She Knows the 250 People Building AI. Here's What They Actually Believe.

Key Takeaway

Build a small “frontier circle” around the people closest to the work you want to understand. This week, identify five practitioners—founders, operators, researchers, or customers—whose work is shaping your field; schedule one conversation, read their latest work, and write down what they see that the wider market misses. Use firsthand insight to form a view before absorbing consensus opinions.

Episode Overview

Sarah discusses how Conviction invests amid rapid AI progress, escalating global competition, and uncertainty about compute, energy, and supply chains. She argues that enduring advantage comes from proximity to exceptional builders, independent technical judgment, and the conviction to act on differentiated information. The conversation also explores AI’s impact on robotics, biology, open-source models, and productivity.

Full Transcript

Transcript of She Knows the 250 People Building AI. Here's What They Actually Believe. from Invest Like The Best. Auto-generated from episode audio; may contain minor errors.

It is a fiercely competitive environment. It seems that many people are particularly concerned about the global competitive environment . Such a situation itself can create anxiety , but it also breaks conventional wisdom. There is a belief that if artificial intelligence research continues to advance and the models themselves improve, exponential intelligence can be achieved within one to two years. Sarah, where should I start this conversation that seems like it's going to be really fun ? Maybe it's because we both have similar interests, but I'm curious what you're thinking today .

It seems like both of us are a bit out of it right now , and since this situation has been going on for quite a while, I feel like things are going to get even more chaotic going forward . It is the same not only with what you do, but also with the world around you. How do you feel right now? What are you thinking? For anyone contemplating how to navigate this period as an investor, the questions naturally arise are how the market will unfold, how it will differ from the past , and what to do if backtesting is not possible .

I had a conversation with an investor friend last night , and the analogy he told me was like this . "I keep saying I want to slam on the brakes, but in reality, I'm driving at 90 miles per hour ." I think the question is whether one side will miss the opportunity or repeat the mistakes made in every boom and bust cycle in the history of technology . I think it is a complex question . Do you think it is correct to say that in our analysis, you are employing a specific strategy—that is, a form of betting or positioning?

Yes, I am a long-term investor. What are your thoughts on the fact that something truly important and fundamental is happening between some research institutes and everyone else in the field of artificial intelligence, and that we must take active steps to ensure this does not lead to uniform results ? I believe in the theory of historical figures, but I would like to answer in a slightly different way . I believe that there are influential people in every field, and if they have appropriate risk capital , networks, or environmental support, the outcome can change .

Therefore, we need to consider what changes will occur in the open source model and U.S. industrial policy . Considering future ecosystem opportunities , does a competitive Western open source model exist ? Is there actually anyone who has created such a model? Would they have had the will to raise funds , invest capital, secure talent, build all the infrastructure, and fight for the frontier ? So I think that individuals, especially entrepreneurs, can influence the results . I do not consider this a war. I collaborate closely with and co-invest with various research institutes, whether large or small, and I have many friends .

However, I would like to make it clear that some people, particularly inside and outside large research institutes, hold the extreme view that someone who owns one or two, or at most three, high-tech models dominates the economy . I do not want such a future. Very clearly, and I do n't think we will rush toward such a future . In theory, are you actively thinking about becoming a 'great woman' ? no. It is not out of my humility that I mean it that way. It's just that there is nothing like that in my goals.

I want to become the best investor in my field . It is not because of a lack of ambition or confidence . Some people seem to be obsessed with the thought, 'I want to be the one who achieves this .' And it was true that I had thought for a long time that I would become a software entrepreneur. From an identity perspective, the thought that I am curious , enjoy understanding things , and want to prove myself right had a significant influence on my decision to become an investor.

And I am greatly motivated by being considerate of others and working with outstanding people . It seemed like using my skills to help them succeed more suited early-stage investment very well . I want to work with the best talents and hope that the companies I invest in will have a positive impact on society . I hope the company can bring about the changes we desire , but I do n't think it is necessary for me to step forward personally. If that is the goal, what do you think is needed to become the best in a highly competitive environment like today ?

What do you think you have done really well so far ? What are the things you have undeniably done well? I am curious to know what effort was required to reach such an outstanding level, and what effort you think will be needed to reach that level in the next year or the near future . Actually, I think it has been quite simple so far . Right? As we discussed on this topic , the environment I observed is that competition was less fierce than before due to corporate growth and generational change .

And tremendous technological changes were taking place. If you understand technology and the community and approach it from the ground up , you will be able to obtain better opportunities and make better decisions than others who lack focus. Right? So, all you have to do is take risks and focus . And then, execution is important. I think that is one of the parts that is quite simple yet difficult . It means that effort is required . So, it seems that it wasn't actually that complicated today .

Rather, I think it is more important what standards you apply to the people you work with. My partner, Mike, and I started based on the relationship and understanding we had already established . So I think that point was useful . Of course, there is likely something more than simply working harder than others . Mike told me an interesting story a few weeks ago , saying that he thinks of about 250 people . They are entrepreneurs, researchers, and people doing the most interesting work in cutting-edge fields.

I mean the people who go to work every morning and make all of this move forward . And I said that one of our company's goals is to stay as close as possible to such people , get to know them well, and support them in every possible way . I think that is a really good idea. That's a great idea. It seems to have meaning beyond simply having many meetings with people starting businesses. From the outside, it will look like much more original things are happening compared to competitors .

We are focusing more and demonstrating better execution capabilities . yes . So I am interested in the factors that have brought success so far . I won't reveal the name , but one of the famous LPs (Limited Partners) conducts an annual survey among other famous LPs regarding where they most want to invest , and it always takes the top spot. So, it means there is something special about the companies you have invested in, their performance to date, and market perception in various aspects . There is something that cannot be explained simply by the reason that "it was an unusual time and we worked hard." So, I am trying to identify the factors behind that success .

