Speaker 1:

From the New York Stock Exchange at the corner of Wall and Broad Streets in New York City. Welcome inside the Ice House. Our podcast from Intercontinental Exchange is your go-to for the latest on markets, leadership, vision, and business. For over 230 years, the NYSE has been the beating heart of global growth. Each week we bring you inspiring stories of innovators, job creators, and the movers and shakers of capitalism here at the NYSE and ISIS exchanges around the world. Now let's go inside the Ice House. Here's your host Lance Glinn.

Lance Glinn:

Welcome into another episode of the Inside the Ice House Podcast back in January on episode 453, we were joined by Sridhar Ramaswamy, CEO of Snowflake. That's NYC ticker symbol SNOW for a wide-ranging conversation on partnerships, the acceleration of AI and Snowflake's ongoing evolution. Sridhar spotlighted the company's powerful capabilities and its mission to simplify complexity, empowering organizations to unlock the full potential of their data. Now as 2025 winds down and we're closer to the end of the year than the beginning, we're thrilled to once again shine a spotlight on Snowflake at the New York Stock Exchange. This time we're joined by Christian Kleinerman, executive vice president of product for a fresh perspective on innovation, strategy and what's next for one of the most dynamic companies in tech. Christian, thanks so much for joining us inside the Ice House.

Christian Kleinerman:

Hey Lance, thank you for having me. Excited.

Lance Glinn:

So I want to begin our conversation with Snowflake's launch of Cortex AI for financial services designed to address some of the sector's most notable challenges, one of the goals to bring AI directly to financial data that already exists within Snowflake to help accelerate tasks. How did you and your team approach just designing AI tools that could work natively with such complex data sets and go about building this new product?

Christian Kleinerman:

Great question and great timing. Since we just announced many enhancements for Cortex AI, specifically on financial services. The thesis of Snowflake all along has been about helping organizations get value out of their data. Of course, that has to be governed with security and privacy at the forefront. And we live in times where AI is disrupting almost all activities around us. And the most interesting thing is that in the enterprise, AI is powered by data. At the end of the day, you can get lots of results from a generic LLM model, asking questions, getting responses, but if you want to put it in the context of an enterprise, it all comes down to data.

And companies wanted to do it with that security and with that privacy in mind. So what we set out to do was to say, how do we bring AI into companies' secure perimeter, the area where they have already entrusted Snowflake with their data? And that's where Cortex AI comes in. We bring state-of-the-art models leading from Anthropic, from OpenAI, coming into Snowflake, and then we realized that there's a need for not only enabling first-party data, which is what a financial service organization can have, but also third-party data.

And one of the things that we're very excited is, we have a number of third-party data providers that are bringing in augmenting data so that they can go and leverage the power of AI within the context of financial services. So we looked at a number of customers, we have lots of great partners, companies like BlackRock have been leveraging Cortex AI to help their internal employees be more productive, provide better services to their users, and that's how it came together. And we're very excited about Cortex AI for financial services.

Lance Glinn:

So a key part of this is Snowflake's new managed model context protocol server, allowing institutions to connect data, connect their data to external AI agents and applications. Can you just unpack what that really means from a technical and strategic perspective and the problems you're solving when deciding to build this kind of infrastructure?

Christian Kleinerman:

Yeah, so it's super important the notion of eliminating silos, that's something that we have been concerned about for a long time, which is, you don't want to have your HR data separate from your financial data, separate from your maybe product telemetry or quality data. And as part of the broader goal to make sure that data is accessible, I'll give credit to Anthropic who introduced and open sourced, this Model Context Protocol, which enables AI services, in particular AI agents to be able to reach out to third-party tools and third-party data and put in context with an AI activity. So now with the Model Context Protocol server that we have introduced for Snowflake, we help our customers take the data that they already have in Snowflake and make it accessible to AI tools, AI agents. So now I can go and ask a question, ask for a task, and if the task, the agent may benefit from querying the HR data or maybe the data from a SaaS app like Salesforce, it can go faulted in through Model Context Protocol service. And at the end of the day, it just helps eliminate silos for data.

