Pod Insights Equity Podcast Snowflake’s transition from storing data to shipping with it About this episode Snowflake is betting that the future of AI isn’t just analyzing data, it’s acting on it. That means a shift away from chatbots and toward autonomous agents that can actually get work done. And Snowflake is reorganizing fast to keep up, from shipping hundreds of AI features to restructuring teams along the way. On this episode of TechCrunch’s Equity podcast, Rebecca Bellan sits down with Snowflake CEO Sridhar Ramaswamy to unpack the company’s transformation and what it signals about where AI is headed next.
Listen to the full episode to hear:
Why Ramaswamy believes the chatbot era is ending and the agentic era is beginning.
How Snowflake is evolving from a data warehouse into an AI and applications platform.
What “shipping with your data” actually looks like in practice.
Why the company is making big internal changes to support its AI push.
Subscribe to Equity on YouTube, Apple Podcasts, Overcast, Spotify and all the casts. You also can follow Equity on X and Threads, at @EquityPod.
Chapters:
00:00 Intro
00:17 Snowflake’s AI shift and agentic future
01:45 Why 2026 marks the end of chatbots
04:09 Cortex Code, Snowflake Intelligence, and new products
06:09 Who benefits: non-technical users & enterprises
07:35 Adoption challenges and why AI pilots fail
12:11 How AI is reshaping jobs and skills
14:39 Layoffs, automation, and the future of documentation
18:37 Snowflake’s evolution into an AI platform
21:04 Competition: Databricks, hyperscalers, and AI giants
25:01 Outro
Learn more about your ad choices. Visit megaphone.fm/adchoices
Transcript 0:00 Speaker 00 This episode of Equity is brought to you by Schneider Electric. Schneider Electric, advancing energy tech.
0:05 Speaker 01 Hello and welcome back to Equity, TechCrunch's flagship podcast about the business of startups. I'm Rebecca Ballon and this is the episode where we bring on industry experts to help us explore a trend in the tech world and dive deep.
0:23 Speaker 01 Snowflake has always been known as the data warehouse, but today owning the data is not enough. You have to ship with it too. And Snowflake understands that. The company has undergone an AI-fueled identity shift in recent months, including laying off entire teams, launching agentic products and dropping more than 400 AI features in the last year. Their CEO, Sridhar Ramaswamy, has pretty much declared 2026 to be the end of the chatbot era and the beginning of the agentic era. And he's here today to walk us through his vision and the recent changes at Snowflake. Sridhar, welcome to the show.
0:56 Speaker 02 Rebecca, great to be here. Thank you for having me.
0:58 Speaker 01 Yeah, so you've had such an interesting background. You co-founded Neva, that was an ad-free, privacy-focused search engine. And then Snowflake actually acquired Neva, what, in 2023? And later you came on as the CEO. You've had quite a wild ride. Where were you at before that? I think Google, right?
1:16 Speaker 02 Well, I'm an academic by training, worked in the search labs in places like The Labs for a while before going to Google. Little did I know that I was joining one of the most amazing businesses of all time, which was the search ads business at Google. I spent over 15 years there, eventually running the Ads and Commerce team at Google, helping them get to $100-plus billion in revenue, which was not something, honestly, that any of us expected to do. And then, of course, Neva and Snowflake after that. I'm blessed to have had so many different opportunities.
1:51 Speaker 01 Yeah, I mean, that's quite a flex, from $1.5 billion to $100 billion in revenue in the 15 years that you were there, right? So, you know, you've said that 2026 is the year that organizations will move beyond chatbots and coding assistants. Can you talk a little bit about what you mean and how Snowflake is embodying that?
2:10 Speaker 02 Yeah, for most of us, AI has meant something like ChatGPT. You could ask good questions, sometimes be right, sometimes be amazing, sometimes be creative, all kind of hit and miss. Stuff that doesn't quite work in the context of an enterprise. But if you have questions about revenue, it better be super, super accurate. And we could tell even last year that coding agents were going to be a big deal. In terms of the speed at which people that only modestly knew coding, say me, could actually get stuff done. What we're seeing this year, though, is these techniques coming together. Agentic AI being used more and more to create complicated applications.
2:54 Speaker 02 For example, our support team actually uses a coding agent with aspects of chat, aspects of tools to pretty much drive how they solve problems. Similarly, enterprises use products like Snowflake Intelligence to ask pretty sophisticated questions about how their business is doing. And built on a solid data foundation, these Agentic AI products are becoming incredibly useful. Just this morning, I wanted to know how our quarter was doing. But then I wanted to know how our different components, our theaters as they were, the US, EMEA, APJ, were contributing to that overall status. Previously, it used to be an analysis that someone had to run for me.