That is precisely the reason why I am focusing like this . With the mindset of making a bet that others could not predict , we carefully considered which markets would be important, and if we were right regarding capability growth and the breadth of influence, what made the founders we were focusing on in this era different, and what aspects might go unnoticed by others . In other words, what is the biggest difference from the current situation, and how should the framework change ? Let me give you two examples .

One of the things we particularly focused on in the first year was workflows, job types, and work fields that we thought were suitable for our model . This is a technology-centric approach. Many people say, "That makes no sense. You have to think about customer problems and approach things from a customer-centric perspective." will say that. However, we pursue both. Well, if we consider Harvey and the function of law , thinking rationally, if we already knew at the end of 2022 that language can be used to predict the next token , we can see that law is a structured language .

I am not a lawyer, but from an outsider's perspective, " Ah, I guess I will have to read a lot of documents ." I thought so. There was a search function and a text generation function, and it seemed that the numerous case law texts found not only within the law firm but also in customary law and history fit together really well . When I say 'focus,' I mean having a very specific perspective on what is currently possible and what is valuable within it. And who will join us?

Like Winston and Gabe, they believed that AI would completely turn legal practice upside down—though it is a bit embarrassing to say right now . They said, "We will carry out an incredibly complex task with the lawyers ." He said. I do n't know when that will happen, though. It could be next year or it could be five years from now, but let's start by saying that we can answer the question by looking at a California landlord-tenant agreement, for example . It's a very trivial matter. However, it can be expected that the Activision Blizzard acquisition will proceed and 85% of the work can be handled .

That's a huge leap, isn't it? What appealed to me at the time was the ambition for what was possible in that era and the technical logic behind why it was possible . It seems that they probably had a decision-making framework different from the approach others took at the time . Ramp is the only platform that makes finance teams more efficient, faster, and stronger, helping companies reduce average annual costs by 5% and focus on growth . Ramp customers are growing their sales at a rate 3.2 times faster than the U.S.

corporate average . Over 70,000 companies, including Visa, Versel, Cursor, Stripe, Notion, 11 Labs, and Shopify, are currently using Ramp . I feel the same way, and you all should too. Learn more at ramp.com/invest . Rogo's Felix is ​​a personal financial agent who generates complete financial reports that can be delivered directly to clients by utilizing a company's templates, contacts, and standards with just a single request . "Please write a report based on these opinions ." Or, "Please update the tracker to fit the context of this email ." If you send an email like this, Felix will send you the completed PowerPoint materials, Excel model, and research results .

Felix delivers work quickly and accurately 24/7, just like the team's existing way of working . Learn more at rogo.ai/felix. OpenAI, Cursor, Anthropic, Perplexity, and Vercel all have something in common. They are all using WorkOS for enterprise-scale adoption , and to do so, they must provide core features such as SSO, SCIM, RBAC, and audit logs . Instead of spending months building these core features yourself , you can use the WorkOS API to leverage all the capabilities from the start. That is why a significant number of the best AI teams you have heard of are already using WorkOS .

WorkOS is the best solution that helps you adapt quickly to the corporate environment and focus on the most important product development. Visit workos.com to get started . What is that group—that is, the Frontier Group of 250 researchers—talking about today ? What is the most interesting and noticeable difference when comparing today with 6 months ago and 12 months ago ? The point is that it is a very competitive environment. It was like that 12 months ago, but it seems to have gotten much worse than 24 or 36 months ago .

It seems that many people are feeling anxious as competition intensifies worldwide. This also breaks conventional wisdom . I will present a belief and raise questions about it . AI research is the belief that recursive self-improvement is possible, meaning that models can improve themselves . We think that within one or two years, we will reach a point where intelligence increases exponentially . It seems that many researchers have come to think this way within the last 12 months. However, what factors are at play in the prediction that we will reach such a situation within two years ?

People smarter than me, for example, Andrei Karfay, said in a very self-aware tone, "I thought it would happen within two years, starting ten years ago ." They also say that. Right? To be honest, he still thinks that way . But who knows? Well, I think that because major research institutes require a lot of computing resources and have such large workforces, the perception that "I can make a difference " seems to have decreased compared to the past . If there are about 200 researchers at OpenAI , achieving the goal is entirely up to each individual.

However, if the current situation is that "$750 billion in computing resources are needed and thousands of researchers are working on this problem," it seems that people feel less responsibility for the results. That 's interesting. Ultimately, securing computing resources is the deciding factor, and the individual efforts of researchers are not important . I think many researchers would agree that, given the current situation, it is one of two things . First, the model will handle my work anyway , and second, the only important thing is the scale of computing .

Well, and both of those things make me feel somewhat helpless. Then what are they doing ? For example, let's assume that I believe in both of those things and that I am one of the top five researchers in the world . That tells me something about my psychology . Because I want to do meaningful work . I believe there are scientists who want to conduct research regardless of whether the work they do is important or not . Right? Yesterday, an entrepreneur asked me, "If you think we cannot change their results even slightly, would you not join that company ?" asked .

He also asked, "If you hadn't invested, would there have been any companies you wouldn't have invested in ? Would investors have rejected them ?" asked. It was probably about 5%, or at most 10% . Because they are resourceful and talented people. They would have found other investors . Because there are many smart investors in the world who want to take such risks . However, perhaps they might not have made the next $100 million computing investment . Right? Well, and I don't think all the researchers doing cutting-edge research at the top labs right now feel that they are essential to the system .

What do you think is the most unhealthy part of everything happening right now ? What are your concerns regarding what people are trying to achieve and where they are ? What are you most worried that will stand in the way of the future you want to see? I think computers are one of them. I think people understand that point very well . I mean the expectation that we will be able to secure sufficient computing resources within the next 5 to 10 years . And people are thinking based on the year 2032 from the current point in time .

I spoke with an infrastructure manager at a hyperscaler earlier this week , and he said that there is nothing that will change the situation on a sufficient scale before 2030 . I thought, 'Ah, this is really depressing.' Well, just, you know, there is also the issue of where to get enough natural gas. Americans and entrepreneurs were able to create new technologies and new capabilities . It developed really quickly in the past . I do not think it is a problem of technology, capability, or capitalism .

