DeepSeek, Stargate and AI's $600 Billion Question with Sequoia's David Cahn

You're listening to Gradient Descent, a show about making machine learning work in the real world. And I'm your host, Lucas Buwald. This is an interview with David Kahn, who is currently a partner in Sequoia, where he invests in AI. And he actually was an early investor in weights and biases when he was at cotu, where he also invested in Hugging Face Runway, and a lot of other well known AI companies. David always has a really interesting perspective on the industry. And in the first part of this interview, or the part that we did first in this interview, we talk a lot about his famous article about AI's $600 billion question. But we also get into more interesting topics around his religion and how that affects his perspective on AI and how he thinks about his career. After we did the interview a week ago, two big pieces of news hit around Stargate and around Deep Seek's much cheaper AI model. So we want to actually go back to him and get his hot take on these new topics. So we also cover those at the beginning of this interview. This is a super fun conversation and I really hope you enjoy it. All right, so we actually recorded the episode last week, but then there was so much news to talk about, we didn't want to release it without getting some perspective on the new things that have happened, maybe starting with the Deep Seek launch. Maybe, David, you should actually describe it first, but then I'd love to hear any kind of quick takeaways that you have from that.

Yeah, well, first, it is wild how much changes in AI in one week. I guess every day right now in AI seems to be pretty dramatic. So as a reminder on Deepseek, Deepseek is this Chinese AI model that was launched late last year and it was trained by a hedge fund in China. And the news this week is the market is really focused on these small models. So Deepseek is a smaller model that is a distilled model, which means that it's trained on some of the outputs from existing models. And the thing that everyone is focused on right now is just how cheap this model was to train. And, and the idea that it may be a signal that AI is now commoditizing, or these foundation models are now commoditizing, which is something people had been talking about for the last six to 12 months. And this was just a clear evidence or a clear data point that this is starting to happen.

And now do you, do you think that there's like a major breakthrough here or what do you think is like, going on that that made this possible?

Well, I don't think it's incredibly surprising. I do think it's sort of an update in people's kind of mental models about how the AI space is working right now. The way I see it, you had this moment in mid 2023 where GPT4 came out. And in my blog post about end of 2024 wrap up predicting 2025, something that I talked about was this idea of everyone getting to parity with GPT4. And so the big event of 2024 was, was Google and Meta and other companies and Xai getting to parity with GPT4 in terms of the quality of their models. Now I think what we've all learned in the last week is China's now caught up, right? And so I think this was a trend line that the industry and the category was on, which is everybody is catching up to GPT4. The bigger question in my mind, and even more important to me than the deep SEQ news, is Ilya Sutskever's talk he gave at NeurIPS in Q4 of last year. And so I've been telling people watch that three times if you want to understand what's going on right now. Which is just to say Ilya Sutskever is the guy who basically invented modern deep learning. He invented the scaling law race that we're now on. And there's a few people of course, like Dario, who trained GPT3, who are big, big parts of that. But Ilya is probably the scientist who sort of is most attributed and most known for this. And basically what Ilya said was pre training is dead or scaling laws are dead is sort of this idea. And that sound bite has been going around for the last six months. But I think listening to the full talk and the ramifications of that, that really begs the question of, okay, we knew people were going to catch up on the existing models. This is always how tech works. You come out with a breakthrough and then people catch up. The question is, where is the next breakthrough going to come from? And so I think that deep seek maybe is overhyped and what Ilya said is maybe a bit underhyped or under focused on.

I guess as an entrepreneur in the space, I've had the experience for 15 years of people keep finding ways to make this stuff cheaper. My investors freak out and say, okay, is this commoditizing? And then it seems like always people end up using that cost savings to actually spend more in aggregate. I'm actually even a little surprised that the stock market is down so much today. Like, what's, what's your perspective?

Yeah, it's funny, I, in my first AI $600 billion question post, I basically said, when this stuff gets cheaper and it's inevitable that it's going to get cheaper, this is really good for startups. So I maybe have a bit of the opposite view in terms of the ramifications for a company like weights and biases, or the ramifications if you're the founder of an AI startup. I do find it funny. The market sort of freaks out at the wrong moments, maybe. And it's because these things are lagging. And so people are actually reacting in my mind to news that was pretty clear six months ago. And so I think what really is going on is as these models get cheaper, it's great for the application layer, it's great for builders. If you're somebody building right now in AI and before you had to pay X dollars for all these GPUs and you had to train these giant models on all these GPUs, and now inference is much cheaper. I think that's really good. And this is a theme that Satya Nadella, the CEO of Microsoft, and to his credit, has been saying this for a while, and I think people have not been listening, but Saya has been saying for a while that these models were going to get distilled, they were going to get a lot cheaper. And in his view, that's really good for Microsoft also, because Microsoft can host these models. And so I don't think that this is quite as doomsday as people would have you believe.

And should we talk about Stargate a little bit? Do you have a perspective there? Maybe you should also describe what that is.

Yeah. Sort of wild to have two major updates in one week, and in many ways updates in opposite directions. In terms of Stargate, is this effort to spend $500 billion on data centers, and then Deep Seq is this idea that, hey, maybe we don't actually need to spend hundreds of billions of dollars on data centers. And so I think part of why folks are focused on this, as they should be, is these are two very different directions that AI could take. And the broader market has sort of been pricing in or thinking about the Stargate direction over the last year or so, which is, hey, models are just going to keep getting bigger and bigger. We're going to have bigger and bigger data centers. Those bigger data centers get us better models. Those are scaling laws. And deep seek, of course, is sort of this move in the opposite Direction, which is, hey, these things are going to get smaller, they're going to get cheaper and we're going to get all this renaissance or building at the app layer where we're going to generate value for end users. And so Stargate, I think everyone saw the news, which is it's a very big financial commitment. It's great for the AI ecosystem potentially, right? You have new data centers popping up. We've known that there's going to be this big race to build data centers and that's continuing. To me, the new information from the Stargate launch was actually the revealed preference from Microsoft's behavior. So Microsoft has had a right of first refusal, a new data center building for quite a while and they've been exercising.

