{"url": "https://www.youtube.com/watch?v=r7HyWFJMAxg", "page_url": "https://www.youtube.com/watch?v=r7HyWFJMAxg", "title": "How to Start a Hard Tech Startup with Sam Altman @ MIT", "show_or_channel": "Vahid Fazel-Rezai", "date": "2016-04-13", "duration_s": 2820, "type": "talk", "speakers": ["Sam Altman"], "confidence": "high", "notes": "MIT hard tech startups", "people": ["sam-altman"], "id": "2016-04-13_vahid-fazel-rezai-how-to-start-a-hard-tech-startup-with-sa", "audio_duration_s": 2820.237625, "worker": "jc-w23", "transcribe_s": 127.1, "model": "large-v3", "segments": [{"start": 1.36, "end": 30.78, "text": "Okay, so thank you for coming out. We have our distinguished guest here, Sam Altman. He's going to talk to us about some hard technology, and I'll let him define that for you guys. Before we get started, I'd just like to thank Sandbox and the Sandbox Fund and the Martin Trust Center for Entrepreneurship. Those are both resources that you guys should definitely check out because they're supporting basically the entrepreneurship ecosystem here."}, {"start": 30.78, "end": 60.78, "text": "Another really cool thing that goes on every Saturday, InCube gets together and basically we hack together on like a ton of different projects. Right now we have about a dozen projects ongoing across a wide range of areas and it's just a lot of fun and it's incredibly productive to come together, have a somewhat social environment, have some tasty food, and build really cool stuff. So you guys should check that out too. And without further ado, I'll"}, {"start": 60.78, "end": 99.14, "text": "Hand it off to Sam. Thank you very much for having me. I will try to make it worthwhile. So Y Combinator has been around for 11 years. We have funded, I think, almost 1,200 startups. In aggregate, they're worth $75 billion, maybe, at this point. We've been very fortunate to be part of some great successes, including some from MIT, I think most notably Dropbox. I think MIT is actually the only school that I've been to every year to come speak at since I've been at YC."}, {"start": 99.14, "end": 125.87, "text": "We really like it here. Recently, in the last couple of years, we have started funding companies that are not only software companies. So we first started with hardware, and then we started funding biotech companies. And we've realized that, for lack of a better term, we call these hard tech companies. Basically, anything that uses atoms as well as bits, and anything that usually requires a technical breakthrough, often but not always, lots of capital and time."}, {"start": 125.87, "end": 154.67, "text": "And we realized that most people weren't funding these companies and that they were really, really important to the world. So what I'm going to try to talk about today is how to start these companies and how we can help you and why you should start these companies. There's a lot of other information about how to start a software startup. So I will sort of incorporate that all by reference. We have put a lot of it online. You can find it on the Y Combinator website. And I'm going to try to talk specifically about what you need to do differently."}, {"start": 154.67, "end": 185.01, "text": "If you're starting a company that is not only a software company. And the main answer, before I get into anything else, is to try to do as little differently as possible. The mistake that most people make is if they're going to start a nuclear reactor company. In fact, we funded a nuclear reactor company from IT. But the mistake that most people make is they try to say, well, I'm different. Because I'm special, because I'm saving the world, the normal rules of a startup don't apply to me. And so I'm going to take a long time."}, {"start": 185.01, "end": 213.33, "text": "I'm going to use a lot of capital. I'm not going to worry about finding a customer. I'm going to go sit in the lab for five years and not talk to anyone because I don't like talking to people and I want to just work on my technology. And this is always been a mistake. And in fact, when we started funding these sort of hard tech companies, we very briefly tried letting them be separate. So they were sort of their own group within YC and they would talk to each other and they would do things on their own way and their own time skills."}, {"start": 213.74, "end": 242.37, "text": "And we realized that was really bad. The problem that most hard tech companies make is that they don't look enough like the software companies that have been so successful in the last couple of decades. Not that they look too much like that. So the most important change, I think, that we made at YC for companies like this is to say that, hey, we don't care what you're doing. You are doing a synthetic biology company. You are doing a mobile photo sharing app. You are doing a nuclear reactor company. You are doing foreign enterprise software. You are going to be in the mix with all the other YC companies."}, {"start": 242.37, "end": 270.05, "text": "And your peers are going to be companies doing whatever it is. Because the market does not care. It does not care if you're trying to save the world. It does not care if you have a noble purpose or not. If you don't manage to actually build a real thing for a real customer that is better and cheaper or ideally both than what they could get otherwise, you will still fail. And so I think this is the most important point I have to make tonight. Don't think you are different from a software startup."