I view it as both a regulatory issue and a matter of cooperation . I'm not talking about things like AI cooperation . If you want to build a data center in New York, you must persuade New Yorkers, or the American people, that they want a data center there . If you want to make nuclear power competitive as base load power, you must convince people that nuclear power is safe and allow sufficient construction . To us, and even to the other countries. That is the only way the price will go down.

Because theoretically, we know how the cost curve will change . I do not think there is a lack of technical and entrepreneurial capabilities to build the abundant and affordable energy required for data centers . That is the first problem. The physical supply chain is a difficult reality . So, what I am worried about is learning how to make something, acquiring tacit knowledge , and securing labor and raw materials . Well, it ca n't be done as quickly as software development or decision-making. However, the only way is to invest.

My concern regarding investment is that I spoke with one of our founders, and some of them are former researchers . They explained to them that the quality of storytelling and the ability to clearly convey the company's ' investment' to investors are very important for success . It will be a business that requires a lot of capital expenditure in the beginning . However, realistically, this is the financial situation of all of us. I am not a researcher, neither are you, and not all capital is invested in researchers .

So, entrepreneurs have to explain their own stories , but one of the difficulties is that many people are trying to jump into technology-related investments . However, basic understanding varies widely, and judgments are often made based on the company's career or background . Other clear signals, such as who invested and who the recommender is, are important , but who the recommender is is also important . So, decision-making becomes a less fundamental factor . For example, I once asked a friend who knows a lot about investing , and we had a bit of a discussion.

I said, " How can this company be so successful? Explain it to me ." My friend asked, " Do you know this person's qualities?" He asked, and I replied, "Yes, I have known this person's qualities for eight years ." He answered. However, I do n't understand the technical theory of the business. It seems like people make large-scale research investments without any intuition or opinion. Not all investments succeed, you know . I'm not saying I can make better decisions, but I think it's important to have intuition.

Having no intuition is a big problem. It is very dangerous to have no perspective on business other than one's personal career . Who is the researcher who impressed you the most, and how did they make you do so? What was it that made them feel special to you ? Returning to the theory of great figures, I believe there are a few people who made truly special contributions worthy of being recorded in history books. What are some examples of such people, and have you ever seen how they act?

My partner Pranav and I met Tony Zhao and Chen Qi at Sunday Robotics. They were pursuing doctoral degrees at Stanford . They had experience working at Toyota Research, DeepMind, and Tesla, so they could not be called simply academics . They were young doctoral students around 25 years old . One person did not finish their doctorate, though . When they started the company, I thought they were impressive, and looking at their achievements now, it took me a while to understand at first , but I think the two of them have almost exclusively created the most interesting ideas in the fields of robotics and AI over the past four years .

It is truly amazing that the two of them accomplished something like that at such a young age. To summarize their idea , it is about how to practically solve the problems of generalization and robustness in robotics by utilizing modern AI . The robotics community believes that if there were only an internet of robot data, perfectly general-purpose robots could be used everywhere. I think it will definitely be realized . However, one of the important research and practical problems is exactly where to obtain that data . One of the ideas Tony and Cheng have been researching is to find a way to collect data in the cheapest possible way, in a manner that supports the distribution of real-world environments and tasks.

So, I think the creativity in considering the practical constraints of data scarcity and the inability to spend infinite money on data is truly amazing . We must approach this as a technical problem that needs to be solved . What is the form and collection method of the data , and how does it interact with the model ? And to what extent can we utilize this cheaply collected data for model training ? This is a really interesting part. It is a results-oriented approach. I was completely captivated at the first meeting .

Fortunately, I accepted right away. It has been less than two years since the company was established . Although nothing is confirmed until the product is released , our entire team believes that we will be able to showcase a beta version of a semi-humanoid robot that performs various tasks in people's homes by the end of this year . Honestly, none of us had thought of that. It is truly amazing that starting with just a few sheets of cardboard in a Stanford basement, they were able to develop a fully functional product and complete production at such a rapid pace.

We have already manufactured hundreds of products . The process of iteratively developing hardware models, collecting data, converting it into actual tasks, and testing them in various real-world environments to verify that they work properly is really cool. I think the speed is astonishing . Many people in the field of robotics believe that the question now is ' when' it will be realized, not 'whether ' it will be realized , so it was surprising that our team said, 'Even if not this year, it will be possible next year.' I am really curious about the moments when you made investment decisions.

As you just mentioned, you said, "Okay, let's do this," at the first meeting. Is it always like that? Or are there any cases where you hesitated but eventually took action and succeeded ? I would like to hear more about the team's investment decision-making process . I would like to know the specific decision-making process, such as ' This is interesting, let's try', 'Let's not try', and 'Under debate' . What is that? What is the process like ? How is it? yes . Well, it varies from company to company.

Personally, I have a very strong intuition for reading people . And I'm not sure if this is the best explanation, but I often know right away that I want to do something. When you know what someone has been doing, communicate with them to listen to their ideas, and gain background knowledge about them, you can say, as if evaluating them on a scale of 1 to 10, " Okay, that's a perfect 8 or 9. " It seems like they are saying that . And what I do after that, before making an actual decision , is to identify where my understanding is lacking , and where their assumptions or my judgment of them are incomplete or wrong .

Right? I try to figure out what I don't know . Because if you are an investor looking for the most ambitious and influential companies, you cannot know every field. Our company covers diverse fields such as biology, defense, robotics, and law . So, from the next day for weeks, I desperately organize my thoughts and try to figure out what others think about these fields and whether they are the kind of people I envision. That is my process . Then, I want to receive feedback, get a re-evaluation of the people, and understand what the core questions are .

I like taking notes. Even in the early days of the company, when I was alone, I used to write notes from start to finish in the Greylock style and send them to an investor friend I trusted to get an objective perspective . I mean funds. You and I talked about what we consider important in a partnership. I am confident in making investment decisions, but others... me I think it can help me develop further . I want others to dig into my logic and make me rethink things .