What do you mean by that?

So if OpenAI wants to build a data center, Azure and Microsoft have the first shot at building a data center. And so if there's a data center for a new model, Saya is the one who's going to build that. Now Saya came out after the Stargate announcement and basically said, hey, I'm good for my $80 billion that I've committed to AI data centers, but I'm not upping my commitment. And Ben Thompson in his newsletter had a really good analysis of this, I thought, which was we're basically moving from equity funded data centers. You can think about Microsoft and Google and Amazon. They're funding these data centers using the cash streams from their existing businesses. So moving from these equity funded data centers to credit funded data centers or levered data centers. And this is just the idea that instead of funding this off the balance sheet of these big companies, these things are going to actually have to get funded from the cash flows that they generate. And so that is a big shift in the market that is, and a very logical one, by the way. It's a late market, it's a mature market shift. And Ben Thompson compares this to the dot com era and how that was sort of the, the last leg of the dot com kind of push which you move from this equity funded to credit funded investments.

And what are the ramifications of that?

I don't know. I think we're all going to find out. Potentially very positive, right? I think if you believe that these data centers are ultimately going to yield AGI, and I think a lot of people do believe that, then potentially this is an incredible move in the right direction. If you think that the deep seat direction is actually the direction the world is going to go and these models are going to get smaller Then maybe it counters that narrative. And so I think there's a tension between these two stories about how the world works. And frankly, none of us know where the world is going to go. And so that's what makes it so interesting.

Does this stuff affect Sequoia in real time? Like do you guys get together and say, okay, like do we need to update our worldview today? Or does Sequoia take like more of like a longer view where you kind of wait and see? I guess when you're investing in private companies you don't have to react so quickly necessarily.

I would say we tend not to be reactive in general and we tend to take a very long term view. And so day to day movements don't really matter for us and we don't think about them and they don't affect our decision making. In this case, what I would say is these updates were very much in line with our broader thematic thesis and just how the world is moving. And so I would say like everybody else, we were surprised by the extent of, of these changes and the extent of the reaction to these changes. But I wouldn't say it's dramatically reshaped our perspective on AI.

I think one of the ways that I've seen you come up outside of knowing you, has been this, this first article that you wrote that was called AI's $200 billion question. And then I think it kind of got updated with AI's $600 billion question. I actually imagine a lot of folks that are listening to this have read the article and of course we'll post it in the show notes, but I thought let's not assume that someone's read it. Maybe you could kind of lay out the case. I think especially it feels like your article might be still a little bit confusing to someone without your finance background. So try to dumb it down for the, the AI audience here, if you don't mind.

Yeah, happy to. I mean, I think just to start, the basic question that I was trying to answer, and this was from my own curiosity at the time, was where's all the revenue? You know, we're spending hundreds of billions of dollars on AI. The hyperscalers are each building out these massive data centers. And so I just didn't have a great intuition for what is the scale of this investment and what's the scale at which the AI ecosystem needs to get to in order for these investments to make sense. And so AI's $200 billion question, which became AI's $600 billion question was basically some form of napkin math that I had sort of come to or intuited over time in terms of thinking through how do we go from data center investment to the amount of revenue that's required to pay back these investments for AI?

And so this, this number is like the total spend on data centers in some period. Right?

So let me walk through the basic math, which is awesome. So in 2024, the idea, when I published the $600 billion piece over the summer, we thought that Nvidia was going to do about 150 billion of run rate revenue in 2024. And so for every dollar that you spend on GPUs, you have to spend a dollar on the data center. Energy generators, power, all this stuff. And so assuming that the people who are buying these GPUs are using them to create AI systems, you can apply this two to one multiplier and say, okay, so for $150 billion being spent on GPUs, we're going to spend $300 billion on data centers. And then the question was, okay, fine, we have this GPU capacity, we have these data centers. But now some startup somewhere, some enterprise somewhere, someone's going to use that data center capacity and they're going to deliver a service. So you're a startup using OpenAI's API, they're calling a data center somewhere in Microsoft's data center. To run that model, you want to earn a margin. And so I just assumed for the sake of argument, that you, Mr. Startup, want to earn a 50% gross margin on your product. And so that means for every dollar you spend on AI, you need to generate $2 of revenue to get a 50% gross margin. And so you basically get this second multiplier. And so you take the $150 billion, you double it to 300 billion to get the amount of investment in data centers, and then you double it again to 600 billion. And you say, okay, 600 billion is the amount of money that all the AI startups combined need to generate. And enterprises and all of the companies using AI combined need to generate to pay back one year of investment. And that's maybe one final important point, which is 600 billion, isn't this like total number into infinity? Actually, if in 2024 we spend 150 billion, that's 600 billion of required revenue. If in 2025 we spend another 150 billion, now it's 1.2 trillion. And it goes up and up and up, and it's almost this debt that we've sort of invested in that we now have to go pay back over time in terms of AI revenue.

And the latest revisionist article I think goes back to this summer and obviously things have been been changing a lot. How would the article change if you were going to write it today?