Lance Glinn:

So Snowflake has always been about enabling ecosystems and this launch seems to really deepen that commitment obviously for financial services. How do you see Snowflake's role evolving as AI becomes just more embedded within that sector?

Christian Kleinerman:

Yeah, I love that the premise of the question is something that we care a lot about, which is how do we work with a number of partners and build that rich ecosystem? At the end of the day, what we're building with Snowflake, we call it the AI data cloud, which is a great technology platform, but an amazing ecosystem of partners. And with MCP, now we can expand that set of partners to additional organizations, whether it's a data provider or a language model provider, and bring those solutions to our customers.

Lance Glinn:

So earlier this year on the Inside the Icehouse podcast, Snowflake, CEO, Sridhar Ramaswamy joined us. Now during our conversation, he shared a powerful insight. I'm going to quote him here, "AI is great, but if you need to be deeply experienced in AI to get any value from it, that's going to limit its application. We love taking complicated technologies and making it easy and safe for people to use." From your vantage point, Christian, how does that philosophy shape the way Snowflake builds its AI products, obviously just like the one we spoke to?

Christian Kleinerman:

I would actually say, that philosophy is the ethos, for lack of a better word, of all of Snowflake. If you look at how we started with simple primitives like compute. If you look at maybe the cloud providers, they have hundreds of different instance types. An when we first introduced Snowflake, which is 10 years ago that it went in general availability, we said, we don't have all these instance types. We have small, medium, large, extra large, to extra large. And that same philosophy that we did for compute, we brought it for many other things. We brought it for administration, we brought it for upgrades, and that same philosophy has carried forward to AI. When a customer of Snowflake wants to leverage AI in the context of the data, they don't need to think about where is the model hosted and did they get GPUs and did they have the right security controls in place? All of that is just faulted, and we just let them ask questions of the data with AI so that philosophy applies to AI, but most important to everything we've done at Snowflake since day one.

Lance Glinn:

And one of the things Sridhar emphasized during our conversation was making obviously AI safe to use. What does safety mean in the context of Snowflake's AI strategy? Is it just about data security or does it also include things like transparency, governance and obviously user control too?

Christian Kleinerman:

I would say it's three parts. One is correctness of answers, which matters a lot. Two is enforcement of security, which is important, and the other one is the ability to do evaluation and monitoring of the solution. It's in the same order. The first one, you can say it's the obvious one, but that was one of the biggest problems with AI a year or two years ago, which is the answer always sounds very high confidence, you have no idea if it's correct or not.

Lance Glinn:

Yeah, but you need to be a hundred percent. Exactly.

Christian Kleinerman:

So we've done a lot of work to have the AI, itself, do checks on the questions and the answers. We have introduced the concept of verified answers. So the AI knows that these answers are human-created and validated, and that increases the confidence of users with responses they get. Security, we already talked a lot of, we honor all the security configurations that customers have. And the last one is evaluation and monitoring. You want to be able to see what types of questions your users are getting. You want to be able to get feedback loops that help you improve the solution. We have all of these integrated into Cortex AI.

Lance Glinn:

So from the question about safety, I want to just spend a couple minutes real quick on the role of governance in AI. And there's this debate I think of how governance impacts the technology, whether it acts as a deterrent and sort of slows it down, or it can be a catalyst for acceleration, right? There's sort of that debate. If done right, how can governance help the latter, right? Help move that technology forward rather than the former and really burden its development.

Christian Kleinerman:

At the end of the day, governance in general is an accelerant because when any technology, and this is not just AI, it is true of data platforms and compute platforms, if there are no bounds and guarantees on what it can and cannot do, then organizations are largely going to lock down the technology or learn, usually the painful way, the consequences of the technology running amok in an organization. But once it's governed, once a central architecture department says, this solution meets our standards and we can deploy with this controls, now you can open up to everyone in the organization. That part of controls is very important part of Cortex AI. We do permissions at the model level, at the version of a model level, we can decide which members of our organization have access to a model and which do not. That gives organizations confidence that they can say, this solution is ready to be deployed at scale.

Lance Glinn:

So obviously we spoke earlier to the financial services sector, talking about Cortex AI in that context. But in industries like it, and you can also take an industry like healthcare for example, right? Governance, it's not optional. It's really foundational in these industries. How does Snowflake's platform help these types of organizations, the ones that are in these foundational governance sectors, operate while staying compliant and innovate while staying compliant?

