many of our customers have big liability limits of something, there is data corruption or anything. And so we don’t rush to put anything in GA until we know we are confident that it will work. And this is why a lot of the features we release and a lot of people are understanding, they take two to three years to build out in a database. Not a feature in an application that you just think you are going to do something and you roll that feature over the next quarter or that quarter, it’s not possible in the database world because you’ve got to get everything right to make sure you don’t have problems.
So the other thing is UniStar is important, Iceberg Tables, a [patented] support for Apache Iceberg Tables. Apache Iceberg seems to be the format that many of the world’s largest companies are choosing for their formats they want to use and enables our customers to run Snowflake using the compute with storage in your own storage. And the performance is just as good as it’s in Snowflake. Then the other thing that is super interesting for us is native apps. Native apps comes next year, that’s when developers can build applications directly on the platform. And that ties in a lot with UniStore as well too because a lot of these applications are — you need transaction like processing speeds to do some of these things. And our UniStore is not to compete with the traditional transactional databases, this is more for new workload types. And people building new applications is where we see that happening. The other thing that’s really exciting to have some of our largest customers excited is we just rolled out a new large warehouse with extended memory, which allows the computational processing of massive datasets. Within Snowflake we have one customer that had been trying to do it in various other public clouds and it was not possible. And they have now done it in Snowflake, and you may think that’s a lot, took them eight minutes to process that amount of data. And I’m talking massive data where it was not possible for them to do before in one data center.
Unidentified Analyst
Any questions from the audience on product differentiation, [Indiscernible] before we move to pricing and people?
Mike Scarpelli
Sure.
Unidentified Analyst
Okay, one here.
Unidentified Analyst
So where do you think we are in the adoption curve of that, and what — are there any kind of technological issues that [staying] fast adoption curve or how do you see that…
Mike Scarpelli
No, there’s no technical issues for slowing down adoption. The biggest thing is we’ve been selling data sharing to our customers from day one as part of our selling motion, and that’s what lands a lot of customers now with aspirations. But they first need to get all their data in Snowflake and whoever they’re sharing their data with needs to be Snowflake on the other end. We’re still in the — what do we have 21% of our customers, are using data sharing today. And that continues to grow every quarter, we give that metric, it’s an important metric, it’s how we measure the business. And that really creates a stickiness, because once you start using Snowflake for data sharing, you can’t get off of data sharing. Once you get into that network effect, you all have that, whoever that supplier or vendor, whatever customers, need to be on Snowflake. And that’s why it’s so important to us and why we track that metric. Quite frankly, half of the — it’s one of the key metrics for whether management bonuses get funded or not the growth in the stable edges.
Unidentified Analyst
What do you guys doing in data streaming [Technical Difficulty]…
Mike Scarpelli
Yes. To date, Conflux is more of a partner of ours. Definitely some of the things that we have are getting us to the edge of it, but we’re not trying to go and replace Kafka, by any means. I think Kafka, it’s a great company.
Unidentified Analyst
[Question inaudible]
Mike Scarpelli
It’s both. And when you talk about [Indiscernible], the most common way that people do data is it’s an FTP download and that is the most archaic way to do things. And I almost guarantee every one of you and your companies does that still today. And that technology has been around, we are replacing a lot of that with customers. But like some of the biggest financial services companies all — and then there’s also a lot of new use cases where we’re doing that data, and clean room technologies is a prime example today.
Unidentified Analyst
[Technical Difficulty]
Mike Scarpelli
We believe that data sharing is the key differentiator, and you must have that to win in this space. And we just don’t see the other players out there having the type of data sharing that we do. I know they all talked about it, but I would say go look at it under the covers and what is it?
Unidentified Analyst
Yes. It leads to the next question around other areas of secret sauce, which is partially people and pricing. But on the people side having those engineers and talent, how differentiated is just the team you have built? Both engineering and sales and marketing?
Mike Scarpelli
Well, I would say the engineering is the key piece not to diminish sales people in any way. But engineers are the key for any technology and product company. And what’s impressive about Snowflake is Benoit and Thierry, the two founders, Thierry, this is a guy who’s made a lot of money. He’s been at it for 10 years. He’s around my age, he’s like 55, 56, whatever. He still reviews code for engineers. And I was talking like Thierry, why do you do that? Isn’t that a waste of your time? And he does it on weekends, it’s like he has no life but he likes to code. And what he said was he goes, you need to teach these young, in his French accent, you need to teach these young engineers that mistakes they make what it costs, and I need to be able to teach them so that they don’t make them again, and they can teach others. So just amazing engineers, I mean, Benoit and then Greg Schakowsky who runs engineering. He super smart, humble guy. He was at Google for 14 years, he ran BigQuery for 10. He is an unbelievable recruiter of top talent. He has brought in a number of — we have a few fellows and [counting], there were 13 fellows at Google, we have a couple of them, we have lot of other senior people that he’s brought in, not just [Barry Harrison] from Facebook, Microsoft. He also was instrumental in building Warsaw, Google’s Warsaw office where they had like 2,000 engineers. And we now have a big presence in Warsaw and a very growing presence there, because we get good engineers there. And then Christian Kleinerman who runs product is amazing. Similar to you, he likes everything.