I think this is interesting and this may surprise people, but it wouldn't change very much. And I'll break it down into the two pieces again. There's a cost piece and the revenue piece. So on the cost side of the equation, there's been a remarkable stabilization in the amount that the big hyperscalers, Microsoft, Google, Amazon are spending on their data centers. And in my most recent piece, which I published in December, I basically laid out the math, which is Microsoft is Now spending roughly $20 billion a quarter on new data centers and that's started to stabilize in Q3, in Q3 and Q4 of the last calendar year. And Google is spending about 13 billion per quarter on data centers and that's also started to stabilize. And then Meta and Amazon, while they haven't fully stabilized, I expect, will stabilize around the same level. So you'll see Amazon stabilize In the low 20s similar to Microsoft, and you'll see Meta stabilize in the low teens, similar to Google. And so what that means is AI. $600 billion question is not going to become AI's trillion dollar question. We sort of reach this state at which it's no longer growing, it's no longer exploding. Now the second piece of that is the revenue piece. When I first published a piece, I basically said, hey, we need to generate $600 billion revenue. Are we doing that? Like, are we succeeding at that objective? And I tried to count as generously as I could. I said, okay, so some billions of dollars of revenue from OpenAI, you have a bunch of startups that are each generating tens or hundreds of millions of revenue. Let's assume the big tech companies are generating $5 billion of revenue, which I think was a very generous assumption. And in the summer of last year, I basically said, hey, there's this $500 billion hole where there's this huge gap between where we need to be and where we are. Now. I think if you fast forward the clock to today, that's also pretty similar. Sure, OpenAI is bigger, anthropic is bigger, but OpenAI is still the lion's share of revenue that's been generated in the AI ecosystem. The big tech companies have not fully unlocked AI revenue in their businesses. We saw this week Google announced that you have to buy their AI product through Gmail. Right. So they're trying to figure out how to distribute this AI product, but we're actually in a very similar place to where we were in July of. Of 2024.

So where then is this money coming from to. To buy all these data centers if the revenue is not being generated? I mean, there's certainly not enough VC investment, I think, to fund hundreds of billions of dollars of spend. So doesn't it have to be inside of enterprises? This is really happening.

This is a great question, and this was something that I. All of these pieces that I'm writing, it's just me thinking out loud. Right. I asked the same question. I didn't know the answer. And then I went and tried to figure it out, and I published this piece about what I called the prisoner's dilemma of AI. And I basically explained that the cloud business itself is about a $500 billion business. And the cloud business is the golden goose for Amazon, Google and Microsoft. And it's a very, very profitable business. And we now live in a world, and I think this is remarkable even in the history of business, right, where seven companies represent 33% of the S&P 500. Right. The magnificent seven. And so what's basically happened is you have these seven oligopolists, monopolists, whatever you want to call them, and they have this golden goose, their cloud business, and they're all trying to protect their golden goose. And so I think the AI what's happening in AI is so fascinating in part because we're living in this unique time where we have this very powerful cloud oligopoly that is competing with one another. And so if you're sitting in the board meeting of Microsoft, you're telling yourself, hey, we don't want to fall behind Google. If we do, we're going to lose our position in this amazingly lucrative oligopoly. And so you have this sense that everybody has to spend in order to compete with one another. And so what? Basically, to answer your question very directly, the dollars that are going to fund these data centers are coming out of existing profits from what is a very lucrative business that's been built over the last decade.

I see. So it's the cloud companies that are just using their own data centers that are. Are funding most of this. Do you have a sense of, like, what they're trying to do with these huge GPU spends?

Well, I think they're trying to win the AI race. Right. And I think there is a genuine optimism in these companies. I do. I think they really believe that there's this huge potential for AI and at the same time, I think they're afraid. They're afraid that if they don't spend, their competitors will spend and they'll fall behind. And in Q2 of last year, shortly after I published the $600 million question, Mark Zuckerberg went on Bloomberg and basically said, hey, the opportunity is immense, but also people are spending a lot of money and we might not exactly pay it back, but we have to, because if we don't, we're going to miss this incredible opportunity. And then Sundar, on Google's earnings call that summer in August, basically said very similar thing. We have to spend or we're going to fall behind. So I think it's a competitive race and everybody's hoping and believes that the revenue is going to catch up. And for what it's worth, I actually also believe that the revenue is going to catch up. I'm very optimistic about AI and so I'm not calling it out because I think that it's never going to happen. It's just a very, very big number. And it's good to have an intuition around where we have to get. And I think it also helps us drive. If we're building startups, if we're investing in startups, okay, we got to build things that are going to be massively impactful to people in order to make these investments work.

Do you have a sense of where those areas are?

I do, and I'll give you one example that I'm very excited about right now. In this 2025 predictions piece that I wrote, I basically said, hey, I think that AI search is the killer app. And I have become really obsessed with Perplexity as a user. And I'll use that as an example, but then I'll explain. I think this is not just one use case. I think this kind of spreads farther. So I use perplexity 10 to 20 times a day. And I think it's this incredible research tool, right? You think about search, pre AI, and it was a navigational tool. It found where the answers were and it told you to go read the answer somewhere. And now with AI search, you basically have this informational search where it can go find, it can go through that index of the web, right? It can go through the information that it's learned from pre training and it can synthesize that information and just give you the answer directly, which is a huge, I mean, that's an order of magnitude improvement in terms of the capability that I'm getting totally and the thing that I think the sense in which I feel like I've sort of seen the future a little bit on perplexity is I think that perplexity is actually mapping itself onto sort of my theory of mind. And I'll explain what that means. People kind of use this word, cognitive architecture a lot to describe how AI agents work. And I've sort of inverted it. Like I have a cognitive architecture, you have a cognitive architecture. I think depending on what we do, our profession, the way that our brain works is sort of changed by that, our patterns of thought are changed. And so I think that you spend a lot of time being an investor and you start to think a certain way, you spend a lot of time being an engineer, you start to think a certain way, you spend a lot of time being a lawyer, you start to think a certain way. Same for a doctor. And if you have friends who are in these professions, you sort of get a sense for some of the patterns of thought that you have to have in these type of careers, the way that you solve problems. And so one opportunity that I'm really excited about is the proliferation of AI search engines for different professions and different types of human beings that map themselves onto our patterns of thought, our theory of mind, if you will. And so what makes Perplexity so powerful for me, it's not just that it's doing this search, it's not just that it's using the models that are Pre trained by OpenAI and Anthropic. What makes it so powerful is it understands how to deliver the information in a way that I'm going to be very productive on it. And I think Perplexity has incredible product market fit with people who are essentially researchers. Right. And as an investor, part of your job is to essentially be a researcher now, to extend that to hopefully people in the audience and other people who are not researchers. Right. And maybe aren't using Perplexity. I think that this will be true in a lot of professions. Harvey is doing this in the legal field. If you're a lawyer, you have certain ways of processing information, certain ways of solving problems. And I think you're going to use Harvey and they're going to train it on a bunch of proprietary data so they're going to have better insights for your profession. And you're going to use that at work maybe 10 times a day, 20 times a day. Open Evidence is a company that's doing this for doctors. You can search and you go through all the medical literature and it understands, again, doctors are trying to solve a patient's problem. Right? So the way they want to see the answer to that question is not how I probably want to see that when I'm doing a search. And so I think that everybody is going to end up having this AI search pal or AI search assistant. And I think it's going to make us as humans and knowledge workers, much, much more productive. And I've seen that uplift in my productivity, and I think we're going to start to see that proliferation. And so I think that's going to drive incredible productivity and revenue uplift over time.