}, {"start": 270.05, "end": 297.39, "text": "Don't think that the rules don't apply to you. They still do. In fact, if anything, they do even a little bit more. And so one of the things that we work with at companies in YC is, okay, fine, you're building a nuclear reactor. That's going to take seven years. What can you get done in three months? And usually people first say, well, I can't do anything in three months. That's completely impossible. And then we talk to them and we figure out, well, you know, maybe you can get your heat system up and running. And maybe you can do a little bit more than that. And universally, by the end of YC,"}, {"start": 297.39, "end": 327.12, "text": "The companies are always able to do more than they thought they could. Momentum becomes a product of itself. Once you start having little incremental wins that build on each other and exponentially, week over week, if you can just get whatever you're working on, no matter how small of a subset it is, to be 10% better every week. This is the whole secret of software startups. Facebook in its first two years started really small, not that good. It got 10% better every week for years. And that is an incredibly powerful force."}, {"start": 327.12, "end": 357.07, "text": "You all understand exponential growth, but it really does work. And so the hard tech companies that we funded, actually another one is called Ginkgo Bioworks. They're also from Boston. They came back here after our program. I remember in our first meeting with them, we sat down and said, hey, let's figure out what you can do that you can make 10% better every week. Let's not care how quickly you think you can move. Let's figure out a way to make your business work that way. And it turns out you can do it. So this is the most important thing. You have to find a small initial project."}, {"start": 357.07, "end": 385.7, "text": "That you can actually do, that there is a real test in the world that you complete and get it done on a short time frame. The second secret, this is a more positive one. There are a lot of ways at this point where it is easier to start a hard company than an easy company. If you are starting the 11,000th photo sharing application, people will roll their eyes. It'll be difficult to hire. It's very hard to say why the 20th employee will join. It's very hard to say why the"}, {"start": 385.7, "end": 415.84, "text": "Next investor will fund you. It's very hard to say why the press will write about you. It's very hard to say why anyone will care until you have massive runway growth. But if you're starting a synthetic biology company, people will join you because they care. People want to help you. And I think this is really badly underestimated. Not that many people try to start these companies. So if you're audacious enough to do it and it matters, people want to help. It's also an automatic competitive advantage. I wasn't making that up when I said there were 11,000 photo sharing companies."}, {"start": 415.84, "end": 444.9, "text": "I had someone count that up for me once, and that was his best guess. The last time I counted there were four private nuclear fusion companies in North America. And so just not that many people tried to do it. A recent company that we had, a YC company also from MIT from about two years ago, was a company called Cruise Automation, hopefully getting acquired by GM soon. When they started that, Kyle was under a lot of pressure to do another video streaming site."}, {"start": 444.9, "end": 474.58, "text": "And he said, no, I want to do this self-driving car company. Two and a half years ago, the company didn't exist. GM just signed a deal to acquire it for a huge amount of money. If he had done the video streaming company, there would have been a huge amount of competition. And what he did instead, he had almost no competition. That also worked for SpaceX. Not a lot of other people were starting rocket companies. And it worked for Tesla. Not a lot of other people were starting electric vehicle companies."}, {"start": 474.74, "end": 503.65, "text": "So don't underestimate how much this works. However, I do think it's worth paying attention to software to see what has worked so well. You know, most of the, say, 10 plus billion dollar companies that have started in the last 20 years have been software companies. And I don't think this is an accident. There are two things about software companies that lend themselves particularly well to startups, and they are cost are low and cycle time is fast. And I think those are the two most important factors for..."}, {"start": 503.65, "end": 533.33, "text": "Software to be able to dominate, or startups to be able to dominate in an area. And so as you're thinking about whether or not a new field is a good bet for a startup, I think these are the two most important things to evaluate. If the costs have come down dramatically, if the cycle time has gotten much faster, then startups can overcome the natural disadvantages they have. Genomics has been a good example of this recently. Another thing that has worked really well for software is the unit economics. It's usually free to give people another copy of software."}, {"start": 533.33, "end": 563.3, "text": "And because software can get to such scale so quickly, you can have the network effects and the monopoly that you need to build a really great business. And the final point I'll make on this, as a general observation, software companies that we see are specific and clear on what they're going to do, and hard tech companies are usually extremely vague and uncertain on what they're doing. And I think because software companies can build things so quickly and change their mind and change what they're doing so quickly, those founders have an easier time doing this."