Nicolai Tangen: Hello friends, I am Nicolai Tangen, CEO of Norges Bank Investment Management (NBIM). Today with me is Sridhar Ramaswamy, who is the CEO of Snowflake. Snowflake is a data platform on which many of the world's largest companies run their work. So when your bank approves a loan, or a hospital brings patient data together, often Snowflake is working behind the scenes. Sridhar spent 15 years at Google, where he grew the advertising business from $1.5 billion to over $100 billion. Then he left Google to start his own company Neeva, and just 2 years later he became the CEO of Snowflake. Now, here at NBIM, we are also investors in Snowflake and use its products on a large scale. Along with this, we have about 2 petabytes of data on Snowflake, which is equivalent to about 20 lakh gigabytes of data. And we do about 30 lakh queries in the database every day. So, a very warm welcome to you, Sridhar.

Sridhar Ramaswamy: Thanks Nicolai. Great to be here. I am very excited about this.

Nicolai Tangen: First of all, how would you explain Snowflake to someone who has never heard of it? In short.

Sridhar Ramaswamy: Yes, so we are a data platform. We are like a cloud computing platform, like AWS, but our special focus is on data. We help you bring data from different systems, analyze it, get insights from it, and then take it to the systems where you take action. That means we are an analytic data platform.

Nicolai Tangen: Who are your clients?

Sridhar Ramaswamy: Oh, if you look at it, about half of the addressable Global 2000 companies, meaning companies outside of China, are our customers. This includes hundreds of customers in financial services, healthcare, advertising, and many other industries. So, we can say that the list is quite long. And we operate in more than 25 countries, while our customers are present in even more countries.

Nicolai Tangen: When the company started in 2012, the idea of separating storage from compute was quite radical. Tell us a bit about that.

Sridhar Ramaswamy: Absolutely. The easiest way to understand this very important concept is to think about how we have always bought or used computers. I tell people that for the last 50 years, whenever a company needed computing, we bought a box. That box had a fixed quantity of storage, compute, and memory, which was fixed beforehand, and usually you had to work with that same box for about 5 years. So if later you needed more memory, nothing could be done. Just wait. And if you bought more compute or CPU power than needed, even then nothing could be done because you had already bought the machine. That's the real thing. By the way, your phone is also a box. It's just a slightly smaller box. And moving data from one box to another was also quite difficult. You had to do a separate integration project for that.

So, this is where Snowflake changed this whole thinking using cloud computing. You can think of the cloud as systems with almost infinite storage and tremendous compute power that many customers can share with each other. We separated compute and storage from each other. We said take as much storage as you need, and take only as much compute power as you need. If you want to do a deep analytic job, because someone inside NBIM got a brilliant idea for investment for which you had to go through every piece of historical information, then you can start 1000 computers in a weekend, do the analysis, shut them down, and then find out how to apply it in practice. This is the magic of Snowflake's model.

Nicolai Tangen: Most software companies charge per seat. But you charge only as much as people use. Why is this method better?

Sridhar Ramaswamy: Because this keeps pricing directly aligned with value creation. One benefit of this is that we are quite good at estimating demand because we have more than 13,000 customers. Therefore, we manage and amortize the fluctuations in demand ourselves, and we are able to give you a more stable and better model. This is beneficial because you do not have to make a commitment beforehand about how much storage or how many CPUs you are going to use continuously. And this means you spin up resources only when they are needed. Most analytical jobs do not have uniform demand. As you know, even in transactional workloads, there is a big difference between day and night. People do not do that many banking transactions at night. That's why there are many such patterns that are already well understood. Because of the consumption model, we are able to give you extremely affordable pricing because we can manage and amortize this entire load on a large scale. This keeps our company completely aligned with value creation. You pay first, but we recognize revenue only when you and your team use Snowflake. We will talk more about this. In the world of AI, this is a very big thing because you pay according to use.

Nicolai Tangen: Absolutely Sridhar. Now, why do you consider a company like Anthropic as your competitor? Because they are completely changing the software world.