It's kind of interesting because, Harvey, I feel like the important part that I would have pointed to is that it has proprietary data sets that you want to look at. And I also like Perplexity, and I feel like I use it for a lot of personal research, maybe a little less professionally than you. You know, it's an interesting point of view. It makes me think maybe Perplexity is going to be even more successful than people say, though, because don't you think most jobs are. If you're going to define research this broadly, wouldn't you say most jobs are primarily research?

I don't know. I don't know. I know that my job has a lot of research and. But I agree with you. I think a lot of, you know, white collar jobs are essentially research jobs. And so I guess the way I think about it is I think there's kind of four dimensions that you can differentiate on inside of these AI products. The first, and I think the one that's most underestimated, is intent extraction. I think Perplexity does a really good job of understanding what my intent is when I do a query. And I think if you specialize by domain, you're probably going to be better at intent extraction than if you don't. So because Perplexity understands to some degree who their user base are, I think they're able to respond. And I've talked to Arvind about this, and I think they're doing a lot of work. This is sort of like hidden work. These AI products look so simple, but there's a lot of work that goes into these intent extraction models. I think the second is proprietary data, which you mentioned. And I think we're in the first inning of this, you know, like it's going to. Now that we understand that we can get so much value out of this data, I think every company that's competing in these fields, there's going to be a race to collect this proprietary Data, maybe some of it's going to be synthetic, maybe some of it's going to be human generated, maybe some of it already exists, we just need to go find it. So I think that's a second dimension. I think a third dimension is like formatting. How do you actually process that data? If you ask ChatGPT, I've done this a bunch. If I ask ChatGPT a question, it gives me this long prose answer. And as a consumer, maybe I'm asking it about philosophy or history. That's exactly how I want to process that information. I do want a long kind of prose answer. If I'm asking a business question, hey, tell me the market size of the auto sector, I just want the answer and I want it as fast as possible. And so I think you see, actually the formatting is different across these two platforms. And then I think the final thing is this cognitive architecture point, which is to what extent are you mapping yourself onto who your user base are? And I think that's beyond intent extraction because I think you can actually design the product and design the experience such that a certain user gets this incredible experience.

Interesting. All right. I want to, I want to take a step back in your career a little bit and sort of understand kind of your, your thinking, you know, when you started and, and you're thinking today. I think you were, you were really early, I guess, like, like weights and biases and like me to kind of get interested in AI and developer infrastructure. Can you talk about, you know, what you were thinking then and maybe what surprised you along the way?

Yeah, yeah, I'll start with developer infrastructure because that came first. And I would say that was my first real strong investment thesis was, you know, I studied computer science in college and I always thought of myself as this 7 out of 10 good software engineer. But I had this perspective that there was going to be a lot more of me, a lot more software engineers like me, and fewer that looked like the Netflix engineer who builds everything themselves. And they're brilliant and they understand how the compiler works and they understand everything low level about the operating system. And so I sort of had this perspective that it's so economically valuable to be a software engineer. You're going to see this proliferation of those, of those people. And so the original investment thesis was what if you could make developers more productive? And that led me to Supabase, which is now used by 30% of YC companies. More than 30%, I think, of YC companies. It led me to Replit, which has tens of millions of Users now. And it also led me down the open source suite where you have things like Confluent Databricks and Starburst which build on these open source projects, which if you were at one of these big tech companies and invested hundreds, thousands, tens of thousands of hours in building this very robust infrastructure, now you can make it available to every developer and not every company doesn't have to build it themselves. So that was the first investment thesis. And I would say I came to AI a bit later after that. And maybe interesting for you, I think the AI thesis I really articulated and after we invested in weights and biases, and so I'll explain how that happened. I can pinpoint, I sort of have this moment in time that I pinpoint as the origin of the AI thesis. And I remember having this insight. I was looking at Snowflake at the time and I loved Snowflake and Snowflake. I had this intuition, hey, Snowflake is on the path to be $100 billion company right now. And what does Snowflake do at the bottom? If you sort of really break it down, Snowflake is a company that makes charts so that you can make good decisions for your business. And that's really economically valuable if you're going to make the right decisions. And I had this moment, or this aha moment, if you will, where I basically said, well, what if you could do more than that? What if it's more than pretty charts that are going to get made with data? What if you can make decisions? What if you can actually automate processes with this data? And naively I just said, okay, maybe that's a trillion dollar opportunity, a snowflake's $100 million opportunity. And so that was sort of the articulation of the thesis. And I think it was influenced also by the work that you and I were doing together. And at that point, this was probably a year into working with weights and biases, and I'd been going to these board meetings and you had shared these anecdotes of companies who were using weights and biases. And I distinctly remember you telling me about John Deere and you said, hey, John Deere is using weights and biases, and now they're using AI to automate how they spread fertilizer. And I just thought, wow, I mean, that's such a phenomenal example of how you can use technology. I remember talking about Qualcomm in the early days. You had these real companies that were using deep learning, which was really nascent at the time. And so I think the combination of seeing it in action and sort of having this sense of what was possible with weights and biases and then looking at the broader market opportunity led me down the AI rabbit hole. And of course we've all been shocked and surprised by how far it's come in the years since then.