}, {"start": 563.3, "end": 592.42, "text": "But of all of those things, I think cost and cycle time are the most important. And if you are in an industry where you don't have, you can't see a way in what you're doing, can drive the cost down dramatically from where it's been, and the cycle time, the iteration speed get more quick, get faster, then I think you have to think really hard about starting a startup there. The good news is I think those two variables are changing in almost every field. And this is a really wonderful time to start a company, doing any of those things."}, {"start": 592.42, "end": 621.14, "text": "So if you can find a new industry where you can make that work, and then you can use those to be able to compete with a small amount of money, time will always be against you. But if you can move more quickly, if you can get that 10% weekly iteration speed, weekly improvement, your big stodgy competitors will not be able to do that. And I think you can really win. It's always tempting to make excuses as a startup. Life is not fair. Big companies don't play by fair rules. Everything is against you as a small company."}, {"start": 621.14, "end": 650.29, "text": "The sooner you can make peace with that, and the sooner you can realize that complaining is not how you're going to win, all of these really shitty things happen, but you just have to find a way. It is your duty as a founder to find a way to get things 10% better every week. You can do amazing things. And although it will feel like slow progress on a short time scale, when you zoom out and look every year, these things actually happen amazingly quickly. The other thing I touched on this earlier,"}, {"start": 650.42, "end": 679.89, "text": "The other thing that's important about a short cycle time is momentum. Startups basically survive on their own momentum. A startup that is winning tends to keep winning. A startup that is losing tends to keep losing. And so one of the great skills that a founder can have and one of the great things a founder can do for a company is figure out how to make it so that the startup is winning from the very beginning, even if the wins are really small. The common mistake of a hard tech company is to sit around and do nothing and talk a lot."}, {"start": 679.89, "end": 708.67, "text": "The companies that actually win and actually create products that change the world, they start doing things, they usually get laughed at for how small they are, but they build up this culture of winning and doing better. And eventually the critics sort of, no one listens to them anymore. And you ride this amazing exponential curve of progress. Momentum attracts good people. People want to come work for a winning team. So if you start having small wins, you're able to hire well. And it makes people, I think, work harder. If you're not going to have anything done for three years,"}, {"start": 708.67, "end": 735.38, "text": "Who cares if you start today or tomorrow? If you're going to have something done this week, that incremental day really matters. A short cycle time will also save you from another mistake that these companies make, which is they build something for your customer. If you hold yourself to, okay, whatever we're building has to be something that we're going to give to a real customer in this short kind of time frame, you will not make the mistake that has killed most hard tech startups we've seen so far."}, {"start": 736.91, "end": 763.41, "text": "Software startups are able to make the cycle time really short by getting very close to their users and shipping them code every day or several times a day. This is a little bit harder as a hard tech company, but it is worth the detour. It is worth going out of your way to actually have a real customer to help you through this process. Tesla recently posted a blog post about how they were going to do this. Tesla is a 10 or 12 year old company, I think, maybe a little bit less."}, {"start": 763.41, "end": 791.57, "text": "They just last month finally released what they really have been waiting to do this whole time, which is a mass market electric vehicle. They had to go through two significant detours on the way to get there. And most founders hate detours. In Tesla's case, first they built this Roadster, which was super expensive, not very good. Well, it was okay. But they used that to make the Model S, which was still really expensive, but great. And now they're going to use the revenue stream from that to build the Model 3, which hopefully will be great and relatively inexpensive."}, {"start": 792.21, "end": 821.47, "text": "Or on SpaceX. They've been clear all along they want to go to Mars, but to get there they have to do these other detours along the way. Resupply the ISS, maybe do this internet network in space. These things are still worth it. Most founders don't want to do these things, because they're like, I have this mission, I want to get there right now. But it's really hard to start from zero and have no intermediate steps between zero and a colony on Mars. And so these detours empirically are worth it, even though intellectually they seem like they shouldn't be."