3:36 Speaker 02 This in natural language, I could not only ask the question, but produce something that looks like a dashboard that could be shared internally. That's the sophistication, which is take on governed data sources, things that are secure and auditable and trusted, which Snowflake has always pioneered. Put this layer of Agentic AI on top of it. It just makes that data so much more usable. And obviously, next up is what kind of actions can you directly take from these Agentic applications that can now take actions in the real world? That's where we see that inexorable push towards what is AI doing in the enterprise?
4:15 Speaker 01 Okay, and then so you launched Project Snowwork in March, right? And Cortex Code as well is a newer product? Can you just walk us like briefly through those?
4:24 Speaker 02 Cortex Code was born of the belief that Agentic AI was going to make it much easier to use on products like Snowflake. A lot of using Snowflake is setting up configuration, connecting to other systems, making sure that data is in good shape, writing code. It's a pretty complex job. It's often tedious, easy to make colossal mistakes. And so making sure that you do it in a safe way is always a concern. Codex Code was born out of the belief that Agent AI had the power to automate much of that, putting the humans in charge of subjective questions, questions of taste and quality, rather than the mechanics of getting work done. It's had a huge impact. Whether it's customers like Boop that were early adopters
5:09 Speaker 02 or partners like Evolve that used it to leverage how quickly they could get work done with Snowflake. Codex Code has had a huge impact in just speed of adoption of various features that are in Snowflake. Now, here's where something funny happened at Snowflake. We made Codex Code available to everyone at Snowflake, not just engineers, everyone. And what we saw was non-technical users. My sales executives started writing applications, started using Agent AI to ask questions. That's when it clicked in my head and in a bunch of other people's heads in Snowflake. A coding agent is not just about coding. It's an abstraction agent. That's really what Snowvergus is about. Use the power of Agent AI to make data, to make actions much simpler,
5:54 Speaker 02 so that a CFO can look at travel and expense data within Snowflake, ask questions in natural language, take follow-up actions also in natural language, all within the safe, governed, auditable environment that Snowflake is. So it's actually a closer progression than you would initially have thought.
6:13 Speaker 01 Interesting. Okay, yeah, I mean, that answers one of my questions, which is who is this for? Is this for the data engineer, the business analyst? You're saying that a lot of non-technical people-
6:20 Speaker 02 This is actually for business users. Yeah, that's the magic. It's my HR team that's writing and saying, oh, gosh, processing resumes now is so much easier. Being able to figure out who is doing a great job hiring is so much easier. Obviously, the finance team is using it to get better insight into the data. That spectrum has basically become everyone within an enterprise.
6:42 Speaker 01 Okay, well, if you know, non-technical users are the ones that are getting the most benefit out of this, there's always that kind of black box effect, right? Like, who is accountable when the agent is wrong?
6:53 Speaker 02 This is a great question. We start on a solid governance foundation. Whenever we create data agents, for example, we help people go through an eval process. That's just a fancy way of saying you're going to ask a bunch of questions, make sure that the answers are right. And every time you make a change, you want to make sure that you didn't break things that were working before. It's what you would think of as standard software engineering. And so whenever capabilities in something like a Snowwork are released, we have them go through the same kind of evaluation process where you ask a series of questions. You make sure that the answers that the coding agent is generating are indeed correct. This is a well-understood science. It just has to be applied diligently
7:37 Speaker 02 every time you add new features to something like a Snowwork.
7:40 Speaker 01 So for your customers or for your customers' customers, what level of autonomy are they actually comfortable with today? Is a lot of this happening in pilot stage? Or is this happening at scale yet?
7:52 Speaker 02 This is rapidly evolving. And I think it depends on the progression of capabilities. More and more folks now are comfortable with the idea of a data agent, for example, built in Snowflake Intelligence, asking and answering business questions. We have customers like Poop that I talked about earlier, or TS Imagine, that have taken the data that they have in Snowflake and have created some pretty amazing experiences for areas like support. And things like autonomous agents, having agents take action, are slow progressions. It's not, you know, you just decide that you're going to turn everything over. I'll give you our own example.