Okay, I want to get more specific about your investments. You've obviously had some big wins and I think some, maybe not wins that you didn't want to name, but we can maybe guess who they are. And I was kind of curious if you have takeaways from that, from investing in AI companies for probably longer than most people, honestly, maybe it's the same patterns as investing in non AI companies. I'm curious about that too. But I know you like to do these kind of retrospectives and sort of think deeply about your decision making process. What have you learned?

Yeah, I'll give a few learnings. I'll start with weights and biases because that's an easy one. I think for me the big lesson from weights and biases was patience. I had been, you know, I'd known who you were because I used Crowdflower, your last startup in con. Amazing. And so for some machine learning projects. So I knew who you were and I thought Cloudflower was pretty cool. And at the time that we invested in weights and biases, the product was already really good. I think OpenAI was using it at that time already and basically everybody legit doing deep learning was using weights and biases at that time. But deep learning was this tiny niche corner within the machine learning world. It wasn't the de facto way that people were doing things. And so the market was still very, very young. And I remember for the first year the market was just maturing. And then at some point, I think early 2020, things just started to hit and take off. And so when I look back at that with hindsight, when you have a great team and a great product and you believe in the market, you just kind of have to wait and eventually the market comes. And of course in our case, the market came with much more force and much more vigor than I think we would have anticipated at that time. Um, but even before ChatGPT, weights and biases was taking off and deep learning was taking off. And so we were in the right place and just had to wait for it to it to work. That's weights and biases. I'll give a second one, which is hugging face. Hugging face. I invested in later and at the time most people knew that it was fascinating and it had a huge community. And it's memorable for me because it's probably the most competitive investment I've ever made. I mean, you remember this, Lucas, because I think we talked about it at the time, but they had 13 term sheets or something crazy. And everybody wanted to invest in this company. And I just knew I had to do anything I could to win. And so I remember flying down to la, I camped out in a hotel for a week, or to Miami, sorry. And I camped out in a hotel for a week. And then I remember building this AI bot at the time that basically produced reasons. It was like a chatbot built on Transformer library. And it basically said, Here are the 10 reasons that we should work together. I had 10 founders called Clem. So that was memorable because sometimes it is clear what the companies that are working are and you just have to know and get in front of them and win. And so that was memorable. I will give you some of the mistakes. Right. And the learnings. Right. Look, I think there's been a number of companies where there's. There's sort of a consistent thread in terms of the mistake that I made and hopefully have learned from. And that was sort of just underestimating how hard go to market is. You meet a founder, it's an amazing founder, the product's fantastic. And over the years you just, you experience viscerally. Right. Being in the trenches is hard to sell. And I think there's two takeaways from that. One is, of course, you need a good sales team. You need a commercially oriented founder who desires to sell, sell. And he wakes up every morning thinking, yeah, I want to grow this business, I want to grow the sales. But I think there's a second piece which is you need to put yourself in the shoes of the buyer also and say, hey, I'm super busy. The buyer has a lot of things going on. And so if your product is 20% better, forget about it. Right? The buyer is just not going to wake up in the morning and allocate the time to go buy your product. And if the market is really competitive, then you have to somehow convince the buyer that you're different. And I've experienced this as well, where you actually have a better product, but you just can't convince anybody that that's the case. It's not only about having the right product and the right team, but you need to be able to unlock the market in doing so.

I'm so surprised to hear you say that. I feel like what I've observed throughout My life is just feeling like the product matters more and more. And we. I actually think it was somebody at a Sequoia event who said all problems manifest as go to market problems, but it's almost never a go to market problem. And that really struck me as like, yeah, that's definitely consistent with my experience. So do you actually prefer founding teams that have go to market oriented people on them?

I think the person you're talking about is Douglioni has this concept of a product life cycle and just really digging deep with founders. He has this great workshop that he basically does where he'll go deep and say, okay, your problem is manifesting as go to market, but is your product good? Is your marketing good? Have you figured out who the audience is? And he'll go deep on that. And I think that's absolutely right. And there's number of problems that can manifest, but I think a lot of those problems are also go to market related in the sense of have you narrowed in on the audience? Can they spend the money that they need to spend? And the fix perhaps is I'm going to go change my product to go narrow in the audience. But ultimately I think that is fundamentally a go to market problem, which is you didn't build the product for a certain audience. And then to answer your question on what I look for in founders, I like technical founders. I think there's all types of founders can succeed. I do like technical founders. I like to invest in technical things. And so you want to invest in people who are much smarter than you if they're building something really technical and challenging. And so I don't think it's about necessarily having a go to market background as a founder, but I think it is about having a passion and an excitement for go to market and understanding that as you scale as a founder, you're going to have to change and evolve. Brian Chesky has this great quote that I distinctly remember from a Sequoia Basecamp event where he basically says, the rate of growth of a company is constrained by the rate of growth of the founder and how you're on this growth curve. And something that I try to experience with my founders is I'm also on this growth curve. We're all on this growth curve together. And I think part of one of the things that's meaningful in my job and working with founders is you kind of get to grow together over the journey of the company as you and I have over the last six years. And so I do think there is this evolution and I've even seen it in weights and biases. Right. You wake up, you're passionate about Go To Market, you care a ton about our customers, happy you're a technical founder, But I don't know, today, are you, you know, you're a technical founder and you're a Go To Market founder, and maybe you're all of the above.