}, {"start": 821.47, "end": 851.26, "text": "And one of the things that we try to do at YC is sit down with a startup and help you figure out what they are. Even if YC never funds you, I strongly suggest you embrace this model. Empirically, it's been difficult to work without this. More hard tech startups fail because they have this grand plan that sounds great in the press and they never do anything than have failed because they started with an initial project that was too small. And it really is somewhat of an art to find this kind of intermediate"}, {"start": 851.26, "end": 882.24, "text": "Goldilocks-sized project. So you shouldn't be afraid of detours, but there are a lot of other things you should be afraid of, and I'm just going to list a couple of them, and then I'm going to try in like five minutes to open this up for questions. Short-term thinking people are really bad. So there are a lot of investors and employees and partners that will tell you how long-term oriented they are and then press you to sell the company in a year or to do something in a year. And in fact, the more time..."}, {"start": 882.24, "end": 911.39, "text": "An investor spends trying to convince you that he or she is a very long-term oriented shareholder, the less you should believe it. Most people want fast. And it is a real, if you're really committed to what you're doing, actually as a side note, if you want to do like a software startup or an easy startup first, I don't think there's anything wrong with that. I don't think you should let anybody feel bad about that. You can do the hard thing next. But if you do want to start with something that you want to spend the next 15, 20 years of your life on, make sure."}, {"start": 911.39, "end": 941.54, "text": "The people you're working with are willing to sign up for that long journey. That is hard to find. In fact, I keep a separate list of investors that I know are willing to hold shares for more than 10 years, which is the typical lifetime of a VC fund. And it's not as long as you would like to believe. There had been a long-term perception challenge, which was the assumption that the best people were working on software companies, because that was the only thing that worked. And if you were working on a hard tech company, you weren't very good."}, {"start": 941.54, "end": 970.8, "text": "That's somewhat changed, but you'll still run into it a little bit. There are definitely regulatory issues. Related to short-term people, although very different, you run into a problem with negative people. There are a lot of things that will go wrong in any startup, but especially a hard tech startup. And freeing yourself from negative people that will just sit around all day and say, this is not going to work, we should give up, this sucks. No matter how strong, no matter how tough you are mentally, that will eventually wear you down."}, {"start": 970.8, "end": 1000.72, "text": "So freeing yourself from those people, I think, is good. I talked earlier about raising money and finding the right investors. Most startups struggle from not being able to raise enough money to survive. There is another failure mode, which is raising too much money. This is a very high-class problem to have. Unfortunately, it will still kill you. A lot of money can ruin cultures and sort of make you unfocused and try to do too many things at once."}, {"start": 1000.72, "end": 1027.63, "text": "Data here once at YC. If you look at the track record of companies that raised $40 million or more in their series A, super high number, not that many in the sample size, they usually fail way more often than the average series A company fails. So there is something to the culture that gets screwed up with too much money. And I think this happens even more with hard tech companies when there's already such a temptation to do things slowly and expensively."}, {"start": 1027.63, "end": 1058.13, "text": "And then finally, don't be good at science and bad at everything else. This is another common, common failure mode that we see. Someone on the team has got to get really good at business. And it is better if it is someone that is already on the team, not someone you bolt on to handle all that stuff. It is not hard. If you can get really good at science, you can get really good at this. It is less fun, but it is possible. I did it, and someone on the team has to. And then finally, I think..."}, {"start": 1058.13, "end": 1085.71, "text": "I just sort of want to close before we open it up for questions on why I think this is important and why I think this is something to consider. One thing is I think in the world today, good ideas are undervalued and underweighted. It has become popular in investing to say, oh, the idea doesn't matter. Just let the team endlessly pivot until they hit something and it'll work. And I think maybe like 70% or 80% of venture capital gets wasted in this general direction."}, {"start": 1085.71, "end": 1113.49, "text": "If you are willing to take the time upfront to really think of a great idea and a new technology that you are committed to and you can explain why it's going to work, you're already way ahead of the game. And because most people don't do that and because most of the investment climate doesn't think about that, almost everybody else, I really believe it's something like 70 or 80% of the market is chasing the current fad. So there was a period when everyone was trying to build the next Facebook."}, {"start": 1113.49, "end": 1144.14, "text": "And then the next workday, and then I don't even know what it is now. But people move together, and you have this crazy schooling effect where it's basically a bunch of fish all going in the same direction. And if you cannot do that, then you'll have an automatic advantage. You will have thought of a new idea that the rest of the world is missing. And very, very few founders do that. Most people, I think, do this because they miss price risk. It's actually not risky to start a company."