8:32 Speaker 02 Our SRE team, their site reliability engineers that help us keep Snowflake up, they started using Cortex Code to help them debug problems faster, to get to the bottom of why is there a customer issue, let's say TechCrunch is experiencing with Snowflake. And what they found was that their time to problem detection was down by orders of magnitude. And now they're driving an autonomous version where they're looking at a large volume of incoming requests and saying which are problems that they can reliably answer. And when you can do that, you can also take actions a little bit more autonomously. And you also want to make sure that these are not like big actions,
9:15 Speaker 02 irreversible actions, things like that. So there is very much that element of figuring out how good is the product? And how believable are the actions? And then you put them in front of a human, make sure that it's working really well. And when you're very comfortable with how right it is pretty much all the time, that's when you say that category of problems you're dealing with can be dealt with autonomously by an agent. It is very much a science. It's not yesterday, no autonomous, and today, everything is autonomous.
9:44 Speaker 01 Yeah, so, so certain things you'll see that are maybe like more rote, they're easier to automate, and those are the ones that will start to get automated first. And mostly, if it's not in those buckets, these are still pilots where we're trying and failing. Like, what, what percentage of AI pilots actually make it to production today?
10:07 Speaker 02 It depends on the product, it depends on the area. With Snowflake Intelligence, for example, once we have one, what we call a use case, which is a pitch for a customer to implement something, most of them make it out to production. But the time that they take to make it out to production can vary anywhere from four weeks to six months, depending on the complexity of the problem. Increasingly, we try to get our customers to try small things and launch them before trying to do bigger things.
10:33 Speaker 01 What's like a number one reason that they might fail, right? Obviously, there's a lot of differentiation and, and you know, what products will, will make it to market. But what causes agentic pilots to fail?
10:44 Speaker 02 The number one reason is going to be, I don't know how to ask questions and get answers reliably from a product like this. And I experienced this firsthand, even within Snowflake. We have access to a sales agent that's been rolled out to 5,000 of our go-to-market team, used every single day. Now, Snowflake is 10,000 people, 5,000 people is a lot. You can agree that that is scaled usage. But on the other hand, we're rolling out a new agent, for example. With all of our financial data, I was telling someone yesterday that I still don't have a good feel for how to query it, how to talk to it effectively, because I don't quite understand the data model effectively, you know? There's a lot of complexity in finance data.
11:28 Speaker 02 Gap revenue is different from raw consumption data, and so you have to know a lot of nuance about how that works before you can feel confident about it. That is an example of something that has to be iterated upon more before it's released to all of the execs at Snowflake itself, for example. And so it really depends on how do you create products that your customers, your users will relate to and reliably get value from it?
11:52 Speaker 01 Yeah, and I guess that bottleneck of not understanding how to query something properly, that means that when you do query it, you're not getting the results that you want, and you're less likely to use the tool, right? And how much, I wonder, not even just at Snowflake, but in the larger environment, are you seeing that employees are kind of reluctant to try to learn this new skill on top of all of their existing work?
12:16 Speaker 02 It's a good question. It's also a layered question.
12:20 Speaker 02 There are many disciplines, especially in software engineering, or with our technical sales, for example, where we're way past the proof points of is this useful at work? A solution engineer working at Snowflake, you know, you can think of it as pre-sales showing what Snowflake is capable of, can create a new demo or a proof of concept in one or two hours, compared to taking a cookie cutter demo or taking two to three weeks to build a new demo. There's just no question about whether AI is useful or not in those cases. Similarly, you know, if you run into software engineers that tell you that they're deciding whether to use AI,
13:05 Speaker 02 I predict that that's an unemployed software engineer not too far from now.
13:09 Speaker 01 Yeah, yeah, that's something we're hearing a lot.
13:11 Speaker 02 It's one thing to say you can't trust AI with everything, but there are disciplines where not using AI right now is not an option. And I have sons that are software engineers. They're in their mid twenties. They use AI natively in what they do. They don't have a choice. They'll not be employed if they don't do this. And so I would say that that's what I mean by it's a nuanced answer. AI is not going to do everything, but there are many places where AI is pretty foundational to how people need to work.
13:37 Speaker 00 Every company today is looking for an edge. Better insights, faster decisions, greater resilience. But as AI drives new growth, it also creates new demands on energy and infrastructure. Schneider Electric is connecting energy, automation, and digital technologies to help organizations make smarter decisions and operate more efficiently. Because the future won't be shaped by AI alone, it will be shaped by energy and industrial intelligence. Schneider Electric, advancing energy tech.
14:03 Speaker 01 Zooming out, you've got kids that are studying software engineering. What do you think? What software categories are you telling them, "Don't go into this. This is going to die"?
14:11 Speaker 02 It's just hard to predict because model capabilities are growing literally by the month.
14:20 Speaker 01 Right.