Unidentified Analyst
Technology [bolt]
Mike Scarpelli
I’ve known Brent for 16 years, because I always joke with him, that there’s not a technology he doesn’t like.
Unidentified Analyst
I am an unabashed technology bolt for sure…
Mike Scarpelli
Yes.
Unidentified Analyst
Pricing. So probably the unsung part of the differentiation, lot of people could copy pricing. But how do you think about that usage based pricing model being an advantage or not relative to competing legacy technologies?
Mike Scarpelli
You know, it’s funny, I’ve talked to a number of companies literally every week, and switching to a consumption model. And one of the biggest challenges of a consumption model is actually having these — there’s no off the shelf systems to actually track your usage and metering, coming up with what is that unit of measurement to be able to do that. And we were fortunate enough when Benoit and Thierry started the company long before I joined the company, they made the decision to do this and they actually used Snowflake to be able to track all the usage, have all the metering data, we are able to pull in all of our contract data into Snowflake, get all the pricing and our pricing tables are all maintained on Snowflake to actually be able to track the usage and in an audible format for our customers if they want, where we produce all of our monthly statements to customers and everything out of Snowflake. Why I think it’s important is, we have over 300 different SaaS applications at Snowflake that we use. It’s probably 20 major ones. But it’s unbelievable even with SaaS applications, the number of licenses you have that are not being fully used. And we have actually using Snowflake. We take all of our Okta data, because you have single sign on and all your authentication has to go. So I know exactly who is using what, how many — when’s the last time they logged into and we actually started going through a process of decommissioning licenses if someone hasn’t logged into something for 90 days or 60 days to manage what — it’s a pain and the ask to manage on those licenses. And as a CFO, you are expensing stuff that you are not even using.
In a consumption model you buy capacity to customer, that’s why I tell people billings is a meaningless thing. I don’t look at billings when someone asked me on the call the other day, because customers just buy capacity. We have many customers that just [co-term], they buy one quarter of capacity. And they consume that and they get — usually it’s a year to consume that if it’s under a one year contract or whatever it is, some is three years, but we do let customers roll it over if they continue to renew with us. And so from a customer’s perspective, they can — they are only expensing what they are actually using, but they can control their usage with Snowflake. And the beautiful thing about Snowflake too from a go-to-market standpoint, we really have one product, it’s Snowflake and that product — it’s one SKU. And that product is used by whether you’re a two man in the shop or a million people or 500,000 in your shop, and there’s three different versions of it. But the beautiful thing is every new release at Snowpark, we have to go and sell that to a customer again. They just have it. Data sharing, they have it. And that just drives usage in the metering engine based upon the rate tables for the different things how to use it. So from a go-to-market, you don’t have to go on an upsell and customer every quarter when you come up with the new feature.
Unidentified Analyst
Two quick questions here. One is on automation. You talked about 300 apps that you have internally. One of the areas we think about where there could be actually increase in priority is around automation, like automating a process. Are you seeing that show up internally as an area you want to invest more in on the software side? And then externally in that data operation space, we are hearing Snowflake basically is becoming a instrumental platform to basically transform your new data operations, and so the role there would be helpful. And then internally, are you prioritizing automation internally or not?
Mike Scarpelli
Yes, and that’s not new though. Since the day I got to the company, I’ve always been investing in automation and systems. And as I tell people, systems don’t leave, systems don’t ask for more money. And I’m huge into automation. And so we have always done that. And we used a lot of Snowflake on Snowflake for all these different use cases that we have. So, yes. I don’t see us slowing down our investments in automation. And I do know customers too want automation. And one of the biggest things for automation is being able to seamlessly get your data into one central place. So it’s not a manual process. And we do see customers, some of our largest customers want all of their ServiceNow data in Snowflake so they can help with their IT operations management for all the analytics. Same thing too with the most common use case for Snowflake is all your CRM data into Snowflake.
Unidentified Analyst
Last question for you. Twenty second answer. What has to go right to get that $10 billion target that you have out there? If you think about the business, execution at this point, new products, what has to go right to get to scales from $2 billion one of the fastest growing enterprise companies [to date]…
Mike Scarpelli
You know, the biggest thing is talent, continuing to attract and retain talent. So we can continue down our path of our product roadmap. And we have a lot of interesting things that will be coming out over the next few years. But I will tell you, we can get that $10 billion with our product set today and the things we have announced already. When we do our forecasting, we forecast based upon historical consumption patterns of products that customers are using. We don’t forecast based upon new products, because you really don’t know until customers start to use it.
Unidentified Analyst
Mike, we’re out of time. Great having you here in Nashville. Thank you so much for coming.
Mike Scarpelli
Thank you for having me.
Unidentified Analyst
Take care.
Be the first to comment