Yeah. Someone. Someone recently thought that I. I came up through Go to Market, which is the first time, like, in my life, as opposed to. To tech, which I was, I was like, a little alarmed by, but also kind of flattered. I was like, I'm not sure, you know, what to make of that. It never happened to me before, so. Well, how would you describe your evolution over the last six years?

Ooh, good question. I would say that the biggest thing that's shaped me over the last, I'll say eight years. I feel like I've seen a full startup cycle now. I've been doing this job for eight years, and I think that in the beginning I was really focused on, hey, I need to look for great founders, I need to look for great companies, and my job is to get in business with them. And then things, you know, I'm going to work with them on the board. But ultimately, the company and the founder are way more important than me. And so, you know, things are going to sort of take a trajectory that they're going to take. And I think the biggest thing that's changed in the last few years is I've sort of come to this idea that you kind of have to be this true believer, this sort of, you have to go native once you invest in a company. And I think that how that manifests is, sure, at the point of investment, everybody believes, right? But asking yourself at the point of investment, in five years, if and when everybody gives up on this company and says that it's not going to work, am I still going to believe how, how deeply am I going to be sort of passionate about this idea, this company, this founder, this mission? And I think that's a really high bar. And I think that the thing that I've learned over the last eight years is you owe that to founders, you know, And I think the venture ecosystem sometimes can be very fomo y. And so you sort of end up in these. In these situations where everybody's clamoring to get into a company. And you do have to ask yourself, what is this going to look like in 8 years, 10 years? This is going to be a really long journey, and what does that look like? And So I think that's really deeply affected my investment decision making in the sense of holding myself to that standard and making sure that that is what I'm doing.

That's a. That's an interesting answer. I'm kind of surprised by. Do you worry that it's possible that with that perspective, you might kind of continue to engage with something that's not working? Like, it does sort of seem to me that maybe a good way to do venture would be sort of to bounce around quickly and really go deep on, like, kind of whatever is winning right now and sort of just shut down, you know, companies when they don't work. Like, I mean, surely the model of early stage investing is that, you know, most of them are not successful. So do you want to be kind of digging it in five years with, you know, companies where they're. They're really not working or sort of encouraging the founders to move on to something else?

Well, I think it's true. I'll give a broader point, right? I think it's true both for companies that are working in. Both in companies that don't. Because I think for companies that are working, and I can give a couple examples of this, but for companies that are working, you might actually not make the investment if you don't think this way, because in the short term, they don't look that compelling. But in the long term, you think to yourself, you know what? I actually really believe in this problem, and I think we need to solve it. And I believe in this founder. And so even if there's a chance that it doesn't work, I'm still going to do it. So let me give the example of the affirmative investment decision, because I think that's actually almost more surprising, which is I remember investing in Runway, it was pre stable diffusion. And you meet Chris and Anastasis and Alejandro, the three founders, and they just deeply cared about this problem. And they hadn't even come up with stable diffusion yet. So this wasn't even part of the story at that time. But I felt these guys really resonate with this problem. Video is taking off. Runway now is the AI video company that generates AI videos. And I've been using Loom every day at that time. And so video is taking off within the enterprise. And these team, if they weren't doing this, they would be artists. This is the thing that they live and die and breathe by. And so I had this feeling, I want to work with these founders again. They hadn't even invented the thing that ended up mattering for the company. But they cared about the problem and I cared about the problem. And I felt in five years I'm still going to want to work with these guys. And so I think that's a good example in the sense of you almost wouldn't make the investment decision if you weren't thinking that way. And then I think a second example of this is even more radical probably is a company called Form Energy. I met this company at the Series B. The founder had run the energy business at Tesla and he had this idea of storing solar and wind power in these giant batteries, I mean these billion dollar batteries on the grid. And that's a radical idea. I think most VCs, you see that and it's too hard, it's just too hard to. And at the time this was early in the company, so it wasn't like the science was fully, fully proven out. It now is. But I thought to myself, wow, in five years, would I want to work with this? Of course, this is probably One of the 10 most important problems facing human civilization right now. Can we get cheap energy storage? And of course now with AI, that's even more important. And the founders were incredible. This founder had run the energy business at Tesla. His co founder, yet Min Chang, was the chief science officer, everybody agreed was one of the top three battery people in the world. And so again, rather than, you know, I decided to dig in and I hired a Harvard professor and we spent months diligencing this technology. And at the end of that, the Harvard professor said, actually, can you pay me in the form of stock in this company? I really believe in what they're doing. And now fast forward five years, the company's built a billion dollar factory in West Virginia. They're manufacturing these batteries and it has the potential to really change the world. And so again, you end up in an affirmative investment decision because you're thinking five years from now as opposed to thinking in the moment.

It also kind of sounds like from your stories that it was a factor as kind of wanting to work with these people in five years, regardless of what happened. It seems like maybe these applications were not as important as the humans here. I'm not sure.

I think it's the general domain plus the humans. Do I care? Am I going to care that this problem gets solved? I think this is especially the case as I Buffett has this concept that I subscribe to which basically says you have this punch card and you get 20 punches on the punch card for the rest of your life and those are investments and so do you really want to use one of those. So I think the problem matters in the sense of great people working on a problem that I don't care about. They're still great. I'm happy to be friends with them, but I don't want to dedicate the rest of my life to that. But of course, yes, you hear a lot of people, people, people in my conversation. And you know me well. Like, at the end of the day, people are the single most important thing. And a very fascinating problem with the wrong people also isn't interesting. So those are both necessary conditions.

What are other domains that you're interested in where you haven't kind of met the team that you want to invest in?