Christian Kleinerman:

The history of Snowflake has been to provide controls and auditability capabilities so that all of those regulated industries can feel confident about how the data is being used. Things like lineage. You got some data. Where did that data get copied? Who else has access to it? And be able to trace all throughout the systems that applies to AI. With Snowflake can say, oh, this data ultimately ended up in this other location and this is what is powering Snowflake intelligence, which is our marquee experience for interacting with data via AI inside the enterprise. And if someone says, well, you got this answer, where did that question come from? Or where did that answer come from? And you can trace it all the way to the root cause. For example, if our CFO wanted to ask questions of what was our revenue yesterday, you better have an ability to go and trace it back. And that's a big part of the governance, applies to financial services, but it applies to every industry. There are other industries that may not be as regulated, but they benefit from the capabilities that we have.

Lance Glinn:

Sure. So Christian, before we get further into Snowflake AI and the rest of our conversation, I want to rewind a little bit and speak more about you and your career and obviously how you got to where you are right now at Snowflake. Just what first drew you into the world of data and technology and really sparked your interest in this path that, like I said, has now brought you to this really big NYC listed enterprise?

Christian Kleinerman:

That's an interesting question. I was building a startup many years ago, like '94, so almost 30 years ago.

Lance Glinn:

I don't want to age you, but a little before my time.

Christian Kleinerman:

A long, long time ago. And I was fascinated by SQL databases. The notion that you can just specify the clarity of what you want and the answer would show up. And you would add more data and the answer would still show up very fast. And I was completely mystified by it. And at some point I'm like, okay, I want to learn enough about this. I want to be part of it. That startup did not go well. Then I did another startup, did not go well, and I'm like, I want to go learn from people that really know how to go build software. Microsoft was probably the single biggest powerhouse at the time. I applied to join Microsoft. I got in, but they wanted me to go to the Internet Explorer team and I'm like, no, I want SQL Server because I wanted databases and the rest is history. I spent almost 14 years in various jobs in database and data management with SQL Server.

Lance Glinn:

So you joined Snowflake in 2018 after years of leadership experience at these other major tech companies. Was there something unique about the company's mission or culture, the vision of potentially Benoit and Terry that made you feel that this was the right place to build?

Christian Kleinerman:

Yes, 100%. I had talked with Snowflake in the early days. I think there were 30, 40 employees. They told me, we're going to do all of this. It's going to be amazing. And frankly, I was a little bit skeptical and I said, this may not be the time for me to join Snowflake, but maybe a year later they published this architecture paper at SIGMOD, a well-known forum for database innovation. And I saw that paper and it's like, oh my God, this is legitimately good. Someone once told me that the definition of true innovation is when you look at something, it's obvious that it's better, but nobody has done it before. And this thing definitely met that definition.

Lance Glinn:

No one had done it before.

Christian Kleinerman:

No one had done it. It made so much sense. And I remember also thinking, it's going to be very painful to compete with this thing. And then Bob, who was the CEO at the time, he calls me, this is now another year later and says, are you ready? And I'm like, I'm ready. I didn't even need any convincing because I could appreciate the depth of the innovation that Benoit and Terry had made.

Lance Glinn:

So Snowflake also, besides the launch of Cortex AI for financial services that we spoke to earlier in our conversation, Snowflake recently reached a pretty impressive milestone celebrating five years since its IPO. You've been part of the company's transformation since that IPO day, even a little bit before that too, playing a key role in its growth into a platform that is truly influencing how enterprises think about AI. What has allowed Snowflake, in your mind, from your perspective, to make the leap it has not just in scale, but really in strategic relevance and overall influence?

Christian Kleinerman:

I think at the center of Snowflake, there's a true customer focus. I've seen organizations that are building products based on their gut or they're building products based on the need for innovate, and we know better than customers and at some point they'll see the light. I don't think there's any of that. We have a clear thesis on how we want the world of data to evolve. We want this connected ecosystem of data participants, but beyond that long-term direction, we innovate with our customers. We think of many of our customers as partners.