Sridhar Ramaswamy: Look, I believe that for many of us working deeply in the software industry, it has now become absolutely clear that the cost economics of software is changing very rapidly. And if you ask me, I would look at it separately from near-term or temporary concerns like the cost of AI models. The reality is that for the past 50 years, making software has largely been a kind of craftsmanship. Some special people make software by working very hard. Making it is also difficult, and integrating it with different systems is also difficult. You can think of the best programmers as concert pianists. Such people cannot be prepared in large numbers suddenly. You put in 10,000 hours, and you have that special talent in your brain. Only then do you become a great musician. It has been somewhat similar with software engineers so far. But in the case of software, what these models represent is an industrialization of software development like we have never seen before. And so if you look at it, this is the biggest reason why I consider model companies and the coding agents they are building to be our biggest competition. So keeping all this in mind, we have also built our own coding agent. These are becoming the front door of computing and the entire information ecosystem. It is starting with software engineers, but its impact will go far beyond that.

Nicolai Tangen: Absolutely. But what about the likes of Amazon and Microsoft, who also have competing products?

Sridhar Ramaswamy: Yes, I mean, no offense, but when it comes to pure data platforms, we have always been top of the charts. And look, companies that operate on such a large scale should never be underestimated because they have great resources to solve the most difficult problems. But as far as I can see, I think these coding agents are the biggest threat to the entire software industry. And in such an environment, finding a path for Snowflake where it not only survives but also thrives, this is my number one challenge right now.

Nicolai Tangen: Okay, so carrying forward the conversation about AI, tell us in what ways you are taking advantage of this revolution.

Sridhar Ramaswamy: Okay. So the thing is that if I look at AI and a software company like Snowflake in a very simplified way, basically at our core we do two things. We build and run excellent software, which is the job of my engineering team, and we help sell and implement our software to customers like you. So a lot of our energy is focused on how to make this fast.

On the sales side, our sellers now have tools that give them instant access to information. A sales agent is present in their phone. Our solution engineers can prepare a customized demo for Nicolai in just 30 minutes, which will have data that looks exactly like your bank's data. With the help of these tools, their ability to achieve better results for customers is truly wonderful.

And if we talk about software engineering, coding is changing completely. I am just coming from a meeting where we were talking about how we should bring even more people into spec-driven coding development, in which you just write an English language spec for what you want to produce, and then you automate the entire process, like writing the first version of the code, testing it, deploying it, and other things. Therefore, software engineering today is not at all what it was just 2 years ago. The biggest challenge before us is that every engineer on our team learns to work this way and takes full advantage of these new possibilities. But we also have superstars who are 50 to 100 times more productive than an average software engineer.

Nicolai Tangen: Okay. So my next question is, after the arrival of AI, how are large companies changing their data estates?

Sridhar Ramaswamy: Two things. First of all, AI makes your data much more accessible. This is what Snowflake Intelligence does. It is like an agentic interface for your data, where you can directly ask: "How are my biggest investments performing? I see this investment going down a bit. Tell me its breakdown. Is this because of the sector or some other problem?" All such questions, for the answers to which analysts had to be consulted earlier, are now at your fingertips. And its direct meaning is also that due to this fast programmable access to data, the entire way of working in the future is also going to change very dramatically.

The second way our customers are benefiting is things like native coding agents. It does not often happen that I go to my customer and say, "I can make something 10 times or 20 times faster that you have to do every day." This is the work we are doing with our internal teams, and this is the value we are delivering to our customers as well. Therefore, whether it is the data creators within a company or the data consumers, both are getting tremendous benefits because of AI.

Nicolai Tangen: Now, we have a thing called MCP that connects us to you. Can you explain in short to our audience what MCP actually is?

Sridhar Ramaswamy: Yes, roughly understand it like this. MCP stands for Model Context Protocol. It is just a way by which any language model or coding agent can talk to a data source and get information from it. And we want our customers to use products like Snowflake Intelligence, which make data available directly and easily. At the same time, our focus is also on interoperability, which means working with the tools you already have. Therefore, every agent built in Snowflake can be accessed remotely from any external front end or interface that you use. That is what MCP does. It is an interoperability layer.

Nicolai Tangen: How will these agents change your work?

Sridhar Ramaswamy: That is a big question, and I feel that the word "agents" is also often not understood correctly by people. Agents should be understood as a combination of a model and some code that has access to various tools underneath, and they smartly know which tool to use when and how. For example, if you tell it, "Write a little program for me," it knows how to make a file, how to write code in it, and then how to call some local tool or system to run that code. But this also works when you ask, "How is my portfolio performing?" The agent knows how to call the portfolio tool, how to analyze that data, and then how to tell you the result. And therefore, agents are changing work at every level. Although you hear about things like coding agents most of the time, the general concept is very powerful. These coding agents are actually a kind of abstraction agents. Therefore, the person who never wants to write a single line of code will also get tremendous benefit from this. They will get access to all their documents, access to structured data in Snowflake, and they will be able to use all of this in a very easy way. For example, if I give you Snowflake Intelligence, which has access to your data, you ask any question related to data, the agent can make a plan for it and help you execute it. This is why there is so much excitement about things like Cloud Work or Snow Work, because it is changing the way people work and think about work. Everything is programmable, and everything is interconnected too. If you have to send an email based on some analysis, there is no need to copy-paste. Just tell your agent, "Send this email to Stephen," and the email will be sent.