14:20 Speaker 02 I think the foundational aspects of software engineering, what constitutes a good system to build, and how do you structure your codebase, develop your systems in a way that's long-term sustainable, that still requires a lot of taste. I think things like that are always going to be needed. On the other hand, if it's a matter of generating webpages, let's say, or documentation, it is very hard to see how that survives as an independent, sustaining profession for the next many years. Things that are pretty easy for agentic platforms to do, generating documentation or reading
15:05 Speaker 02 documentation to write little code snippets, those are all things that are largely automatable at this point.
15:12 Speaker 01 Okay. Interesting that you say that because you recently, Snowflake in March, laid off quite a lot of workers and you made some targeted cuts and you cut your entire writing team, right? So I'm kind of curious about that because if you are pushing for agentic first, do they not need access to docs that are in working order to read them? If agents are relying on documentation, why are you cutting the people who create it? Or are you just betting that AI can generate and maintain docs better than humans?
15:39 Speaker 02 In all situations, whenever we make changes to jobs, we try to make sure that folks have other opportunities within Snupfling. That's always the aspiration with which we go into areas. And things like documentation have become much more of an interactive exercise, where the continuum between when a product requirement document is written by a product manager that says, "This is what we should build," to when an engineer says, "This is how you build it," to someone saying, "We need to figure out how to use this feature that has been built." That's now all done under the purview of coding agents. The way I learn, for example, a new piece of code
16:23 Speaker 02 is to literally point a coding agent at a codebase and say, "How do I use..." That session, some ways, becomes like the first version of potential new documentation that you can write. And in fact, that's a very common archetype for how do you abstract in this new world of AI and coding agents, which is you go do something by hand, and then you ask the coding agent, "If I needed to help somebody do this automatically, what would I do?" It writes this thing called a skill that then asks you for input, and my experience becomes a packaged skill for you to try. And this process is very recursive, because you build better documentation, better skills, on the basis of trying and the existing skill that there is.
17:08 Speaker 01 So then you did, in cutting that team, you believe that you're able to fully automate the tasks that were being done by them?
17:17 Speaker 02 Some tasks can be automated. Some folks from that team went on to join a new group that's going to be figuring out what are the tools that we need now in order to be able to generate documentation. Effectively, how do you have the same guidelines for all of the documentation that is written? You don't want one page looking dramatically different from another page. And so the work shifts over to these kinds of pan documentation issues. And that's a common pattern that you see, which, as I said, the specific work that you do can be generalized. Into a skill that is repeatable and a set of skills that help you get some things done can now be harmonized by putting another layer off. And this is how we should structure how people get things done.
18:01 Speaker 02 The work is shifting over to a different place. People that used to write SQL queries, even within my own team, they write skills that make it easy to create agents that now write C. If work is going up the level of abstraction.
18:15 Speaker 01 Yeah, okay. So then what other roles are disappearing or expanding, right? Like what skills are you hiring for today that you weren't two years ago?
18:22 Speaker 02 Roles are changing all the time. The kind of places where we continue to hire a lot include areas like sales, any place where there is communication with external entities, things like that. Those are jobs where we continue to hire, um, hire a lot. We continue to hire lots of software engineers, especially ones that are AI native. Because we find that being born into this world just helps you be a lot more effective in today's world. And we are hiring into areas like our developer relations team because they help effectively educate a very large group of developers outside on how to better use Cortex Core, how to better use these higher level abstractions.
19:04 Speaker 02 These roles are changing and evolving and they will continue to be areas that we will invest in.
19:09 Speaker 01 Now layoffs are not easy, but we're seeing it across the industry. And we're seeing a bigger push towards, you know, obviously embracing more AI. That's just the way, the world that we live in today. Snowflake has had its own shift, right? You're trying to avoid this, what you call the dumb data pipeline, right? It feels like you're in a mid-identity shift from this data warehouse to an AI platform. Can you talk a little bit more about what your thinking is, your strategy is going forward?
19:38 Speaker 02 Dear point. How data gets consumed, how work gets done is changing pretty rapidly. Yes, Snowflake was born as a cloud data warehouse platform. That was who we were 10 years ago. We added on other things like Collage. Data has always been an incredible ally to making businesses just run efficiently. You know which part of your advertising is working? Or you know where you're spending money if you're closing your books on Snowflake and it helps you be more efficient. I think what AI is doing is it is unlocking the power of data in ways that were unimaginable before.