I truly am on speed dial basis with dozens and dozens of customers, and that's how we know that what we're innovating and building in the product adds value to them. And they also, as customers, understand that we've got their back when... it's already [inaudible 00:16:44], when you're a smaller company, and I'm say compared to the Amazons, Microsofts, Googles of the world, you succeed only if your customer succeed. And that's the lens that we look at how we operate, and it has been true from where we're a few hundred employees and it's true at the scale that we operate today.

Lance Glinn:

So would you say you're not just working for the customers, but you're really working with the customers to a certain extent.

Christian Kleinerman:

100%. I do believe that it's a very close partnership, and in many instances we call ourselves partners of our customers.

Lance Glinn:

So as of Snowflake's fiscal second quarter, more than 6,100 customers are using its AI capabilities every week. A remarkable number. You joined the company, as I said, back in 2018, and a lot has changed obviously since then, not just in technology but in customer expectations and how products are built. So on the topic of those customer expectations, looking back over your tenure, the last seven years, how has your approach to product development and customer engagement evolved to meet the shifting demands of the user in this AI driven era?

Christian Kleinerman:

I feel that there are some elements that have not changed. The importance on security and governance.

Lance Glinn:

It still rings true 2018 to now, yeah.

Christian Kleinerman:

It's unwavered. And the few times that we've tried to move an ounce faster on those fronts, customers are like, slow down. We need to be careful. But there are other areas that definitely have forced us to change how we operate. AI and the speed of innovation of AI is the clearest example. We've traditionally had two large product launch events a year. So we do Summit, which is our conference in June, and we do our build conference usually in November. And for the most part, we could innovate, we could still give technology to our customers, but just make major announcements at those two points. In the age of AI, if you have something available in July and you're going to wait until November, you are there.

So it has changed not only the pace at which we share technology with our customers, but frankly the pace at which we innovate ourselves. The speed at which new models are being available, the speed at which standard, we talked about Model Context Protocol, there's Agent to Agent [inaudible 00:19:02] by Google, so the space is moving fast. So I would say that there's a much higher agility on everything we do. And I will give credit to Sridhar. He has also done his part in saying the company needs to be operating in this new clock speed that the world is at. So definitely that piece is different.

Lance Glinn:

You make a good point. You have something ready in July, can't hold it until November now, not anymore. So during our conversation with Sridhar in January of 2025, he spoke to the role of partnerships. Specifically during our conversation, he referenced collaborations with Microsoft and Anthropic. So how have just those partnerships as well as ones with, say, entities including OpenAI and Workday and so on and so forth, how have they enabled the company to serve customers in ways that may not have been possible before these collaborations?

Christian Kleinerman:

Yeah, so we are looking at many company engagements, partner engagements with the full strategic lens on how can we help each other, how can we create the proverbial one plus one equals three? Microsoft in particular, since you mentioned it, we've had a great relationship with Microsoft for many years, but it had traditionally been around the technologies that surround data. So we had the Azure ML team or maybe the Dataverse team, PowerBI to a degree, but we would've never thought of partnering with the core data team because we're competitors. And interestingly enough, in the last year, year and a half, we realized that we are frankly more aligned than not in many of the capabilities that we want to enable for our customers. We found common ground in the emergence of open file formats and open table formats. So we said, you know what? Why don't we hold hands and we bring Apache Iceberg implementations to market? Why don't we hold hands and make sure that data that is in Microsoft's OneLake is available to Snowflake?

Data that is in Snowflake is available to Microsoft Fabric. And we've done this, and I would say we have probably one of the deepest partnerships in the industry. And you could argue and say, but you guys are competing. And at the end of the day, what we're doing is, yes, we compete, but most importantly, we want to give customers the architecture and the choice that they need. With Anthropic and with OpenAI, we have very close partnerships on model distribution. Claude's latest model dropped yesterday. A couple of days ago. And we were a day zero launch partner for them. Same thing has been true with GPT-5 and the latest models from OpenAI.

Lance Glinn:

So Snowflake, still on the conversation of partnerships, and I think this speaks to the success of Snowflake's partnerships. So Snowflake recently named Morgan Stanley's 2025 strategic Partner of the year. So first and foremost, congratulations on that accomplishment. But with a partnership like that, how do you really approach co-creation of value, especially with a firm that operates in such a regulated space, ensuring that there's obviously positives coming at it from both sides or for both sides.