}, {"start": 1144.18, "end": 1169.26, "text": "If you do it, like let's say you have nothing, you start a company, you work really hard on it for a couple of years, and you fail. You still have nothing. You have an incredible experience. You've probably learned a lot. I think the actual risk is you have this great idea, and you never try, and you spend the rest of your life regretting it. I think I have met far more people that fall into that camp than wish they hadn't started in the first place."}, {"start": 1169.26, "end": 1200.59, "text": "So most people want the safe path. They want to do the non-risky thing. They want to do the thing where if it fails, they can say, well, it failed, but I had too many competitors, but it was the right idea, the right space at the time. You're actually no better off being able to claim that than say, yeah, I tried to start a synthetic bio company and it didn't work out. So don't do that. Okay, why should you do all of this? Here is the fundamental thing that motivates us at YC. We had in the U.S. 100, 200 years now of incredible economic growth."}, {"start": 1200.59, "end": 1229.9, "text": "We had 100 years of incredible territorial expansion. We had 100 years after that of incredible innovation. And for that time, we had massive GDP growth every year. Everybody's life got better every year. And in a world like that, democracy works. Because most people want their lives to get better year after year. Now we're in a world of 2% a year economic growth. Maybe it's even lower than that. In a world like that,"}, {"start": 1230.64, "end": 1260.02, "text": "Democracy does not work. And I think you're seeing it in this current presidential cycle. You have people, you know, for your life to get better this year, if the pie is not getting bigger, you have to vote against me and my life has to get worse. And this is really bad. And so I think the only hope for the current political system in the US to keep working is to reignite innovation and have that reignite economic growth. And if we don't do that, we will be stuck in the same zero-sum cycle."}, {"start": 1260.4, "end": 1306.83, "text": "And so the good news is it's totally possible. The good news is there has probably never been a better time than today to start a startup, and specifically a hard tech startup. And I think there will be multiple hundred plus billion dollar transformational companies created, probably many of them from people at MIT, that really do kind of fix this existential risk for the country. So I hope you do that. We'd love to talk to you if you do. And I think I have about 20 minutes for questions. So we will try using the feedback tool. If it doesn't work, then I will..."}, {"start": 1306.96, "end": 1336.1, "text": "Take questions at the mic. All right. How do you pursue small achievable milestones if you are, e.g., a nuclear fusion company with only one very hard goal? Well, it is very unlikely that you build the system and there's only one thing you're unsure of. You build the entire thing and you're sure everything will work, except there is the final test of whether it produces net gain. There's usually something like a first plasma you have to create. You have to see if the plasma is stable under the heat."}, {"start": 1336.1, "end": 1364.98, "text": "Pressure conditions are going to be looked at. You have to see if you can get the magnets up to high enough field strength for everything to work. You have to see what happens to the plasma once you have unstable particles in there. And so there are all these things along the way. And what you figure out are, what are the biggest risks on one side, and what are the risks we can test first? And where do those intersect so we can plan the right order? And again, in practice, you can usually say, well, okay, let's see if we can just get a plasma in the system stable for X period of time."}, {"start": 1364.98, "end": 1395.46, "text": "And that would be like a good first goal. How can you really tell if someone is in it for the long term? This is really tough. I don't think you can tell this perfectly because I don't think most people truly know. I think everyone wants to find employees or co-founders that are 100% committed to the idea and they're going to see it through thick and thin. And I think people get more committed over time. And it's unreasonable to ask someone on day one, are you going to do this for the next 20 years? If they say yes, they are best lying to themselves."}, {"start": 1395.46, "end": 1422.94, "text": "Because honestly, how can you know that? You just, you have to work together. I think the best that you can ask for is someone that has a history of sticking to projects, and that someone seems to really believe whatever you're doing. And if someone shares the mission with you, and has in the past, and however long their career has been, stuck with things, I think that's all you can ask for at the beginning. How did Rigetti Quantum Computing demonstrate growth?"}, {"start": 1422.94, "end": 1452.1, "text": "So Rigetti is a quantum computing company, as it sounds like, that YC funded, I think, in the summer of 2014. They've done phenomenally well. They just raised a bunch of money, and they're pretty, maybe as far along as anybody to creating a usable quantum computer. In our very first office hours, Chad and I sat down, and we made a very long list that took up several whiteboards worth of space on the steps between where he was today."}, {"start": 1452.1, "end": 1482.1, "text": "And getting to a working 40 qubit system. And we did the exercise I just talked about. And he had a plan. He got done what he said he would get done in three months. He actually got that done in two months. And at this point, they're fabbing their own chips. So again, there's always some incremental milestone. Actually, his first milestone was not even a hardware problem at all. There was a software thing they needed to prove for their particular design. And so that was easy to do."