So, in the past, I used to write notes and send them to people like John Lilly or Dylan Field . The method was that I would write a note, and Bella, Pranav, or Mike would figure out what they wanted to know about that person, and together we would complete the overall picture . Other people seem to be a bit more balanced in their decision-making process, don't they? They keep thinking and thinking until they reach certainty, whereas I tend to start from there and reason backward .

However, what Mike and I have in common is that we start from a certain point and explain what drives us. So, earlier this week, my partner Bella and I met with a really interesting company. Well, I thought, "Okay, instinctively I think positively, but I don't know much about the scientific side, so I need to verify if this makes sense." I thought so . Um, so, if I do n't understand it properly, could it be an 8 or a 9? That's not true, is it? So that is why I think, 'I could be like that too.' And as we talked about , this is exactly one of the things I am worried about at this time .

If we don't have a solid intuition about the betting itself , what are we doing here? Right? Hmm , I'm not sure if what I'm thinking is fair or not . Because if Bret Taylor said, " Yeah, that's exactly what Limit means, " or did something like dog streaming, I would say, "Of course ." But he wouldn't do that . yes . yes . So, in some ways, even if I do n't understand it in my head , that person is so outstanding that I'm just going to support them .

Brett is an example of that. Yes, but I think these things are so closely intertwined in my mind . yes . Because that is precisely one of the elements that makes a person great. For example, yesterday I tried to help a recruiter from a company, and they asked, "What do you see in these people ?" They asked me that. So I said, "He's just the perfect fit ." That's what I answered. "Right? That's really useful information. He is always right. Their industrial logic is flawless. They also possess excellent character .

They are really good at recruiting talent , and in my view, they have the right perspective on the world. I have seen them repeatedly make the right decisions, even when I was wrong. I really respect that . So when people say, 'These people are amazing,' I think they are amazing because I trust their judgment. If I didn't understand what they were doing, I couldn't offer an opinion on their decisions, could I? Brett is working on enterprise AI, you know? I understand what he does . Yes.

If you were to represent your time on a pie chart... yes. Not only when I first started the company, but even now, there would be work involving meeting new companies and helping existing ones." Looking at companies, talking to researchers , spending time with other people , talking to candidates, and so on... Honestly, I do n't know exactly what I should do. Hmm, exactly what should I do? To summarize, could you tell me how you allocate your time and how much you spend on each task? I think I spend about 2/3 to 3/5 of my time on work related to my portfolio company.

Well, things like hiring, helping people organize their thoughts, trying to influence the external ecosystem, and attracting investment. The next most significant part is looking at companies . Hmm, I'm not sure if this is right or wrong . It bothers me , and I try changing the numbers , but it seems like I look at roughly 4 to 6 new companies a week . It's not that much . At my previous company, I looked at 500 companies during the first few months . So I was really surprised .

Hmm, now that I have a better grasp of it, I think I can make judgments with more confidence. Right? And I spend the remaining time helping one of my partners review a project or meeting people from whom I can learn about the world . If you are a researcher or an early-stage investor at a large company, the ecosystem is so vast that your perspective can become narrow, right? However, since we are a small company, we operate in an ecosystem-centric manner . So, I learned a great deal by meeting investors with different perspectives from mine and people who invest in various asset classes .

I had never spent this much time with investors in the public offering market before , and their way of looking at the world was really informative . Yes, right? I spend time with various investors regardless of size or asset class , and I also interact with companies. For example, I am currently meeting with a pharmaceutical company that is contemplating how AI will transform their business , and it is very interesting to see how they think about the future. I am doing various things, but I am also saving some spare time.

I just spend my time learning and satisfying my curiosity . And as you know, there are other parts that serve as a bridge to DC. I mean external communication. What are your thoughts on financing ? I stick to the beliefs I hold when advising entrepreneurs. You need to understand why people oppose what you do and their questions , but you shouldn't just say what they want to hear . Right? When I started raising funds for my first fund , it wasn't that complicated because I had already known many LPs for a long time .

However, a friend of mine who is a private equity investor said, "You need a very differentiated story for LPs ." "Well, and I think we need to clearly define exactly what we will do at each stage of the funnel. To be honest, I'm not entirely sure yet , but we need to raise funds, experiment, and figure it out . So, I did n't create slides to discuss the specifics of everything we are planning to do right now. I believed that we wouldn't know for sure until we experienced it firsthand, thought it over, and went out into the market .

However, there were definitely LPs who didn't like that approach . They asked, ' You explained your background and investment history to people in two pages, claimed to be skilled at discovering top talent, adept at helping and genuinely supporting them, and asserted that you possess both investment judgment and talent recruitment skills—don't you think that's just the starting point ?'" That's what I said. I mostly just said, " Understood." I answered . I will present a few ideas regarding company culture that I envision , put them into action, and find a solution.

" And," I understand how difficult this is from the perspective of an LP (Limited Partner) . And I sincerely respect those who have invested in us from the beginning . They sent a memo to the investment committee saying, "That person is going to work really hard. Let's wait and see." You probably think that. Right? That's the difficult part, isn't it? Sometimes, what people want is clear storytelling. I cannot give advice to other managers regarding this part . Because the results differ for each person. However, if you tell people what you are going to do and what you believe, things become much simpler.

As an investor, I get incredibly excited when someone convinces me that the world works differently than I thought . So I think that perhaps I might be able to persuade people that the world actually works this way. Over the past four years, I have met far more investment managers than venture capitalists (VCs) . I have met many people, especially those doing creative work, and you also like investment managers with a strong entrepreneurial spirit, just like me . I don't want to copy the company they founded exactly, but I think it's really cool how they approach business in a creative way, like Philip, Thomas, and Josh from Thrive .