Christian Kleinerman:

When I spoke earlier about the relationship with our customers, that it's not a vendor-customer relationship, but it's truly partnership. The relationship with Morgan Stanley stands out as a great example. They came to us and they said, this is an example, I have maybe a dozen or more. But an example from 12 months ago, they said, in talking to the regulator, they need a solution to make sure that some data is truly immutable. It means that once it's written, nobody can delete it. Not even a system administrator. Snowflake didn't have such a capability. And instead of Morgan Stanley come and saying, you need to do this specifically, they told us the requirement.

We work with them. We spend actually a number of iterations and jointly we ended up creating something that's called database snapshots. It's now a feature of Snowflake and it addresses that requirement. Of course for Morgan, but for others that [inaudible 00:23:10] subject to a same regulation. So the partnership with them is, I would say second to none. I speak with them on a very regular basis, and I don't even think of them as a customer. I think of them as a true partner. We compare notes on who is doing what that is interesting in the industry, compare notes on company startups, partners that can help them or that can help us, and we're always working together.

Lance Glinn:

So Christian, as we begin to wrap up our conversation, and this is kind of a tough question because you talked about the pace of which things are going now and it's constantly changing. Say we have a million use cases for AI, it could be 10 million by next Monday. It is constantly, constantly moving. So you've seen this data space and AI really evolve dramatically over the last decade. Just general, what excites you most about going forward and what it could possibly bring to all of us sitting here?

Christian Kleinerman:

Got it. There's a lot to be excited about. I've always believed in the power of data. I've seen study after study where companies that are leveraging data are outperforming their peers. At this point, you don't have to take my word. There's-

Lance Glinn:

Clear-cut evidence that is the case.

Christian Kleinerman:

...evidence on that. And the coolest thing of what's happening right now is we're in the era of AI. And AI is the ultimate application of data, it's the ultimate technology that will help organizations get value of data. So I would say, in some ways we've been preparing our entire shorter life of Snowflake, but maybe for me, a longer career in data is preparing for this moment where the ability to leverage data gets maximized and amplified. So that's the number one thing that I think is exciting for me personally. It's exciting for Snowflake and hopefully it's exciting for all of our customers.

Lance Glinn:

And so we spoke to Snowflake success since its IPO five years ago here in the New York Stock Exchange. Obviously the company has come a long way with still so much space for continued growth and continued success. But when you think about the next five years under Sridhar's leadership with the leadership team as a whole, what's the vision? How do you see Snowflake continuing to evolve and continuing to grow?

Christian Kleinerman:

Yeah, we want to help organizations with the entire life cycle of data, from when data is originally created, whether it's an application or a sensor or a device, all the way when it's ready for AI to get value out of that data, and we want to create that ecosystem of parties that collaborate based on data products. That's the AI data cloud. That's how we think about technology plus ecosystem. And over time, and this is very true on the financial services industry today, over time, once you are a Snowflake customer, you're part of a network, you're part of people that can collaborate with data, that's the broader vision and it's ambition, it's ambitious, it's large, and we're making great progress in that direction.

Lance Glinn:

Well, I love highlighting Snowflake here on the Inside the ICE House podcast. So many great things happening for the company. Christian, thank you so much for joining us.

Christian Kleinerman:

No, thank you for having me. And we also love the partnership with the New York Stock Exchange.

Speaker 1:

That's our conversation for this week. Remember to rate, review, and subscribe wherever you listen and follow us on X at Icehouse podcast. From the New York Stock Exchange, we'll talk to you again next week. Inside the Ice House. Information contained in this podcast was obtained in part from publicly available sources and not independently verified. Neither ICE nor its affiliates make any representations or warranties expressed or implied as to the accuracy or completeness of the information, and do not sponsor, approve or endorse any of the content herein, all of which is presented solely for informational and educational purposes. Nothing herein constitutes an offer to sell a solicitation of an offer to buy any security or a recommendation of any security or trading practice. Some portions of the preceding conversation may have been edited for the purpose of length or clarity.