20:20 Speaker 02 Previously, if you wanted to access data, you needed somebody to build a dashboard or write a SQL query for you. Agentic systems like Snowflake Intelligence democratize data access. They put data at the hands of people that need to look at the detail to figure out what to do. I think what Agentic AI is going to do is give these data platforms the ability, say, to call into other systems using technical protocols like MCP. So I predict that in the near future, you will not really be able to tell between what's called an analytic system that looks at the past to tell you what's going on and more of an operational system where you take action on something that you want to do right now,
21:03 Speaker 02 like send an email or update a sales use case. That's where we see the world heading. We see the world that is going to be full of agents with these kinds of capabilities. Some of them are going to be pure data agents. Others are probably going to look more like the applications that you and I have been using for 10 plus years. But we want to be there, both in providing the data platform underneath that's going to power these new systems, but also be there with instances of these systems that end users can use. You see this with Snowflake Intelligence and you see this with Snowwork.
21:36 Speaker 01 Right, so it's not just the data as a moat. There's also the implementation of that data as the moat. And I'm curious, is that enough for differentiation when you've got Databricks that is kind of eating into Snowflake's turf? They recently raised a massive valuation, $4 billion at a $134 billion valuation in December. So what's the real difference beyond we have better data governance? I'm curious if you've started to embrace unstructured data. Snowflake has always been the structured data area, but most enterprises, their data is kind of a mess.
22:13 Speaker 02 First of all, the biggest answer is innovation. Transformative things are possible with technology today. And looking forward to where you want to go is really the path for success. Snowflake Intelligence was a product like that. It looked forward to what Agent AI was going to do before Agent AI was a thing. We announced that in November. Similarly, we could see the surging power of coding agents. Middle of last year, which is when Cortex Code was born, it's going to be one of our fastest growing products, actually both of these, both Snowflake Intelligence and Cortex Code.
22:51 Speaker 02 The way we differentiate is by looking into this future and creating products that our customers want. I won't comment about private company valuations. It's sort of really hard to tell how much of that revenue is durable enterprise revenue. We are a public company, we report stats. Our RPO, future commitment our customers have made, is $10 billion, $9.8 billion. The reported numbers from Databricks are very fuzzy, it's like a fraction of that. So I'm very proud of the business that we are building and the durability of that business. We had a customer that signed up a 400-plus million-dollar contract with us.
23:34 Speaker 02 That tells you how much trust they have in our future. Areas like unstructured data, 100%. We have things like AI SQL, Cortex Search, which help people index and make sense of this data. It is going to be a key component of AI. We feel good about what we're doing. Do I continue to push myself, the company and the teams very hard in terms of being able to do better? Absolutely. You're going to see a slew of MCP connectors, for example, coming out that support unstructured data much, much better. That's the speed at which innovation needs to happen.
24:08 Speaker 01 Well, while we're talking about competition, I did want to just also ask, you know, at what point does the neutral platform become a competitor to your own customers' AI stacks? Like, are you worried about AWS, Google, Azure, you know, eating your lunch?
24:25 Speaker 02 There's a ton of competition and innovating, creating value for customers is absolutely the way to go. I would, by the way, add the big model creators to that list. Anthropic and OpenAI want to be the front door to every enterprise. They clearly make among the most impressive models in the world. So there is no competition. But on the other hand, we collaborate actively with all of these folks. We have very large commitments to spend, multi-billion dollars in the case of Azure and AWS. Slightly more modest, but still nothing to sneeze at when it comes to Anthropic and OpenAI. And what drives all of this is the growth of our business.
25:09 Speaker 02 I think this is a time of unprecedented opportunity. If anything, I'm focused on how do we accelerate and go even faster. Yes, we will absolutely compete with the hyperscalers on a number of fronts, but we also collaborate with them very effectively with many, many customers. And that's a little bit of what partnerships are all about, which is you figure out the areas where you're strictly better for your customers and lean together and work. And all of these folks are very mature in how they do this.
25:34 Speaker 01 Awesome. Well, that's a great place to end it. Sridhar, what, where can our listeners find you online?
25:40 Speaker 02 I'm on X, I'm on LinkedIn. I always read and listen to all of the comments that come in there. And I'm at amazing podcasts like this one every now and then.
25:52 Speaker 01 Okay, great. Well, thank you so much for joining us. It was really a pleasure to have you. Please keep us posted on any news that's coming up for Snowflake. And to our listeners, you can find me on LinkedIn, Blue Sky X, and you can find Equity at Equity Pod on Blue Sky and X. Talk to you next time.
Previous Space: the final frontier of AI infrastructure Next Luma AI's Amit Jain on why most world model companies are getting it completely wrong Page updated October 12, 2026 · transcript and summary are machine-generated