I also think it is excellent that you encourage others to try new things and express their opinions through investment management companies . How were you both influenced by your parents, who are both entrepreneurs ? I believe that my parents were always by my side throughout my childhood and teenage years, and that they were always devoted to me despite being busy . Well, when trying to solve these things , for example, I was a pretty independent child. Or maybe that aspect just helped me. I'm not sure what type I was, but I think I was quite independent.

My family might make me feel like the center of the world, but it helped me to know that they are also people with other interests who want to spend time on other things . I think my family's values ​​are very similar to my parents' . Right? Well , there are things like honesty, thinking for yourself, independent thinking, and focus on family, teamwork, and family love. And independent thinking is probably the least common among them . Many people want to be kind, friendly, and work hard. However, for my parents, it was a moral issue .

My parents told me, "You must never care what other people think ." Well, I think I might lean a bit toward the extreme , but I do think about it sometimes . Sometimes you worry about what other people might think, right? yes . What do you want people to think ? Or is it a worry you wish people wouldn't think about? I am worried about stimulating competitiveness in the industry. Because I am a kind person . I want to be friends with everyone . Right? Well, I do n't dislike competition itself, but I do n't think the pure zero-sum competitive landscape seen in traditional Series A and Series B investment firms—where they say, "I own 18 to 25% of this company and will be on the board, and you get nothing"—helps with collaboration .

Although there are issues in terms of incentives, due to the nature of the open market, people are willing to share their best ideas , and then everyone rushes to invest . I think this is a really special situation. I like talking about how games and entertainment will go in completely different directions, but I think people are underestimating this aspect too much. I am a person with a positive mindset, so I am attracted to that direction. I am always very interested in sources of inspiration . I am curious about this from two perspectives.

Who has been the greatest inspiration to you throughout your entire life , who inspires you the most at this moment, and why ? Watching my parents start a company, I thought, 'That's really cool .' It was a dynamic where we were pitted against giant corporations . Giant corporations are trying to bring us down , but we believed that we could secure customers and achieve success through better technology and products. I consider entrepreneurship important . It is because I dream of a better future and have the belief that I can create something out of nothing .

I think it's really amazing that they can do that so quickly. I liked these things because they are similar to what I want to try . There is tremendous courage in that, isn't there ? And optimism, too. Actually, I find the negative perspective prevalent in the world these days a bit off-putting. Entrepreneurs, especially those from Generation Z and beyond, seem to think that building networks is absolutely crucial, just as if marketing and branding are everything . Of course, no one denies that point, but there are people who are too cynical about the way the world works or entrepreneurship .

It's as if they think everything is wrapped up in connections or things like Twitter . I think that idea makes no sense. If you focus on values , treat people well , work with great people , and have a valuable vision, you can succeed in far more cases than you think. I really hate the cynical view that treats everything like a game, whether it's marketing or financing . Well, I get inspiration from people like Coders. Coders is truly admirable, and the same goes for Tuhin .

Tuhin says that if you do the right thing to the customer, you can win, but I think it is a much more complex issue than that. I think Tuhin is right, and it seems to actually be effective. yes . One of the biggest controversies currently is that non-American open- source models are demonstrating competitiveness at the forefront of AI model performance . These models appear to be based on American models to some extent . How should this situation be resolved? What significance does this hold for the future of AI ?

Companies prefer open source because it allows them to develop their own models. For example, models such as base 10 are used for various inference tasks. In this situation, what do you think is right, what should happen, what will happen in the future, and what the impact will be on the business ? This is a very important and interesting question . Well, I have my perspective on what is healthy for corporations, the U.S., the ecosystem, and individuals, and there are actually things that have already happened. Well, the reality is that over the past three years, increasingly competitive open source models have emerged in all areas.

While it was mainly led by China, of course the US and Europe were also involved, and recently there are Thinky, Poolside, people waiting for Reflection, Nvidia models, Mistral, etc. We will have, and already have, very powerful open source models in Western countries . Well, Pandora has already been mentioned, and it can be said that these models are being used everywhere . Even if you do not have an economic perspective on research institutes, consider how capability diffusion might occur in the economy . Today, using models from state-of-the-art technology providers is very often too expensive, too sensitive, or too slow .

And I think that as artificial intelligence becomes capable of doing more things, these problems will become even more severe . This is because it actually costs a considerable amount of money . Therefore, I believe it is objectively true that if technology becomes more widespread, it will be utilized in a wider variety of ways . I hope that happens. It is irresponsible not to properly assess safety as the model evolves . For example, some companies are utilizing cutting-edge model technology for defensive cybersecurity or biological research .

If it is effective for these uses, it can be similarly effective for offensive purposes, namely in the fields of biological weapons or biosecurity . We must face this reality and consider what other ways there are to control it . Well, however, attempts to block technological advancement and the resulting openness can be problematic . For example, if the use of open source models is restricted in the United States, it is essentially the same as restricting law-abiding U.S. companies. It involves slowing down the company's growth , diverting profits elsewhere , or restricting specific uses .

This is because attackers or people who actually intend to exploit these technologies are not affected by these restrictions. Right? Ultimately, it amounts to restricting one's own citizens . In my opinion, thorough testing and understanding of these models are necessary . People are very concerned about vulnerabilities, such as backdoors, in Chinese models . However, what actually needs to be done is to conduct very strict safety tests on these issues . Instead of raising issues based solely on speculation without sufficient research , you should collect as much information as possible .

We are fully capable of doing so . Well, it’s not an easy task, but it is possible . As Sam said, I think a future will come where we can have extensive access to information at prices so low that there is no need to even measure it . Open source will support this. Companies want to determine their own destiny in terms of economy and capabilities . And if these models and the technology for training them, building tools, and performing tasks become increasingly democratized, it will be possible to introduce these technologies into the economy .

The economy is huge, isn't it? Every individual has various use cases that researchers at cutting-edge laboratories cannot even imagine. You cannot imagine the diversity of reality. And even if you trust the model to figure out what to do , it still has to reach that goal, right? And I think the best way for that to be realized is through a corporate ecosystem and affordable infrastructure, just as they have always existed in the economy. Have you ever considered another version of American history—a scenario where we build about a thousand nuclear power plants like the AP1000, making energy prices so low that there is no need to even measure them ?

