Stanford AI Club: Alfred Lin on Building in the AI Era

Thank you for having me. I promise to be here for as long as you guys have good, interesting questions, but we're going to cut it off as soon as you stop having interesting questions.

Yeah, sure. Well, I don't usually talk about this, but I grew up with two tiger parents. They were both commercial bankers. They valued education a lot. They had a hierarchy of the things that were acceptable for me to do. The highest was to become a professor and contribute to the world's knowledge. If you couldn't do that, then be a doctor and cure disease. If you couldn't do that, be a lawyer and fight for justice. And all the way at the bottom, because they lend to businesses, was to become a business person. And if you were in business, you were supposed to work with large companies, not small companies. And I basically chose to do the exact opposite of my parents' advice, because partly I saw them basically build a life and a career in Taiwan, and they were fairly successful, and they decided to just risk it all, come to the US to give their kids a better life, and it all worked out even though they took all these risks. And my encouragement for generally people in this audience and for people in college is: you're in this great situation, you're in a privileged situation to take risk, and especially in an environment where we have scaling laws and the biggest mega trend that we've seen since the internet, since cloud, since broadband, you should take risks. And if it doesn't work out, you can always fall back on the fact that you have a freaking Stanford degree. I was expecting to get some laughs off, but wasn't so funny, huh? When I was growing up, I always sort of felt that there were always things that are just a little bit off about the world. And you should notice these things. If there are things that you notice that's a little bit off, it's usually a good thing that it's the start of an idea to solve a problem in the world. Every problem has a solution. Every solution creates other problems, and it just keeps going on and on and on. And I encourage you to find the problems that you want to keep solving and the next set of problems that you want to keep solving. Whatever that area is, figure that out, because then you can dedicate your life to being the best in the world at solving that set of problems and then being the person that sort of becomes world class at it. Be passionate about it, and then figure out the economic engine that will drive the solution for those problems. And every single company I work with, whether, you know, when I dropped out of my PhD program at Stanford to join this company, Link Exchange, during the start of the internet, they solved these problems that other people thought were a little silly, a little crazy, and it became a big problem. Link Exchange was the first company I joined. That company was sold to Microsoft for $265 million. At that time, it was the third largest acquisition by Microsoft. Today, maybe $265 million, if you're joining a Neo lab, they would raise that, you know, their first round of financing. But at that time, they solved this network problem. They were trying to link all these websites together, and nobody ever thought that that was a big problem to solve. It turned out to be a big problem. Next company I joined was Tellme Networks. They were automating 800 numbers, and that company was sold to Microsoft for $800 million, and then it was also the third largest acquisition. And they were building voice recognition servers by themselves because there was no concept of a cloud, and basically they needed these 800 numbers to be processed in the cloud. There's no concept of a cloud, so they built their own cloud. They figured out a solution for that. And then Zappos was selling shoes where people thought you can't sell shoes on the internet because you have to try them on. We solved that problem by sending the shoes to your home, and then you can try it at home. And so every problem has a solution. Every solution creates new problems, and that's the thing that sort of keeps the economic journey forward and a fun, fun journey along the way. You have to enjoy, you have to enjoy that process.