}, {"start": 1482.1, "end": 1513.18, "text": "And they went on from there. What is my opinion on having the government as your customer? Or planning to have the government as a customer? I don't think people do this enough. I think there is a common belief that the government is a terrible customer. With some exceptions, people in government are not bad people. They're not stupid people. They want to do the right thing. They want things to be better and cheaper and faster. And they're definitely slower than other customers."}, {"start": 1513.18, "end": 1541.84, "text": "But I think they're fundamentally good. And we have more and more customers now that sell to government. And I would encourage you all not to discount this. It went really out of fashion for a long time, I think largely because VC Groupthink decided that the government was always a bad customer. They're sometimes a bad customer. You know, they're like a bad customer to sell enterprise software to. But I think you shouldn't rule it out, and you should be open to it. And, you know, they want to do this. They want to be better."}, {"start": 1541.84, "end": 1572.32, "text": "Is hard tech even more prohibitive for women? More than software startups? Yeah, maybe a little, because I think, unfortunately, people still have this science bias about men and women that really sucks. So I think maybe a little. I think startups in general are a little bit harder, but please don't not do it, because it's going to be harder. It is not as hard for you as the other things that are going to get in your way of a startup. So if you want to do this, and you're worried that"}, {"start": 1572.32, "end": 1601.89, "text": "This is going to be, this is the reason not, please don't do that. I think, you know, some of our more exciting hard tech companies are run by women, and they're doing a phenomenal job, so it's definitely possible there's existence proof of it, and I would encourage you to do it. What do I think of the HBO show Silicon Valley coming from a startup guy who worked a bit with the show? You know, I watched, no offense, I thought it was brilliantly written, but I watched like 10 minutes of it, and I just couldn't handle it."}, {"start": 1601.89, "end": 1633.02, "text": "So I have only seen like one part of one episode. It was kind of funny. It was like pretty painful to watch. How do you get investors to care about small wins, the hard tech startup? There's a good answer to this. If they don't care about that, you don't want them as your investor. You want the kind of people that understand that this is the only path that has ever worked for these companies. Do I have any specific or different advice for hard tech software startups?"}, {"start": 1633.02, "end": 1662.26, "text": "Well, the only one I'm closely involved with is OpenAI, which is trying to develop general AI, and that is definitely a hard tech startup in every way except the hardware component. The thing that has worked well there has been getting the right people on board and setting the right short-term milestones. So figuring out what our attack vector on how you build AI was really important."}, {"start": 1662.26, "end": 1687.6, "text": "Because once you have that, once the whole team is on board with that, and once the team has agreed on the intermediate milestones to that particular attack vector, then you don't have the problem of a difficult to manage team going in different directions. There are all these theories of management and how you do this. The one I like the best is get smart, motivated people, have everyone that you bring on believe in the same direction and the same mission, and then you don't have to do that much management."}, {"start": 1687.6, "end": 1717.97, "text": "And so that's what we've tried to do. I think that is generally good advice for startups. Is being scared of failure normal? Yes, it is. And everybody who tells you it's not is lying. Failing sucks. And once you do it a few times, it sucks less. But it is really scary. Here's the issue. It's your whole life, but everybody else is busy with their whole life. So what people are fundamentally worried about, I think, when it comes to failing is that everyone is going to spend,"}, {"start": 1717.97, "end": 1745.23, "text": "As much time thinking about you as you spend thinking about yourself. But really, they don't think about you that much. They're thinking about themselves, and you're probably not thinking about them. And so if you fail, people will be like, all right, that's fine. What are you going to do next? I think one of the things that is really good about Silicon Valley, and one of the areas that honestly Boston and Cambridge could stand to improve a little bit, is in Silicon Valley, you are measured on the best thing you ever do. And people kind of forget about the ones that aren't good."}, {"start": 1745.23, "end": 1772.5, "text": "If you never do anything good, that's still really bad. But if you do one amazing thing and then three really bad things that don't go anywhere, that's fine. Whereas I think in a lot of other places, you are measured on your biggest failure, not your biggest success. I think that's going to change. It's normal to be afraid of this. But don't let it stop you. How does one get good at business if you're an engineering student?"