Just like what China is doing at its nuclear power plants right now . However, due to a combination of various circumstances, we failed to seize that opportunity. And have you ever worried about such a situation in relation to information as well ? Abundant, accessible, affordable, and valuable information seems inevitable to us, but can you imagine a world without such information? Yes, of course. And I think we can easily imagine a world where we do not have such information in a competitive way. This is because it is essential for economic competitiveness and national security .

Right? For example, imagine that there is no world in the United States where the industrial base is rebuilt without automation . It doesn't make sense if we have to produce more goods and build a resilient supply chain without importing labor, when our own employees are expensive and lack some skills . Yes, then who will produce these things? Well, Americans do not want to work inhumane jobs for $13 an hour , and they should not . I do not think that becoming competitive is inevitable, and I believe we must make active decisions.

In my opinion, a likely scenario is that people are very reasonably afraid of the impact of AI on jobs , dislike the monopoly by a few tech companies , refuse to belong to a permanent lower class, and link this to anti-capitalist tendencies . Could this way of thinking slow down the development of energy, infrastructure, and industrial production capacity ? One of the most important input factors is computing power; without it, we naturally wouldn't be competitive, or at least wouldn't be able to be independent, right? I think we will be talking much more about computing independence in the future.

So this seems like a big problem . Vanta helps over 16,000 fast-growing enterprises, such as Ramp, Cursor, and Harvey, automate security and compliance to stay audit-ready 24/7 . Vanta is the best agentless cross-platform solution that now enables companies like yours to monitor risks occurring during inter-audit periods, vendors, AI tools, and the entire environment . Every new tool introduced by the team and every vendor enabling AI capabilities carries the potential for problems. Most security programs were not designed with the rapid growth rate of AI in mind .

Vanta Agents identify problems , devise solutions, and reduce vendor evaluation time by up to 50%, just like GRC engineers working in the background 24/7. Whether you are a fast- growing startup or a global enterprise, Vanta helps you earn and prove your trust. Invest like the best companies. We are offering listeners a $ 1,000 discount at vanta.com/invest. Ridgeline is the first AI-embedded end-to-end system for investment management firms that handles portfolio accounting, reconciliation, reporting, trading, and compliance on a single integrated platform . Many companies are switching from existing technology to Ridgeline because Ridgeline's AI capabilities are far ahead of any other product in the investment management software industry .

I hear a lot of talk about AI from many investment managers, and they fall into two main categories. While some people don't even know where to start , others are confident that they can build their own order management system over the weekend . However, the reality is that governance controls and a single source of data are always essential for running an investment firm, and these requirements do not change no matter how enthusiastic people are about AI . Ridgeline is built upon this very foundation , and for this reason, we are confident that the companies leading the way in the AI ​​era will be those using Ridgeline's integrated platform .

If you are seriously considering your company's AI strategy, please review Ridgeline . You can request a demo at ridgeline.ai . Then, what exactly does computing independence mean ? What are the missing elements for securing computing independence ? The most significant factor would likely be the expansion of state-of-the-art semiconductor manufacturing facilities within the United States . However, the upper class sector is also formidable . In particular, computing independence is an essential element for all components, such as certain glass materials whose supply is monopolized by a single company called TSMC.

However, if computing independence is the key , what should be done? Should I invest in a company that solves this problem? What is the essence of the problem? What is needed to achieve computing independence ? If we consider the factors required to support all use cases—such as cooling, power supply, training, and inference—in a GPU-filled data center in reverse, I believe that computing independence is ultimately very closely related to energy independence. Well, all such inputs have a global supply chain. Just as I am holding a TSMC mug right now , parts of that supply chain are located in regions that are difficult or unstable for the U.S.

and its allies to access, making it feel like it is pierced through like a thin sieve . Therefore, a completely different form of world is needed that requires significant investment, national security, and energy policy . For example, Jacob Hellberg is working on a project called Pax Silica . "Can we invest more in each part of the supply chain and find independent paths ?" This is the question. That does not mean everything has to be made in the United States . Because a comparative advantage clearly exists.

However, securing multiple supply sources is a position that every company desires. If we look at the fields we have invested in , we have invested in solving data center workforce shortages , robotics, nuclear energy, and alternative chip architectures . We are continuously keeping an eye on companies such as data center construction companies, solar installers, and solar and battery installers . This is because the core of this project is building actual capacity . This is a highly operational business where operations, technology, and real estate work in combination.

Of course, financing is also important. Funding will likely be the most important factor . We have not invested in this field yet, but we have been reviewing it very closely . It is not that I am against it, but fundamentally, it is because I am a technology investor . I would like to understand what durable product assets companies are building . What is the biggest controversy within Conviction? You have a really interesting team . My partner Mike is an amazing engineer with truly outstanding technical talent .

It seems like you present a lot of interesting perspectives because there are many young talents in the company . It seems like there will be lively discussions about important issues. What are some of the major topics of debate these days? I host a podcast called 'No Priors,' and the premise that what we used to believe about the market is no longer true is interesting. We also regularly look into companies in fields other than traditional software . The question is whether it is really possible to make a profit in a market like this .

For example, venture investment in semiconductor companies has long been a really bad business, with the exception of Lip-Bu-Tan . The rate of return was really bad. Yes, that was really the case. However, our team partner Bella looked into several of these companies and found that there is clearly demand. So now I think we have all reached the same conclusion . In other words, today's market is different from the past. We are witnessing a large-scale demand for integration regarding accelerators or supply chain independence . This is because major buyers also want it.

We cannot all rely solely on one of TSMC's production lines . TSMC is highly valuable , and this changes the company's risk equation. Therefore, we start with a fairly consistent belief about the direction we should take or the problems we need to solve. And much debate is whether the market is favorable to venture capital firms . Even if that is not the case, the space market is not a favorable market for venture capital firms , but is there a possibility ? And considering the distribution of the results, is it worth investing in ?