One area that I'm fascinated by is what is the future of the global supply chain? Talk about another big. I mean, we're in this situation right now where the US is trying to decouple from China, and I meet a lot of companies that are making Made in America things. And you go six layers down in the supply chain, and even the made in America version is using Chinese suppliers six layers down. China is just so much cheaper on so many components. And you think about, I call this the Shenzhen problem. You go to Shenzhen and everybody's running these amazing automated factories. Not only is the labor cheaper, but the machinery is much more advanced, the automation is much more advanced. And you have these agglomeration economies where everybody is working together. And so one problem that I'm fascinated by is it's not enough to sort of assemble things in America. We're going to have to rebuild a lot of the supply chain and probably to pinpoint one specific thing, we're going to have to build some version of Shenzhen in the West. Is that in America? Is that in Mexico? Is that in Canada? I don't know, but I think that's an incredibly important and strategic problem. And we're just starting to scratch the surface of it. Companies that are manufacturing these things that are in these kind of categories are just beginning scaled manufacturing. So that I think is something that's fascinating. And you look at America today, we've sort of gone through this phase of globalization for the last 70 years, and we sort of outsourced a lot of our manufacturing capability. And rebuilding that muscle is going to be really difficult and really important.

It does seem like there's going to be a new genre of this kind of thing with the advances in robotics that seem clearly coming, doesn't it?

I agree. And the AI robotics space, to your point on robotics, it is very Clear that there's going to be some application of this new AI technology in robotics. How is that going to manifest? I hope founders figure it out. I'm not going to figure it out, probably, but I think that the intersection of AI and robotics is going to be very fruitful. And it would shock me if in the next 20 years we don't have some version of these robots in our homes, in factories, et cetera.

I think one thing that I really admire about you and have always admired about you is how quiet and humbly you present. Actually, I think this interview might not even capture how humbly you typically show up. And it's in kind of contrast, I think, to, to most of your peers. And I think I had a perception of venture capital that you really needed to be very sort of aggressive and kind of visible in order to be successful to win these competitive deals. I'm just curious because I think you kind of bring more of maybe a computer scientist or sort of a nerd's personality to the industry you're in. So I thought people might be interested to hear about how you've kind of leaned into your authentic personality, but also had an incredibly successful career and won deals like Hugging Face that were quite competitive.

First of all, thank you. Very nice. I think that there's so many ways to be successful in any career, and venture capital, of course, is one of them. And I try to just be authentic to who I am. I grew up with. The people who I'm investing in are kind of the people I grew up with and was friends with in high school and as a kid. Right. And I think I try to develop individual one on one relationships. Maybe why I'm like more boisterous on this call. Right. Is, you know, we know each other, right? We're friends. We've been friends for a long time. And I think one thing I've tried to do in my role is just see each one on one relationship with a founder as the atomic unit of work to be done. The work to be done is to build a relationship with a founder to understand what they're doing, to believe in what they're doing, and then to be with them through thick and thin over time. And I think if you do that, you can build really strong relationships with people. And at least in my experience, you don't need to have this huge public Persona. And it's not necessary as long as you. And in the one on one context, you should be aggressive. And I think I am somewhat aggressive in a one on one context, which is, hey, I really think that what you're doing is great, or I really think that you should think about this. But I think that context with which that should happen is not the board meeting, it's not in front of all these people, and it's not this public thing. I think it actually is a one on one conversation, a walk around the block, a conversation over coffee. And I think that I deeply care about people. I think I deeply care about all of the people that I work with. And I have this phrase that I use all the time, this mantra, if you will, which is everybody knows everything. I just think humans are so much more transparent than we think we are. I think you know what I'm thinking right now. I think I know what you're thinking right now. Right. I just think we can connect at this individual level and we don't need to do this. There doesn't have to be a divergence between me as an individual relating to you as an individual, and then me as an individual somehow relating to this group of people or relating in a larger group.

I guess you also are. I mean, maybe I'm misread in this way too. I think you obviously are incredibly competitive. Like, I think the story about going to win the hugging face deal by like actually using their technology is a pretty badass story of a particular kind of sales that I think we can all learn from also. But I appreciate the, I appreciate how you can authentically be non transactional in these, in these relationships. I think it's. I'm so happy to see someone with that style be so successful.

Thank you.

Here's one that I actually didn't know you thought this, but when I was just looking at your website and reading your blog posts and stuff in preparation for this, you said that you think there's a trade off between looking smart and being right, which I thought was a really interesting perspective that I've thought a lot about. I think, especially as a founder, when you're trying to present to vcs, there's this thing of, okay, I'm trying to simplify what I'm doing, but it's hard to really simplify it. How much should I bend reality in order to make something digestible? And I've always actually kind of felt that being able to handle that complexity and not need to simplify things, you can kind of lead you to better decisions. But I was kind of curious how you kind of came up with this, with this concept and if you have any kind of stories of how it's kind of Helped you how you've engaged with it.

