Transcript
a lot more money is going into those leading companies. The real question the founders have to ask themselves is, can a coding agent somewhat create what I'm creating here, or do I have the true moat and staying power because of my domain expertise, because of what I've built around the data assets themselves? And so I think that's what investors are gonna be looking for. That's what we're looking for. Hello everyone, and welcome back to Snowflake Ventures. I'm now seated with Janaye Turay, Research Lead at Crunchbase, along with Harsha Karpe, Head of Snowflake Ventures. Such a pleasure to sit down with both of you. Thank you again for making the time. Thanks for having us. Thanks, this is great. Really looking forward to diving in. Snowflake and Crunchbase recently released a report examining where capital is flowing across the AI data cloud ecosystem, and that what it says about the future of enterprise AI. Janaye, let's start with you. What stood out most in the data? Yeah, well, I think what was really interesting in this report is we matched the whole Snowflake partner ecosystem with Crunchbase data to see if we could find some interesting insights. And what we found is since 2020, around 113 billion has been invested in Snowflake partners that are venture-backed. So it was more than 1,300 companies, part of the ecosystem. And the peaks were 2021, which matches Crunchbase data. We saw a lot of funding flow in 2021. But what's also interesting is that we saw that pick up again in 2025. So around 25 billion invested, over 25 billion, in the Snowflake partner network in 2025. In 2021, it was 34 billion, so higher. That was the market peak. But ventures picking up again. What was different is it's much more concentrated in 2025. So fewer companies, but larger checks in the partner network. Great perspective. Thank you, Janaye. Now, Harsha, the report shows capital increasingly flowing toward AI-ready data infrastructure, governance, and agentic applications. What's driving investor conviction in those areas specifically? Yeah, and I think that you kind of covered the spectrum of it, right? I think initially when the AI boom happened, generative AI, you could have said, hey, the agentic solutions, we're getting a lot of that concentrated capital. I think that was the case. And then I think what people started to realize is there was a lot of prototypes, a lot of experimentation. But to really get those into production, you needed the proper infrastructure, you needed the proper governance. And so I think that three-legged stool, if you want to call it, is what has really come together to say these are now the type of companies that have staying power. So I think all the VCs are now looking beyond, do you have the agentic capabilities and are you solving a problem? But how are you doing it? Do you have the governance considered properly? Do you have the infrastructure hammered out? Where is this going to be deployed? How secure is it? And that's obviously where Snowflake plays, right? Governance and infrastructure and security components of it. So that's why I think we're starting to now see where the companies that have true staying power really understand and cover all three of those. Great perspective. Thank you. And Janaye, back to you. The report highlights a major shift from broad market exuberance in 2021 to a much more concentrated investment environment today. What does that evolution tell us about the AI market and how it's maturing? Yeah, I think what we're seeing in our data is there are companies who are breaking out. And so there was a lot of capital, not only into the foundation models, but any company that's breaking out due to AI, some of these AI native companies are seeing a lot of capital flow to them. We're also seeing if a company has a pedigreed founder, there's a lot of interest. And then there's markets, which are just very hot. And if there's a few leaders, others are saying, we also want to find companies in the sector. We think this is growing. So there were all these growing sectors. But what we see with investors is when there is conviction, a lot more money is going into those leading companies. I think broadly though in the market, in 2021, it was very software driven. What we're seeing that's very different today is there's a much broader array. So there's energy, there's physical AI, there's infrastructure, there's semiconductors, there's robotics, and manufacturing, as well as all the software. So I think the sectors has massively broadened. And I think that's a much healthier ecosystem than what we saw in 2021. Great perspective. Thank you for that. And Harsha, what signals tell you a startup is truly ready for enterprise AI adoption, not just say experimentation? Well, we kind of have the benefit of being a little bit up the stack when we look at investing in that we're looking at companies that already have some signals of traction. So our investments tend to be a little bit more in the series A and later. And so we had the benefit of, these are companies that already have some customers, they already have some revenue. And of course, we want to see joint customers with Snowflake. But those are the signals we actually look for because I think trying to make bets on, is this the next thing that's going to solve our problem? Like without the customer traction, you're back in that paradigm of experimentation, right? And so having even a few customers who have validated and you can talk to them, get some perspective on how this is working, what they would like to see beyond what there is, that then gives us the signals to say, okay, let's work with that company to say, how do we address those gaps that have come up? But it really is that simple for us. It's actual market traction, actual customer traction, I should say, with real live customers and not just POCs and pilots. It's wonderful though, very much in it together, it sounds like, Harsha. Now taking a step back and very much looking at the future and