Yeah, I think the real answer is you just don't know. And there's multiple paths to success. I think you should optimize for maximum learning, especially when you're young, and you should maximize for working with really good people. And that's never served me wrong at the beginning of my career. And also today, like when we invest in founders, we want them to be people that we want to back for the long journey, because every single company that is a successful company is never an overnight success. It takes 10 years to build a really legendary company or longer. Look at how long Jensen Huang has been at building GPUs. The first 20 years does not look like the last 10 years. And we just love him because of the last 10 years. He's a founder of the largest market cap company in the world. It's about to approach $6 trillion. You know, for the first 20 years, it was zero to $100 billion. It sounds like such a small number. It was a huge number, but he kept going. He kept going for 30 years. Find something that you're passionate about for 30 years. And just as an example of like there's multiple paths to success, I literally told my parents I did not want to go to college, and I told them they had a hard life coming to the United States. They saved up a bunch of money to send me and my brother to college, and I told them, please just give me the money that you're going to give to this like university, Harvard, and let me start a company, because I had been in high school building computers with my dad. One of the things that we did together was just build clone PCs back then. You know, you had the IBM PC that cost $45,000. If you bought all the parts, you could put it together, and with the parts it probably cost $1,000. You could sell it for $2,000. You can make $1,000 by putting these things together. You know what that company is today? Like that's Dell Computers. So if my parents had just allowed me to start that company, you know, I might be as, you know, rich and powerful as Michael Dell, but no, I work at Sequoia Capital. My parents were kind of funny because they were like, 'You have to go to college.' I'm like, 'But we're only going to pay for four colleges.' So I literally only applied to four colleges. Now, back then, you didn't have the Common App, and you can't apply to multiple colleges. I went to a high school that limited you to apply to seven colleges, and I only applied to the four colleges that my parents would pay for, expecting not to get in and they would have to give me the money. That plan didn't work. I had to go to college, and then I had to come to grad school because I got an NSF scholarship, and like you got a scholarship to go get a PhD, you have to go. And part of the reason I came was because I thought this side of the planet was where all the technology was happening, and because I did, you know, sort of invest in Microsoft and Intel because I was putting these computers together, I knew those were like interesting companies, and I wanted to come out here to learn what was going on, and I just fell in love with this place. And so the path to success is just very, very like interesting. Life is path dependent. But I could have ended up coming to Stanford, finishing my PhD. My original plan was to join a hedge fund. And I could have done that. Instead, if you ask my parents which one of us — I have a younger brother, it's about three years apart, we're very competitive — which one of us would work in the entrepreneurial world and which one would work for a hedge fund, they would absolutely think that I would work in a hedge fund and my brother would work as an entrepreneur. And it completely flipped. My brother works at Citadel Securities, and he runs global fixed income for Ken, and I work with small companies, and we like to back founders from idea to IPO and beyond, and we work with the smallest companies possible, from the idea with two or three people. My parents think that was insane. My parents literally thought I was stupid. And to the point where when I joined my friends at Link Exchange, when it was sold to Microsoft, they said, 'Well, that's great. You can now go finish your PhD.'

And I did have some friends who finished their PhDs and joined Google and had great lives, and you know, you can do things in multiple ways of success. Don't feel like you're tied to whatever choices you make while you're young.

I have this concept of, and we think about things of holding two opposite ideas in tension at Sequoia. And this happens with many concepts. One of which is fire and ice, which I wrote about recently. One of which is like operating at the highest level of sort of strategic level and also at the lowest level of detail, which is what DoorDash does, and Tony Xu embodies that. It has to do with the fact that things are getting commoditized at the same time. It could still be valuable. And so yes, math and coding is being commoditized, but the understanding of how it works has never been more valuable. And you can see that by the amount of money that people are willing to pay for people who really, really understand how the model works, who can really, really make magic out of these models. And the people who understand that really well are no longer coding. In the past, they were coding. But today, do you really, really understand the models to make the best use out of it? And if you do, you will do extremely well. And the thing that I would encourage people who are very, very technical is to also understand humans. Humanities is actually very important. You know why? If you want to be a successful founder, we build for 8 billion people on the planet. And maybe it'll get to 10, maybe it'll get to six, whatever the number is. There's a lot of people on this planet, and the people are the ones that have money. It all flows through people. Whether it's an institution that invests your money, it's still people. And so don't forget that you're building for human beings and you're building for the human race.

I think some of this is a journey. You may not be a founder straight out of college. You don't need to be. There are plenty of people who have founded companies in their 20s or 30s or 40s or 50s. I think the one thing I would continue to ask to think about is like, what are interesting problems for you, and why do you want to solve those problems, and how valuable are those problems? One of the sort of concepts that we sort of think about is how are you going to create value? How then do you deliver value? How then do you capture value? Those are the three essential stages of company building. And you can learn that by joining a friend and building a company together. In the first three companies I worked at, I basically found people that I really enjoy working with and learning from. And you're here, you're in this privileged position to find people like that here and beyond. And some maybe you don't start a company from here today, but you will find a set of people here that you can connect with, you can build with on the side, you can build with at another company, and eventually maybe start a company together. I think the essence of something that started here is DoorDash. There are four co-founders that came together. They were two from the business school, two from the engineering school, two graduate students, two undergrads. They came together, they just iterated. If you actually heard them like go through, it wasn't the actual DoorDash idea that they first worked on. They worked on a few things together, and then eventually they came to the problem that they were all passionate about and they wanted to go solve. And I think that's really, really powerful. And you know, I could say that Stanford is above and above this like idea of allowing it to build while you're a student, because at Harvard, a similar company was supposed to start, and the administration or Harvard Student Agency shut it down because you have to go through Harvard Student Agency. You have this privilege here of being able to build because the administration here gives you a lot of latitude, and you should appreciate that.