There may also be possibilities in the fields of solar power, batteries, nuclear power, turbine manufacturing, robotics , and biology . The software market isn't as popular as it used to be, like it was 10 years ago . yes . Well, there are new topics of debate in each field . For example, biology is an interesting field , and as I have experienced the data firsthand, I have now firmly taken a side. Which one, you ask? You can create and secure tremendous value through biological models .

Various AI software can also emerge in the field of biology. We made our first investment in a company called Chai Discovery . Chai plays an important role in accelerating parts of the research and development process by collaborating with several top 10 pharmaceutical companies . I have looked into various types of computational biology companies for over five years , and the conventional wisdom was that the only way to make money in the biotechnology or pharmaceutical industry was to develop drugs . After developing the drug and attracting biotech investment, you decide how much risk you are willing to take and what that means for the company's capital structure .

Traditional biotech companies discover research leaders , have them own 40% of the company's shares , and develop new drug candidates by combining various factors . But most of them fail. Therefore, the investment method and result distribution are completely different. For example, software investors cannot make money by selling software to pharmaceutical companies or building platform businesses in the pharmaceutical industry . So the debate is whether the situation will change if the model changes . I think it will definitely change. Of course, regulatory issues must be resolved , and safety concerns are not easily overcome in a rapidly changing world, but I believe that the development of treatments will accelerate tremendously .

And some of the value creation will come from young companies , and we have high expectations for investing in such companies . What changed your mind ? What was the turning point that changed from 'I'm not sure yet' to 'I'm certain of this' ? Well, we invested even when we didn't know much yet. I understand. yes . Well, I think that's exactly part of a fun adventure . We thought, 'It is possible , and worth a try .' yes. There are no geniuses here. I thought, ' This is a $10 million contract .' And another important point is that we spoke with the client's scientists or department heads—that is, tool users—or with people related to the tools adopted by end users, specifically the product-driven growth tools that are proving effective in their respective fields .

I thought, 'The customer knows best whether the tool is valuable or not.' Well, I think the moment a kind of 'blind idea' comes to mind in the pharmaceutical industry is when AI leads to the emergence of completely new indications or new drugs, thereby changing the very flow of the industry . Then I think a massive wave of investment will occur, and naturally, it should. I believe the pharmaceutical industry and the overall healthcare industry are rapidly recognizing the potential of AI, thinking, 'This will definitely bring about change.' I was really impressed by the speed at which you said, "I'm going to completely change the business." Why do I call that certainty ?

That is exactly an 'ideal ambition' . Right? The most traditional method of early-stage investment is to invest in a company early to secure a significant stake, never sell that stake, and continue investing until the company succeeds, is sold, or disappears, right? I think that method is very simple and creates a sense of shared understanding . Well, Mike and I have both been through the company's growth process , and we have experienced times when things did n't go well at first or the chances of success were unclear .

For example, companies like Figma, Notion, and Rippling . You can't invest expecting it will take 4 to 5 years for those companies to succeed, can you ? However, I believe that everyone grows through their own experiences, specifically their investment experiences. Likewise, the first few years of Base 10 were also unclear. Therefore, I think it is important to have a perspective that is not easily apparent in the market . Although others may not possess it due to various reasons such as market conditions or the backgrounds of the members , I believe it is important to put doubts aside and act with conviction .

That is exactly how it can serve as an excellent example for partners building a business. Well, I am trying to build a partnership , and I firmly believe that investment is a team sport. However, while talking to a friend who runs another investment company, I heard him say, "We do not have individual ownership of the investments ." I thought, "That's ridiculous. How can you run a business like that? " Someone has to have the power to make decisions, right ? So, I don't know if this is right or wrong, but "Patrick, you have to make a decision.

Please show us your conviction. Can you persuade us? How can we help you ?" I cannot think of any other investment method other than saying that . Therefore, I think it is best to concentrate all information on one person . great. What did you learn about taking risks ? It seems that behind certainty lies a kind of leap of faith . If you knew everything and everything was clear, it would naturally be reflected in the price, so there would be no opportunity to make a profit .

yes . In that case, is a leap of faith always necessary? Is that risk-taking in another sense ? You have already made a significant investment and have extensive experience in this field. Unlike my close friends at Founders Fund, I do not have an instinct to think differently from others . However, I think it is very important to firmly hold onto your own thoughts without paying too much attention to what others think . Right? Here, 'other people' refers to the dominant discourse of the era or the opinions of various entities that play an important role within the ecosystem.

I just think it is important to find the truth . To explain this in relation to risk-taking, if you discover the truth but the price is incorrectly set, sticking to that price puts you in an advantageous position . You need asymmetric information and the confidence to stick to opinions that others do not yet agree with . So I think a lot about how I can secure better information than others and protect myself from unnecessary noise. Could you explain it in more detail? That is a good way to ask what certainty looks like today.

You are creating a shell called certainty. What is the secret to doing these two things well ? I mean telling the truth and protecting yourself from noise . Yes, I want to spend time learning from the people who actually make the world . Just like the founders of portfolio companies or many founders who do amazing things outside of their portfolios . They are really smart people who surprise me, pioneer new fields , or believe in things I couldn't even imagine . Right? I think these three are in different categories.

Let me give you an example . Mikey Shulman of Suno is building an amazing business in the field of music production. But I stupidly refused. Mikey's mutual friend and investor asked me to invest, but I just refused. I was wondering how many people would want to make music . I do make music, though. Hmm, but could that transform into a social network ? How much will be consumed? There are too many questions, and I think my intuition is wrong. Actually, I think I underestimated the amount of creative activity and things that AI tools can express and enjoy.

I learned something while talking about Mikey, his business, and the things people are trying to do . So, everything is really informative—founders running companies that create behavioral patterns we cannot understand, people conducting research in interesting directions, companies presenting plans for AI, and so on. It seems that the circular logic regarding how people think about the strategies of large research institutes today and how their applications can survive is not very helpful for decision-making. In my opinion, every organization has a few key priorities. Let's assume that the priority of OpenAI, Anthropic, and DeepMind is AGI ( General Artificial Intelligence) .