Nicolai Tangen: Absolutely. Now let me tell you one thing: we have spent years cleaning up our data, and to be honest, it was not a fun job. Data is often scattered. There are duplicates, there are gaps, and decades-old legacy systems are somehow connected to each other. In such a situation, how big of an obstacle does this become for the use of AI and the adoption of your product?

Sridhar Ramaswamy: Earlier, this used to be a very big obstacle in getting the right value from data. But I feel that this problem is getting a bit easier every day. This is the real magic of applying coding agents to data problems. Earlier, if you had to add just one additional column to a complex dataset, you had to make changes in the data pipeline, and this work could easily take a programmer a week because they had to go into every small detail and work hard. Now we have things like "Skills", which you can understand as programs written in English language. They automate the entire process from start to finish. Someone can just start the work and come back after an hour to see that everything is completed. We are working on things like agent-driven migrations. This is a thing where you can migrate data from a legacy system to Snowflake in just a few days or a few weeks. Earlier, the same work used to take not just quarters but several years, and overall AI can prove to be extremely powerful in the entire journey of data modernization. This is the reason why we are investing at a very big level in this area.

Nicolai Tangen: I wish we had this 5 years ago, but okay.

Nicolai Tangen: So what about GDPR? Is it limiting or slowing down the use of AI? I mean, look at our personal data protection laws.

Sridhar Ramaswamy: Look, a big part of Snowflake's benefits is that we offer excellent governance features. Companies have to make sure that rules are followed, which is an extra burden. Look, I am an American, and everyone criticizes GDPR, and there have been some unintended consequences as well. But there are many aspects of GDPR that show thinking that was really ahead of its time.

Nicolai Tangen: Is GDPR good or bad for Europe?

Sridhar Ramaswamy: I think it is quite mixed. It has had many unintended negative consequences. But it has also taught that regulations should be made very carefully, and people should think deeply about every aspect of it.

Nicolai Tangen: What are its unintended negative consequences?

Sridhar Ramaswamy: Yes, first I'll tell you its positive points. As a consumer, you can tell a company, "Delete whatever information you have about me." This has happened because of GDPR. This is a very big positive point. It forced every company, including my Google Ads team, to track everything we knew about Nicolai. So this is a very positive thing. On the other hand, the walls of prompts that come up in front of you. At one point, you just give up and click yes. I think this is an unintended consequence. I think it has made doing business more expensive for every company in Europe, and ultimately, large companies benefited the most from it because they have enough resources to follow the rules, maintain compliance, and spend money on it. Whereas every new company starting in Europe has to struggle a lot from the beginning to follow all these rules. So by unintended consequences, I mean this kind of thing.

Nicolai Tangen: A different question. Will we have our data centers in space as well?

Sridhar Ramaswamy: To be honest, to answer this question, I am not the right person. But mathematically, it seems quite difficult.

Nicolai Tangen: Okay. So what will quantum computing unlock for Snowflake?

Sridhar Ramaswamy: That is a good question. I feel that first of all, quantum computing, like large AI models, is a very big security risk. And making sure that your data remains secure is a high priority for us. It promises to bring completely new and different ways of looking at things like optimization and search, and we are confident that we will be able to take good advantage of these new technologies and methods. I think that even in the world of quantum computing, the importance of core infrastructure will remain.

Nicolai Tangen: Can we talk a bit about your early days? You are still quite young. But when you were even younger, after leaving Google, you started Neeva. What did you learn from that failure?