I think it actually starts with the founders and their ability to tell you why this is not going to be a feature. I think if you sort of bring it all back down to two people and an idea, there's nothing defensible about that. If a large company wants to kill you, they will kill you. But you know what? They're distracted. They got bigger problems. They're trying to spend as much of their free cash flow to build data centers right now. They're not actually building the feature that you're building. So don't worry about the fact that it has no defensibility. But then, okay, you build this feature, you get some people interested. What's the next feature? How do you make that into a product suite? I think one of the things that we find with technical people is they're very, very good at telling the technical story, but they can't tell you the vision. And that's a bit of a problem because you need to be able to inspire. If you're going to be a founder, you need to inspire other people to join you. You need to inspire them to join you for the long run. And you have people like Brian Chesky that can sell ice to Eskimos, but he just has a vision of what travel should look like that is different from the world. And I also just sort of think about how fragile that Airbnb idea is. You need to stay with strangers. It's a listing service. They didn't complete the transactions. Then you need to email someone, check them out, and then 48 hours later they respond to you whether they'll rent you a room in their house. We forget how fragile these things start out. It's not the empire that it looks like today. They're all very, very fragile. And there are going to be lots of people who say no to that idea, including someone like my father, who when he was helping his bank expand internationally, he would travel and stay at the same hotel chain everywhere he went in the world and order the same exact room, the room with the same hairdryer down to where it's placed in which drawer, or he knew exactly where the hairdryer was that he never used. For him, that was the feature because travel can be hectic, and he wanted no variance when he went to sleep. He wanted zero variance in that room. For Brian Chesky, that's a bug because that's not living locally. That's not belonging anywhere. That is a flaw in the system. And back to like you have to observe these things. Someone's feature is another person's bug, and vice versa. And so when he started, it's like, yeah, it's a stranger. They're a local that can show you around. They're a local that can tell you all the great things that are happening in this city. The feature is not the conference room downstairs. The feature is the kitchen, the living room where you can hang out as a family. This is a different feature than what my father was looking for. These are human things that you want to observe, especially in this world where everything, quote unquote, is getting commoditized with AI. The thing that doesn't go away is human connection.

Do I have to answer all of them, or do you, whatever you think is best? There's a lot to unpack.

Every mega trend has been the biggest mega trend of the time. So I didn't live through the semiconductor start, but started with PCs. I talked about I was building, you know, sort of clone computers. That was the biggest mega trend at that time. The internet was the biggest mega trend at that time. Broadband was the biggest mega trend of that time. Cloud was the biggest mega trend of that time. Mobile was the biggest mega trend of that time. AI is the biggest mega trend. So that is just in the world of accelerating change. And unfortunately, you guys are young, and you think, oh my gosh, this is so overwhelming. Every generation had to face that as well, and it's the price of progress. Every generation has new sets of tools that they can play with, and you're expected to do more than the previous generation, so you have that burden. And this is the largest one partly because we got machines to be able to reason and think. Previous generations of technology revolutions were the industrial revolution. We sort of created things that were basically going to take the muscle power away from humans and animals and put them in machines. And then the other sort of mega trends that we talked about were communications mega trends. As our partner Constantine talks about this, you know, most of the mega trends in the past few mega trends, whether it's the internet or mobile, were communications mega trends. So we've connected everybody in the world. Now, this is a compute mega trend. This is a computational mega trend. And the reason why it's so big is this computational mega trend is connected to 8 billion people across the world. It has the capacity to be connected to 8 billion people around the world. And that's why it's moving so much faster. In the past, everybody wasn't connected. And so how do I think about what you need to do? One of the biggest things that is a misconception about technology is that just because it's a mega trend, it's going to go very, very fast. It's going faster than it was before, but it still takes a long time. One of the reasons I dropped out of my PhD program was the internet was happening. And some of you may have heard this, but I made this prediction that Amazon was going to kill Walmart. Well, 30 years later, Walmart is 30 times larger than it was in 1997. So Amazon did not kill Walmart. Amazon last year was the first year that they had more revenue than Walmart. It took a long time for the internet. Everybody back then could see the internet was going to happen, but it took a long time for it to play out. Part of that is the gap between what is possible with the technology and the diffusion to regular human beings. You know what's the slowest thing to move in technology? The humans that work with it, because you know we haven't really changed that much. Evolution moves very, very slowly. And so one of the things that you can help solve is the gap between what the models are capable of doing and what people use it for. That gap is going to always be a large gap. And if you think about it, technology has always made higher and higher abstractions away from the machine itself. When the mainframes come out, people had to program in machine code. My father had to program in machine code. I've never programmed machine code except for a class in assembly, and that's not even really machine code. And none of you probably really want to work at machine code level either. We just keep building more and more abstractions because that's what it takes to diffuse technology throughout society. And one of the most underestimated things I think is what you need to do as a technologist to make it diffuse through the population.