what's next, as we look toward 2026, how are both of you thinking about this next phase of capital formation in the agentic AI economy and what will investors and enterprise buyers be rewarding most? Harsha, why don't we start with you, maybe start with that investor and enterprise side. Yeah, I think it goes back to having some true signals of adoption where these things are going into production, but I think there are some signals also of the way these companies are architected, right? Have they considered that, what you talked about in the first question, have they considered governance? What's the infrastructure layer? How are they doing the agentic piece? Are they AI enabled solutions in the right way? I think the breadth and scale of what they're trying to solve, like it needs to be those more complex problems, right? We can throw a workflow at a very simple problem, but coding agents can also do that, right? So that's the real question the founders have to ask themselves is, can a coding agent somewhat create what I'm creating here? Or do I have the true moat and staying power because of my domain expertise, because of what I've built around the data assets themselves? And so I think that's what investors are gonna be looking for. That's what we're looking for is the true moats that are being defined by, I have something that can't be disrupted by a build scenario at an enterprise customer because I'm either bringing proprietary data that expands and compounds the value over time, or I've really got that domain expertise and I know this very complex problem that a simple workflow solution is not gonna be able to solve. And so I think you're gonna start to see a little more discipline around that in the VC environment and definitely even from us. Great breakdown. And Janaye, I'd love your perspective from the broader market data. Yeah. Well, I think what's interesting and it's just to ground the discussion, Q1 based on crunch-based data global venture was 300 billion. That is the largest quarter ever that we've seen in venture. And 2021 was over 700 billion in a year. So we're already almost half of 2021. And everyone said in 2021, there was over investment. Again, I think it was, there was no new technology. In 2025 and 2026, there was a whole AI revolution. So I think that this feels, even though it's much bigger, it's a lot more grounded. We saw 122 billion go into open AI. But what that tells me, and I think the report also shows that, is a lot of the partners in the Snowflake ecosystem, around 44%, we predicted would be fundraising in the next period. And then we also looked at how many companies would exit. And that number is pretty high. It was 34%. And some of that is IPOs. It's a smaller proportion, but a lot of it is M&A. And so what we're seeing is there was a lot more coming, money coming in, but there was a lot more M&A activity we saw in 2025 and 2026. I think we're seeing Snowflake acquire. And so the market is really heating up because there's so much opportunity and we're only three years in. Very well said. Thank you for that. Finally, this report is a great example of Snowflake and Crunchbase very much combining ecosystem insights with market intelligence. What conversations do you hope this research sparks across the industry? Sinead, let's start with you. Yeah, well, I think for Crunchbase, we wanna be this underlying private market data. And AI has really opened up our business because we used to bring in a lot of data from many, many different sources through processes. Now we're bringing in a lot of data through AI. So there's a lot more product, there's a lot more up-to-date information. And then on top of that, we're building insights. So we're able to do these predictions about companies that we think are likely to raise or companies likely to go public. And then just this week, as part of Snowflake, we announced micro industries. So you can look at companies from a much more granular perspective. This is sort of AI generated off of products and services. So I think a much more interesting way to dive in. I've been playing around with it and look at autonomous trucking and seeing what comes up or in AI inference. And so there are all these micro industries about 5,000 where you can investigate. If you think something's interesting, you can do some deeper research. So I think this AI moment for us as a product, as a data product, it's a lot more exciting. And it's great to be here with Snowflake to kind of celebrate and build that data. Wonderful to hear. And Harsha, from your seat. Yeah, I think for us, selfishly, I'm looking at this to be something that the founders or the next generation of founders are picking up when they think about what they should be trying to solve and how they should be trying to solve it. You and I talked about this a little bit earlier about I can do so much with AI today that I can go and build things that I couldn't build before. I can build something that turns on a light in China. But why? That question of why am I building it? What am I trying to solve? And I hope what this kind of spurs is that relentless focus on customers. When you think about starting your company and you think about how you're gonna go to market, is it a complex problem that really has that staying power, has that mode? Just getting that thought process going in that level so that when we're coming into the investment discussions, we've got the founders that are like, yeah, I've been thinking about this. I understand that I need to have the governance thought about the infrastructure, the agentic solutions, but I also need to understand what problem I'm solving. And by the way, I'm an expert in this. I am bringing something to the table that is gonna really define that moat for you. And obviously for us, do that with Snowflake. It's wonderful to hear. Such a pleasure sitting down with both of you. Thank you again. Thanks, Ryan. And for the audience watching, thank you so much for tuning in. We'll catch you next time.