}, {"start": 1774.22, "end": 1807.39, "text": "Unlike engineering, which requires lots and lots of study and lots and lots of practice, the tenets of business are much simpler. So I think you do a program like Y Combinator or anything else. I would not go to business school. I think you can learn it better and in less time. But I think the best way to do it is just sort of hands-on practical education. Is there a good reason for a hard tech company not to move to the Bay Area if they have the opportunity? Yes."}, {"start": 1807.39, "end": 1836.18, "text": "So I'm generally very biased to thinking the Bay Area is a good place to be. YC actually started in Boston and went to the Bay Area, and I think that was a good decision for us. There are two trends more recently that I think are a risk for the Bay Area's long-term dominance. One is that we cannot get our act together around housing policy, and a one-bedroom apartment in San Francisco costs, I think, like $5,000 a month at this point. Maybe a little bit less, but it's on that order. It's really gone crazy."}, {"start": 1836.18, "end": 1866.16, "text": "If you are a startup constrained by capital, that is a real problem. And our leaders have been unable to fix it so far. The other problem comes down to the long-term focus. I think in the Bay Area, people join a company and think about being there for two years in general. And they often stay longer if the company is doing really well. But I think an advantage that a place like Boston has is that people are willing to make longer-term rental commitments. And that's really important."}, {"start": 1866.45, "end": 1898.3, "text": "To building a company like this. What if one wants to build a world-changing company in the long term but become wealthy in the short term? You could just start a company that won't take as long as you think. The problem is a lot of people have this idea and markets are efficient. And so the easy companies to start are never that easy. And people that think they're going to build a company and get a $10 million exit in 12 months."}, {"start": 1898.3, "end": 1928.11, "text": "Generally get disappointed for a whole bunch of reasons. But that strategy never works as well as people seem to think it's going to work. I think the most certain way to become wealthy in the short term, or at least wealthy enough, is to go work at a company like Google or Facebook and get a crazy pay package. But what a lot of people want to do is they actually want to become wealthy so that they can start the hard company. And I understand that. I think that's a noble goal."}, {"start": 1928.14, "end": 1956.85, "text": "I don't think it's necessary. I think in today's funding climate, you can raise money, even if you're a company that would have been unfashionable to fund five years ago. So if you only want to become wealthy because you have this life's work company you want to build later, I would push on that, and I would say you don't have to. If you want to become wealthy because you want a bunch of sports cars, I understand that too, and then starting a hard company now doesn't help you. So I think it ends up as a personal decision, but it's worth..."}, {"start": 1956.85, "end": 1984.08, "text": "It's worth really thinking about the motivations there. Now that Imagine K-12 is part of YC, what role will YC play in supporting Imagine K-12 startups? Education is now a vertical that we're going to do more companies in, just like biotech or hardware enterprise. They'll have access to the same services as YC companies. Does this actually go to Sam? Yes, it does."}, {"start": 1984.08, "end": 2012.56, "text": "What do I think of Google selling Boston Dynamics? I don't know enough about it to say. If they want to donate it to OpenAI, we'll take it. Well, we have to look at it first. Yeah, I really don't know enough. I've heard that it's pretty hard-coded and that there's not a lot of actual learning going on in the algorithms there, but the hardware is clearly really impressive. Do I discourage trying to find a co-founder through co-founder meetups?"}, {"start": 2012.56, "end": 2042.46, "text": "Generally, yes, we have seen one or two cases where it's worked out. Most of the time it hasn't. The trick with startups is that startups are very difficult, and at some point, in most startups, the expected value dips below the x-axis. If you have EV here and time here, at some point it goes like this and goes negative. If you have a previous history with someone, you can sort of violate the laws of thermodynamics and keep working together even when you shouldn't. Without it, it usually falls apart. And startups bring out the worst in you."}, {"start": 2042.46, "end": 2071.97, "text": "Even close friends. And so a pre-existing relationship seems really important. We have tried to host co-founder meetups. We have tried to fund people from co-founder meetups. It has generally not worked out. I don't think that means it can't work out. But at this point, my default advice to someone would be better off to be a solo founder than to start a company with someone that you've just met at a co-founder meetup, given the data that YC has. It's way better to..."}, {"start": 2071.97, "end": 2102.22, "text": "Try working together on some project first before you sort of commit your time to doing a startup together. How much time do I spend learning and reading about new science? Not as much as I would like. I'm trying to do more. I spent all afternoon with some groups here, which was fascinating. And I am temporarily running YC Research, which is our nonprofit research arm, until we hire somebody else. So in that capacity, I've gotten to spend a lot more time with it. And that's been really fun."