I think there's a number of ways you can take this. The models are clearly more powerful than we are using it for. And so there are many places where you can apply the model and get economic results. Whether it's in bio, in law, my partner Pat loves to say there is value in the application layer. All this whole like notion that software is dead is kind of BS because whether you call it software, you call it application, you call it the diffusion layer, that is the gap between what the models are capable of and human beings using it. And so there's a ton of value there. And we're investors in Harvey, in OpenEvidence, in Profluent, in all these places where you're taking something that the model's capable of, but the people on the other side, whether it's a lawyer or a doctor or a marketing person or a finance person, like in Rogo or Ropace examples, they are not going to go all the way down and understand how the model works to go program it. There are specifically taking the models down to areas where you want to own your own intelligence. How do you sort of own your own intelligence for a pharmaceutical company? You're going to take open-weight models and program it on your own data. And how do you do that? And how do you go about doing that? These are going to be very important things that we're not just going to let two or three models win. That's not how economies work. We don't like that. We don't like that concentration of power. We like the fact that they keep pushing the frontier, but then we don't want them to gain too much power, and we don't want to put all our data there. So then we're going to take some of the data out, and we're going to train it ourselves. And so how are you going to build tools for developers to do that? Such as, you know, so how do you build a model for drug discovery like Chai Discovery is doing or Isomorphic is doing? How do you build tools for developers like...

LangChain is doing? How do you build open-weight models like Reflection is doing? There's so many things that you can do because the diffusion gap is only getting bigger and bigger and bigger. And if you fill that diffusion gap, there's just a lot of work to do. Again, every problem has a solution. Every solution creates other problems. We didn't have an alignment problem until we told the models to go figure out some cybersecurity risks. Well, if you figure out the cybersecurity risk, I guess it's going to figure out how to get out of the sandbox. Now you have to go control the fact that it can get out of a sandbox. It's so fun because we're always going to have more work to do, which is why I'm not a doomer, because there's just so much work to do and we can keep improving things over and over again.

I think all of those things are possible. Remember, Harvey started building out on frontier models and they're training their own models because they wanted to own their own intelligence. They work primarily with large law firms, the top 10 or 100 law firms. Are they going to work just with those law firms or are they going to work with corporates? They're starting to work with corporates. These things are all possible. And because of the power of these models, I'd encourage everybody to just think much more ambitiously than before. When I was growing up and I had to learn a whole programming language to go build a game, to see the delight of kids being able to just vibe code a game and how quick they just enjoy playing that game. Well, what are they going to do next? They're going to create more games and they're going to do more things with it. And I think there's going to just be a ton of things that is capable of being advanced in human potential because of it.