Sridhar Ramaswamy: Yes, I mean, there is a part of me that is very idealistic and very romantic about the world I want to see. I consider this a positive thing. Yes, and Neeva was started with the true belief that building a better search engine would not only be good for people but would also prove to be a wonderful business. And one of the difficult lessons you learn is that consumer products are not adopted unless they give a much better experience than the existing options. At one level, I knew search could be better. But I did not have a concrete plan on how my view of search and information could be 10 or 100 times better. Ultimately, it failed because of that. Because we built a search engine that was slightly better and had slightly better privacy. Of course, you learn the hard way that privacy is also like exercise; we like to talk about it, but in reality we don't practice it. Those were difficult lessons, but we had wonderful people in our team. We built excellent technology, and ultimately that went on to become the foundation of AI in Snowflake. So something very wonderful came out of that.

Nicolai Tangen: Yes, I really like the fact that in the US, this kind of thing becomes a learning experience or learning curve. In Europe, they are not seen this way, right? So, very good.

Nicolai Tangen: After that, you went to Snowflake and were made CEO. The share price didn't particularly like this news. Tell me about the early days of Snowflake.

Sridhar Ramaswamy: Yes, I believe that I became the CEO of Snowflake at a very difficult time and in a very critical situation when the company's growth was actually starting to slow down. In fact, on the day I became CEO, we kept our guidance for the coming year a full 5 percentage points lower than the market consensus. This was the real reason for the heavy fall in the market. Of course, it also didn't help that an unknown man who had never played the role of CEO before was replacing the legendary Frank Slootman. In my opinion, facing such difficult situations and emerging from them is the mark of greatness. So we just got straight to work. I had full confidence in the company. It also helped that I had been with this company for the previous 6 months, and then we started focusing on making great products again. Ultimately, that's what makes a difference. People talk about a lot of things. But ultimately, if you are in the software business, the question is: are you building products that people really love? This sole focus on creating more value has kept driving us forward. With this thinking, we built products like Cortex. A little search will show you how much excitement and love there is for this product. It is every software engineer's dream to build a product that people love, and that is what makes real value for the company.

Nicolai Tangen: What is common among software engineers?

Sridhar Ramaswamy: I would say that in the pre-AI era, software engineering meant the ability to think at a high level to solve a problem, and at the same time, the ability to apply things correctly by paying attention to details at an almost passionate level. You had to be capable of doing both at the same time. This is a discipline that does not forgive mistakes at all. If you left out even a comma, the compiler wouldn't say, "No problem, I'll add it." It would show an error, and you would have to fix it yourself. I feel that AI is now changing this entire picture. I feel that the modern AI-driven software engineer has become much more conceptual. Now he is managing a whole team of agents himself. They are using their judgment and understanding about which problems to solve and how solving them makes sense and is useful in the context of the company and the product being built. These are the people who are very quickly learning ways to solve problems using new and excellent tools. This entire discipline is changing every day, every week, and I feel that a year from now, its picture will be completely different. I will give you a specific example. My younger son is 24 years old. He finished studying software engineering from a very good university three or four years ago. He was very proud that he is a systems programmer who understands fine technical details very well. Designing streaming systems with 3, 4, or 5 milliseconds latency was his specialty. And now he works in an AI lab. Recently he told me, "Dad, nothing I learned in school and after that is helping me be successful in my current job. I had to learn almost everything from scratch." This gives you an idea of how much things are changing.

Nicolai Tangen: That's amazing.

Nicolai Tangen: By the way, at Snowflake, you have something like a weekly war room. What is that?

Sridhar Ramaswamy: Yes, this is a concept we applied early on. At Snowflake, over time, work became very specialized. I mean, every step between a product being made and being used by a customer like you became a specialized discipline. You have software engineers, product managers, designers, product marketing managers, technical program managers. And this stack keeps growing. This works fine when you have a perfect product and your target is to optimize it at every scale and deliver it to people. But this doesn't work when things are changing rapidly and you have to build a new product. In such a case, you need a tight feedback loop between the product makers and users. We need to talk to your team every day, every week, continuously. We have to be present on Slack channels because we have to constantly understand and solve things. Honestly, that was the purpose of war rooms. They brought everyone responsible for a new area into one place. I also used to participate in them, and we would discuss how much progress we made this week. This was a way to speed up communication. As you know, in management, you can organize horizontally or vertically. Both have their advantages and disadvantages. Actually, this was an example of organizing people in a virtual but vertical way to get outcomes. I still run such war rooms today. In many of our war rooms, planning happens on Monday, and in the world of AI, I want to see results by Friday. This is the fast pace at which we are constantly changing and evolving today.

Nicolai Tangen: You call yourself an email machine.

Sridhar Ramaswamy: [Laughter] I process emails for a living.