Great question. I got this from my dad. My dad is a serial entrepreneur and professor and has done a lot of tried nobly throughout his life on many things. And, and I remember growing up, I go to school and you sort of have the teacher and they know everything and you're just the kid. And my dad always had this view, well, the teacher's just a person, you know, and they know a lot of things, but they're a human being. They know some things and not others. You go to college, you have the university president, they're wearing their robes and their degrees and they stand in front of the audience, you know, the whole student body, and they make all these proclamations. And then we saw on October 7th, they're just people. And that's not even to criticize them, but just to say they're human and humans are flawed and infallible. And so I think I grew up with this idea that there's a distinction between form and substance, that form is not indicative of substance. Just because something looks one way doesn't mean it actually is that way. And so you have to really be penetrating and try to look through form. And so much of what we do as a society, as we put up form, right, in order to sort of present certain way or be a certain way. And actually the substance is much deeper. And so I try to do that. And so I would say one thing I love about Silicon Valley is I think Silicon Valley actually embodies some of this. You think about the story of Steve Jobs showing up to Sequoia without shoes on and barefoot, and somehow Sequoia invested. Don Valentine thought it was a good enough substantive idea that he was able to look through the sort of presentation. And I try to do this in my day to day investing and I think it's hard coming back to looking smart versus being right. One way to look really smart is to invest in things that everybody else wants to invest in for a while. That generally works and you look pretty smart. And so there's plenty of investments that I've looked at where I said to myself, if I do this, everyone's going to think I'm so brilliant for two years. But I don't really believe that in seven years. I just can't at sort of this deep level in my gut, I just don't see how that works. And so I wouldn't do that. And on the flip side, you have things where for one reason or another, you're going to look dumb for a couple of years, people are just going to think, oh, this person overpaid, for example, or oh, this idea will never work. And there's a bunch of reasons why you might look not great for a couple years, but substantively you believe that it's going to work. And so this tension between looking smart and being right, I experience it very viscerally on a day to day basis because I think that the decisions that I'm going to make are actually going to be quite different based on which of those I'm optimizing for. And one thing I've discovered, and I think this took me years to sort of figure out and really hone, is that the people I want to work with are also people who are running that algorithm. This is one thing that I just really care about is I don't want to work with people who are presenting something that isn't the true substance of who they are and what they're doing. And this intimacy that you can build with a person when you're both presenting your true self and you're both being authentic to what you believe in allows you to build trust. And that trust is very durable. And so when I look at founders, I also want to work with founders who they're just going to tell it like it is and we're just going to have direct conversations. And by the way, those direct conversations can also be kind of serious conversations. Right? Because if you're not doing this form versus substance thing, I can say, hey, I really don't know about this thing. You should talk to somebody else about it. It's not my area of expertise. And as a. And in conversation you can also say tough things. Right? Hey, I think this person is not the right fit for your team. And because we're both presenting this real substance, we can have this intimate dialogue.

Interesting. And maybe you could segue into another question, which I'm glad that you left on the list because I was curious to get your opinion on, which is I wonder about if your religion influences your thoughts on AI. And I ask this because you're one of the few people and so cabal, I think you talk about religion more than most and I do feel like AI sort of has almost like.

I.

Don'T know, like religion scale implications when you think about AGI and things like that. So maybe this is. Nobody had a question, but I wonder if there's any connection there for you.

I'm deeply fascinated about religion and we had some conversations about, you know, religion in the past. I find all religions to be Interesting, because I think religion is trying to answer this question, what does it mean to be human? And that question of, as you say, has a lot of relevance in AI. And I've read the major books of Christianity, Judaism, Islam, the Mahabharata, the Tao. I've read sort of the books of all the major religions in my own sort of personal quest, my own curiosity to understand, you know, what does it mean to be human? What is this experience? And I think that religion institution, I think of religion as like this like package, like institutional religion is the wrapping paper. And so once you kind of tear away the wrapping paper, what's inside is similar, I think in different cultures and contexts. And what's inside is this moment of confrontation, confrontation with consciousness, confrontation with God, confrontation with truth, confrontation with things that are really essential and important. And so bringing it back to AI. You probably don't know this or remember or know that this influenced me so much, but you sent me geb one year for Christmas as a gift. This is Godel Escherbach, which is a book about, by an AI researcher about AI and he sort of offers his own hypothesis, which is a non religious hypothesis, which says that this unique element of human consciousness is this idea of self reflection, of recursiveness. And he runs that through math, in the work of Godel, in art, through the work of Escher, and in music through the work of Bach. And he says that this is, you know, essentially the uniqueness of the human condition. And so why does that matter for the world today and the work that we're doing in AI? Well, there's this open question of can AI be self reflective? There's this open question of whether we're going to get, we're probably going to get a very intelligent AI whose IQ is higher than mine. And I think we're going to get this very high IQ AI and maybe they're going to be better at making decisions than the rest of us. You sort of have this sci fi image of the computer who's like the President of the United States. You can imagine a world where AI is far more intelligent than humans. But will that AI have this self recursive, self reflective property, this mirroring in consciousness where it knows that it exists and it uses that as a lens through which to perceive the world. And I think that humans very much that is so core to how we experience the world, not just rationally and intellectually, but also emotionally. We can experience awe, we can experience empathy. Empathy is so fundamentally a result of this self reflective property. I know how I experience things, and so I can craft hypotheses about how you experience things. And so I don't have an answer to this question. I think this is one of the important questions. It's obviously one of the big questions. That has always been one of the big questions. But for sure, it has relevance today in AI, as we try to imagine, are we going to have an unconscious, really intelligent AI or we can have a conscious, really intelligent AI. And fundamentally, that's a religious question in many ways. And in order to understand it, you need to grapple with the history of religion. And maybe I'll say one final thing on that, which is religion's been around for thousands of years. One of the things that I find so fascinating about all these different traditions is you have continuity of analysis and thought and work. We bootstrap on the work of the last few years. Religion is kind of bootstrapped for longer than most intellectual threads. The Judeo Christian tradition, you have these. You have these academies. In Judaism, you had these academies that were thousand years in length, and you had generation after generation after generation penetrating the same questions. And the same is true, of course, in Christianity and in other traditions. And so you actually can get a lot of depth and a lot of meaning and a lot of understanding in. And these were the smartest people around back then, maybe the smartest people around today. They're all working on AI. I think one thing that's cool about AI is everybody is focused on AI, but for a long, long time, the smartest people in the world were focused on these deep questions. And so we'll see what happens next. Awesome.

Thanks. Great place to end. Thanks, David. Thanks so much for listening to this episode of Gradient Descent. Please stay tuned for future episod.