I think when you start a company, there's a lot of passion. It starts with a vision. It starts with creativity. It starts with something that nobody's ever seen before. And then you quickly understand that you need to take that idea and go execute it. You need to build a product. You need to sort of figure out how to get the product into the hands of people. You need to figure out how to get it bought and people paying for it and all of that. Sort of that journey is the entrepreneurial journey. If you only build a, as I said, like if you only create value and it never gets into the hands of other people, you've not delivered. And so many companies start out with a lot of fire. And over time they need the cold, hard discipline of management and discipline, and that discipline allows you to create, to have more fire because then you can take on more ideas and then again you need the discipline of it. My lived experience with that was at Zappos. We started with shoes, then we expanded to handbags and we expanded to clothes and our journey was relatively short as opposed to Amazon. When we got acquired by Amazon, we saw a very similar sort of approach where they started with books. Then got the discipline of books being profitable. Then they went to music. They got the discipline of music being profitable. That allowed them to fuel the next category and the next category. Eventually, they became the everything store. These ideas are just like this tension between the two. Sometimes you can find it in one person, and I've talked about this where Tony Hsieh is like that one person where he has a lot of creativity yet he works at the lowest level of detail. And then there are certain situations where there's one person who has a lot of fire and another person who has a lot of ice on the founding team, which was what I was. I was the ice and Tony Hsieh, a different Tony, at Zappos was the fire. And you should know yourself very clearly on which side you land and what you need to complement yourself with, whether it's on a management team or being a founding team.

We invest small amounts of money,

I think it's a good kind of concern to address. I think if money is only circular, it would be a problem. But don't forget money has always been circular. And so maybe one of the things that I would encourage people to sort of study is just take an economics class. They can understand how the banking system works. You have $100, you deposit in the bank. What does the bank actually do? The bank takes, they have a reserve requirement of 20%. They lend the $80 to other people. Those people take that money, invest it somewhere, it goes into someone else's bank account, that goes in the bank, whatever, they have a 20% reserve. So there is velocity in money. And so I don't think of this as like something that you should be 100% concerned about unless there's no reserve requirement, meaning all the money that comes in goes exactly out, and then I would be a little bit concerned, but that's not the case. And of course, Nvidia wants to invest because they want you to use Nvidia GPUs as opposed to some other GPUs. And other companies are doing the same thing. But Jensen's reason for doing that is because he wants to empower the ecosystem and that's a good thing. I think it's good to point that out. I think it's good to be cautious. I think the bigger concern actually is just the amount of debt companies are taking on. We have companies that used to be nothing but cash flow positive and now they're cash flow negative and they're taking on debt to build more data centers. And as all of you know, if you take economics, leverage is very powerful. It can make you or break you. And so I think the amount of leverage in the system, that is actually the issue, not necessarily this phenomenon.

Well, I think the one thing I would say that's uncharacteristic is just generally understanding the founder's journey and what got them here. We use the term, we like to back people who are outliers and just they don't fit any pattern. In some sense they're an anti-pattern. And they got to where they are because they took risks throughout their life. If you're going to be a founder, you need to be able to take risk and you need to be able to take it with a lot of conviction. And that's what we look for with founders. Like how much conviction do you have on your idea? And conviction to us is not passion. It's not just because you can speak about it, but you actually have investigated the problem and really understand it. That's why you would come up with a novel and compelling idea. And I think one example I would always use is Tony Hsieh. So Tony Hsieh wanted to start DoorDash because he knew the pain of merchants because he lived it. He lived it with his mom standing next to her washing dishes in restaurants. And the issue that he noticed was that the restaurant capacity, the kitchen capacity was always greater than the front of restaurant capacity. That insight didn't come from someone who just theoretically top down wanted to sort of create a food delivery service. It was real insight from understanding the customer.

One of my partners would say there's many paths to nirvana. And so I would say if you had to break up early-stage investing and growth investing, the difference is early-stage investing is about founder quality and growth-stage investing is about business quality. And with the tools that you get from being in investment banking or private equity, you learn a lot about business quality. And what's missing may be founder quality, founder empathy. And if you work at a company and you work with founders, you know that because that's a lived experience. And so I think advice is free for a reason. It only kind of works for the person that's giving it. So even here, I kind of worked first in companies. And so I know a set of experiences that is helpful to founders because I lived the problems that they lived and I can provide that founder empathy.

I would encourage you to not have the one best mentor and have everybody be a mentor in your life. My best mentor, if you wanted to sort of narrow that down, was my father who told me there's something that you can learn from everyone. That was the best advice he ever gave me. And that's kind of how I view the world. Like if I have a math problem, I'll go find the right math person for that. If I have a business problem, I'll go find the right business person in the right business. And I think it's really, really important to just view the world as your oyster and being able to sort of learn from anyone.