Or should I call it ASI (Artificial Intelligence )? In a direction that allows for safe profit generation . The next steps would include ChatGPT (virtual reality programming), advertising, coding, etc. And after that, it could be expanded into collaboration, that is, a feature that allows various users to perform richer tasks on more diverse interfaces . I believe we need to assess the competitiveness and realistic scope of each project and compare it with the companies or opportunities we intend to invest in . I don't think it's very helpful to come up with a grandiose strategic framework about which field will win .

It seems many investors pour their energy into that aspect, but I would rather ask myself, "If we have entered about 1% of the market now , how can we expand to the remaining 99%?" I think we need to focus on that. Well, so I think that is exactly the kind of place we intend to pour our energy into. Now, to wrap up, just for fun— this is purely speculation and not a serious prediction— but a question asked just for amusement: what do you think will happen one year from now?

When you put together the amazing people you hang out with , the research community, and entrepreneurs like the Sunday Founders, the world is really changing fast . One year is a long time, and it feels like the age of a dog turned upside down. What do you think will be the most noticeable change in the tech world one year from now? I expect to see the Jevons paradox actually realized in a year. Because agents and products will emerge that handle daily tasks more efficiently in every area of ​​our lives .

Just like the changes that have occurred in software engineering. Right? Among the companies I know, there are some that say, "We are growing much faster," and I expect that kind of analogy will happen in all other fields as well. Right? We can see that at our company as well; for example, one of our portfolio companies has only one and a half employees in its marketing department. It's a great company. This company has many customers , and it also has to carry out traditional marketing activities like creating sales support content .

However, the marketing manager was very interested in expanding their influence , so they created an autonomous marketing department within the company. So, if you learn faster in all fields and reduce repetitive tasks, you can somehow utilize that time. Thanks to AI, has productivity increased, so have your working hours actually decreased ? Yes, I ended up working more . I think this is exactly Jensen's core wisdom. We will all be doing more work in the future , and to do so, we need to ensure that people have access to the necessary tools and receive education .

You really did an amazing job. It was really great getting to know your world thanks to Colossus . I think it's really unique . I think you are a great example showing that there is always room for excellence . When you started Conviction, there were many early-stage investment firms, but now, four years later, if you ask the people you worked with, they will say that you brought a really big change to their lives . There is always room for excellence . I think it is a really great lesson, especially when considering how we approached the initial investment process with a "we will just dive in" attitude instead of explaining "how we differentiate ourselves at every stage." In the end, I was so frustrated that I even thought, " This is a game of execution ." That's really true .

I completely agree. Well, I ask everyone a traditional closing question. What was the kindest thing someone did for you? I really like this question. I think there are many people in Silicon Valley who have achieved tremendous success , but they do n't seem to care much about their background . Uh, they had a conversation with you and asked, "Could you possibly help me?" Or, "It seems you have an interesting idea ." Or, "You look promising ." The moment they say that, the most important factor determining their willingness to invest in a relationship or person is precisely their evaluation of the idea and the person.

It is truly amazing. Most ecosystems don't work that way . So there were people like Asim Chandna, Anil Bus Lee, Joseph Ansanelli, and Reid Hoffman, who hired me when I joined Grey Rock at the age of 23 . People mocked young venture capitalists, saying, "They're like barnacles clinging to the ecosystem. It's a terrible experience for entrepreneurs. They try to give advice to others when they know nothing . Who gave that kid money?" They often say that . So I said, " Well, first of all, my job at the time was to help others succeed.

And it was about learning. You can do any job or any task well by imitating and thinking based on fundamental principles, right ? That is why our company hires people with little experience, and back then, I thought, 'Wow, thank you so much for giving an opportunity to a stranger like me and taking the time to teach me how to become an investor.' " He said that. Among the people who gave me advice when I started the company, there were a few who would have considered this kind of story very trivial .

It's just my opinion. He said that. For example, people like Ravi Gupta, who is now the co-CEO of a new company called Ithaca, Dylan Field and Elena Natalinskaya, and longtime partner and friend John Lilly . You can definitely do it, right? There were a few people who told me that . Anyway, I tried to do it, but the encouragement from people who believed I could become great gave me great strength. Especially including our early LP members. I will be forever grateful to those who took risks with me .

It is a truly beautiful thing. The world runs on faith, you know. It means confirming someone's ability even without evidence yet. That's a really wonderful thing. Yes, and I think that is exactly faith in people . We've talked a lot about how people's thoughts and our opinions on them are intertwined . But I think that is beautiful. After all, you don't need any special advantage to come up with ideas . yes. So I am really grateful for the fact that people appreciate my idea and trust us.

I learned a lot by watching you work and talking with you. It was a really fun time . Thank you for the invitation. You know that small benefits accumulate over time, right? This applies to investment as well as company operations . The expenditure system is a capital allocation strategy . Ramp basically makes this system smarter. It provides better data, better decision-making, and better long-term economic efficiency. Learn more at ramp.com/invest . As your business grows, Vanta scales with you to automate compliance and provide a single source of information on security and risk.

Learn more at vanta.com/invest . Leading AI and software companies such as OpenAI, Cursor, and Perplexity use WorkOS to leap to enterprise-level in an instant, rather than in months . Visit workos.com to skip the hassle of building infrastructure and focus on product development . Ridgeline is redefining asset management technology as a true partner, not just a software vendor . Ridgeline has helped businesses scale fivefold to achieve faster growth, smarter operations , and a competitive advantage . Visit ridgelineapps.com to see what benefits we can offer your company.

Every investment company is unique, and generic AI cannot understand your company's processes . Rogo is an AI platform specialized for Wall Street that connects to your data, understands processes, and delivers tangible results . Check it out at rogo.ai/invest .

Go deeper

• All Invest Like The Best episode guides