Snowflake, Inc. Q3 FY2025 earnings call

• Strong third quarter with product revenue growth, outperforming expectations and increasing FY2025 product revenue guide. Customers value Snowflake as the easiest and most cost-effective enterprise data platform. • Launched the same number of tier 1 features to general availability in Q3 as in all of fiscal 2024. Snowflake Cortex AI has over 1,000 deployed use cases and over 3,200 accounts using AI and ML features. • Interoperability features have a >$200 million run rate as of the end of Q3, with partnerships with Microsoft and ServiceNow to increase data interoperability. • Snowflake world tours had 29,000 attendees across 24 in-person events, with a 40% year-over-year increase in attendance. • Focus on cost management, including centralizing teams, removing redundant layers, and using AI for efficiency. • Notable customers like a global telecom giant using Snowflake for network performance data.

Product revenue for the quarter was $900 million, up 29% year-on-year. Remaining performance obligations totaled $5.7 billion with year-over-year growth accelerating to 55%. Snowpark is well on track to represent 3% of product revenue and growing nicely. Approximately 500 accounts are adopting Iceberg, and the contribution from data engineering features like Snowpark, dynamic tables, connectors, and Snowpipe Streaming is expected to more than offset potential loss of storage revenue.

• For the fourth quarter, expects product revenue between $906 million and $911 million, representing 23% year-over-year growth. • Increases FY2025 product revenue guidance to approximately $3.43 billion, 29% year-over-year growth. • Increases non-GAAP product gross margin guidance to 76%, non-GAAP operating margin guidance to 5%, and expects approximately 26% non-GAAP adjusted free cash flow margin for the year.

• Competitors' technology is highly complex, leading to higher risk of engineering mistakes. • Potential impact of economic conditions on customer spending and budget allocation.

Q: Hi, thank you Mike. It's impressive to see the strength here simultaneously in both the consumption revenue and the bookings, especially given the prioritization of consumption incentives this year. I'm curious to what you might attribute that and specifically whether Iceberg tables might have contributed all or whether Snowpark might have picked up in any meaningful way? And then I have a quick follow-up.

A: I would say we're starting to see the positive benefit of Iceberg with a number of customers that are now bringing new workloads that are now being addressed by Snowflake and Iceberg tables. But I would just say it's broad-based demand across our customers. Yes, there is a few verticals that were very strong technology, financial services and health care. But it's really broad-based, and we are seeing the uptick, as Sridhar was mentioning, a lot of the data engineering stuff, as well too and Snowpark is part of that.

Q: Hi thank you very much. One for you, Sridhar and one for Mike. Sridhar, there is a narrative, which you will definitely dispute that the core of the Snowflake Data platform that structured data does not really have a long runway in the world of generative AI and that also Snowflake has a lot to prove with respect to generative AI on the unstructured data. What proof-points can you talk to the quarter that would invalidate that bearish view and reinforce your conviction. And one for you, Mike, with the headwinds from storage not being as much as was dialed-in, should we safely assume that you're guiding to product revenue, 29% for the year or takeaway 3 points for Snowpark container services that the core is actually at a point where you can see it be stable going to next year. I know you're not giving guidance. And if we can start to think about doing the dream of all the new products to be largely incremental to that growth, right? Thank you so much and congratulations.

A: Thank you, Kash. On the core business side, analytics still is going to be pretty important, getting the most important data about your business, but increasingly being able to act on it quickly in real time, is the thing that is going to set great companies apart. If you look at the best companies that have been created in the past like 2, 2.5 decades. These are companies that have [integrated] (ph) data into the core of how they are operating and when I talk to customers, not just about the analytics, the view of clean data, but also about being able to act on it, being able to see trends, being able to figure out things like guest experiences like our customers Hyatt and Disney do at scale. So there's a very long runway because analytics flows over seamlessly and fluidly into things like machine learning. And AI then becomes even more of an accelerant because you can now go from unstructured data to structured data very, very easily and that's the magic of products like Cortex AI and the new things that we announced in build, where you can bring multimodal models. Imagine a world in which you just write a SQL statement that goes to act on a PDF and produces a bunch of structured information out on the other side. It redefines what you and I think of as analytics because just a lot more can be done. And so that's the world that we are driving towards, and that's where investments in companies like Datavolo that bring even more data into Snowflake is exciting and empowering for us, as a data platform. And then on the other side, when it comes to unstructured data or just AI applications, as I said in my remarks, we have over 1,000 deployed use cases. And in all of them, these are not tie deployments. We work with our customers. We make sure that they get value from it. That's the first thing that I tell all our customers. AI needs to be a business accelerant. It's not a hobby and the products that we have created, which do things like take trustability on the work that we are doing with the TruEra acquisition, for example, that brings observability to how people create AI applications are the ones that are creating rock-solid applications and where increasingly, the difference between structured and unstructured is going to be less and less meaningful as we go forward. And then in terms of examples, there are a ton of them, Siemens, Bayer, these are all folks -- Zoom, these are all folks that have used our AI product, got immense value and talked publicly about them. So I feel very good about where we are executing on that side.

Q: Hi, thank you. Sridhar, can you talk a little bit about the acquisition from today? Like historically, you always said you wanted to do some ETL, but there's obviously quite a few ETL players in the market. How do you see that evolving between what do you want to do? What do the other players want to do? And what does it bring to Snowflake? And then I have a follow-up for Mike.

A: I mean, first of all, these are all very, very, very large spaces the overall vision, especially with the [ERA] (ph) interoperable data that's upon us, is we think that there is a very large opportunity for Snowflake to help our customers act on all of their data, not just the gold data that they used to put into Snowflake for analytics. Anecdotal, but the kind of examples that I get from talking to our customers is they have hundreds, sometimes thousand times as much data sitting in cloud storage, as they will do in a structured data platform. And more and more, they also feel like it is important that they own that data. And so you're seeing a shift in which things like application data is getting de-siloed, deconstructed so that great new things that you can imagine, both for data transformation, data engineering, but also AI is going to happen. And this is the context in which something like a Datavolo is really important for us. It comes with over 100 different connectors out of the box. It is going to run as part of Snowflake. It can also be deployed in customer VPCs, which lets us bring data in -- from places where normally we would not be able to run Snowflake on. And so it's really a first multiplier for the data that our data engineering pipelines can take on, but our AI products can be built on. But as I said, this is a very, very large space, estimated just data engineering -- data products as a whole, we think will be on the order of several hundred billion dollars 10 years from now. And so there's going to be lots of company. Our value-add is this easy integrated, highly efficient platform that we can bring for our customers and things like Datavolo are an important piece here.