Nicolai Tangen: [Laughter] And me too. But honestly, it is all about information.

Sridhar Ramaswamy: Yes, I mean, no offense. But when it comes to management, I look at it from a similar perspective. I like to work hard. I like to go into details, and at the same time, understand and explain big principles, or help teams make important decisions. Therefore, I consume a lot of information. This is all about building context and strengthening my intuition so that you can make those one, two, or three big decisions in a year that have a deep and long-term impact on your company. Therefore, I see the CEO's role as having a deep and good understanding of what is going on in the company. And then based on that, helping to move in the direction of the right results. By the way, I don't think that giving new ideas, bringing new ideas, or deciding everything is the sole responsibility of the CEO. The real thing is that you prepare an environment where the best ideas can come forward, and your role is of a facilitator. You are the face of the team when you say, "Yes, we should move in this direction. We should launch a coding agent like Cortex Code because that is what will secure our future." Ultimately, it is all about supporting your team.

Nicolai Tangen: What is the most important part of corporate culture?

Sridhar Ramaswamy: I think one of the most important things, but first I'll start with some baseline. Like civility for each other, mutual respect, and ultimately, equal opportunities for everyone. These are basic things that should be the most important baseline for any company, at least any company I am a part of. And I do not accept any behavior that goes against this. I do not tolerate shouting, yelling, or disrespectful behavior from anyone in the company. Such behavior is not acceptable for anyone. And I feel that setting this kind of culture is very important. And besides that, what I emphasize the most is an open culture where things are judged on their merit. Where people do not have to think twice before proving me or anyone else wrong or contradicting us. Where we debate openly on things, where we all come together, and where we reach decisions very quickly that determine where we need to go so that we can execute them together as a team. But that open and honest communication and teamwork is something that I think is extremely critical. A lot of people, and you see this everywhere in the world, are afraid to say clearly what they are seeing.

Nicolai Tangen: By the way, you grew up in Tamil Nadu. So how did that shape you?

Sridhar Ramaswamy: I grew up in Tamil Nadu. I was also in Bengaluru, which is in a neighboring state, until I went to college. I grew up in a lower middle class neighborhood. For most of my childhood, our family of four had only one living room and one bedroom. But this was a family that believed deeply in education to move forward, and in which there was an intellectual curiosity about everything. None of my parents went to college. They only finished high school. But they always put a lot of emphasis on education, and there was nothing they wouldn't do to get a better life for me and my sister and to educate us. So basically, the values are that being knowledgeable, being smart, it is also important to be helpful, and the fact that hard work can make a better life and a better future for all of us. And these are qualities that I still carry with me today.

And along with this, I would also like to tell you that my parents were also extremely open-minded. I went to a college where my parents did not want to send me. It was not easy for them to send their younger son 300 miles away, and they had a lot of hesitation about it. But still, they supported me. Whether it was the decision to go to that college or to choose a life partner, they always supported me. I feel that this quality of openness is also very important. If I had to sum up all these things in a few words, I would say that it is about the importance of education, the importance of hard work, and the ability to adapt to a changing world. These are all things that I keep very close to my heart. And these are indeed the qualities that I talk to my children about.

Nicolai Tangen: If you were to get another degree today, what would it be? What are you curious about?

Sridhar Ramaswamy: So when I was growing up, I was very interested in cell biology and how the human body works. I think it is a very interesting subject. This is a topic that attracts me because on one hand it is very complicated and on the other hand its depth is also limitless. I think studying that area and understanding it would be a very satisfying experience for me.

Nicolai Tangen: Last question: what is your advice to today's youth?

Sridhar Ramaswamy: The same things I said, and I would like to add one more thing to it, and that is that hard work really matters. Being good at whatever you do matters. Having the ability to change your thinking. It is very important to keep learning constantly and adapt to new situations. And the third thing I would say is the ability to face failure and recover from it. Very, very important. It is very necessary that you try new things. But at the same time, it also has to be accepted that in many of them you will fail, and there is nothing wrong with that. Your identity is not defined by one win or one failure. You are much more than that, no matter what you did. I feel that having the right thinking about yourself is very important. So just remember these three things: hard work, resilience, and perseverance.

Nicolai Tangen: Sridhar, there could be no better place to end the conversation. Thank you so much for taking the time. Keep doing great work.

Sridhar Ramaswamy: Thank you.

Nicolai Tangen: Thank you.

Sridhar Ramaswamy: Thank you so much, Nicolai.