}, {"start": 2102.32, "end": 2130.61, "text": "What is YC doing to keep itself on top? That is a great question. But the answer is really simple. YC is a network effect. It's a network more than anything else. And we have a network effect, which is that we have been very fortunate to fund many of the best startups of the last 10 years. And so all we have to do to keep ourselves on top is keep doing that in order for it to keep getting stronger. If we get distracted by all the other stuff that distracts people that fund startups, or, you know,"}, {"start": 2130.61, "end": 2161.78, "text": "Any number of other things that hurt businesses, then we fail. But that is our metric. And we just keep trying to make that single metric get better and better every year. So it is both a little bit depressing in that it's sort of this like endless specificity and task, but it is rewarding in that it's actually really fun to do that. And we know exactly how to do it. It's very hard to see somebody external killing in YC. We'd have to like really scrub ourselves, which is, you know, like that's a place that I'm happy to be in."}, {"start": 2161.78, "end": 2191.14, "text": "Hardware is not as easy to scale as software. What is my advice about launching? So this is true, but it is not that true. It was really true a few years ago, and every six months since then it has gotten easier and easier to launch a hardware company and deliver the first 100,000, 10,000 units. And my answer about what to do about launching is work with any of the number of programs, YC or others, that are good at advising hardware companies about how to build things quickly."}, {"start": 2191.14, "end": 2223.7, "text": "I won't say this is a solved problem, but it's getting close. This is no longer the problem that kills companies. There is a well-understood playbook about how to do this. I can't answer it in seven minutes, but YC or any number of other people can advise you about how to do this. Does YC invest in Chinese and Indian companies? What do we think about the future of China and India's venture environment? Yes, we do. We love to. We invest in companies from all over the world, operating all over the world. China and India are two markets of special interest to us, given the growth rates there."}, {"start": 2223.7, "end": 2251.54, "text": "And the populations. But what we say at YC is we want to fund anything that we think can be a $10 billion company. And there are so few of those at the stage we look at that we can't have any other criteria. So we're open to people in any area, any sector. We want to do that. Is it better to enter a market with existing competition or a market completely free of competition?"}, {"start": 2251.54, "end": 2282.58, "text": "I don't have a general answer to that, because some markets are brand new, some markets get missed over, and some markets have a lot of competitors, but they're all not very good. So rather than try to worry about competitors, I would just think about, do you believe you can build a $10 billion company in this particular market? And if you can, that's great. All right, anyone else have a question? I can keep going through these otherwise. Okay, perfect."}, {"start": 2284.46, "end": 2312.88, "text": "Would we ever consider having one in the East Coast? Would we ever consider having a female founders conference in the East Coast? We might at some point, but I think A, people really like the excuse to come to the Bay Area, and so when we have floated the idea with small groups of doing it somewhere else, we've gotten a pretty negative response to that. And B, it is so much easier for us to host events in the Bay Area where we have our office and infrastructure and all of our companies to support it, that when we have hosted events..."}, {"start": 2312.88, "end": 2340.4, "text": "In other countries, it's just been a huge amount more work and we can do less of them. So we don't rule it out, but we have no current plans to do that. What percent of our portfolio is a women-led company? I think for the last batch, 23% of the companies had a woman on the founding team. I don't know the statistic off the top of my head for how many of those are CEOs, but I believe it's about half. So that would get you to like 11 or something like that."}, {"start": 2341.33, "end": 2369.76, "text": "How important are patents for hard tech startups that are trying to move quickly in those first three months or so? How important are patents for hard tech startups that are trying to move quickly? So there's this thing called the provisional patent, which is what we advise all YC companies to do. It takes about an afternoon of work, costs about $1,000, and all it does is give you a year later to file the patent. But it basically just sort of like is a timestamp. And that has worked really well. That is what all YC hard tech startups do."}, {"start": 2369.76, "end": 2410.99, "text": "And then you have time later, if things seem promising, to spend the $15,000 or $20,000 and days it takes for a patent. So that is one of the rare examples of something that is worth the time, even though it is not strictly in the sphere of talking to users and building product. I'll come right back there. What is the best location for a health IT or biotech startup? I think the two obvious choices at this point would be Boston or Silicon Valley. I think you can make both of those work. I also think you can make other areas in the world work."}, {"start": 2411.6, "end": 2440.75, "text": "I think the Bay Area is really good. I think Boston is really good. But I think you should not constrict yourself just to those two places. If there's a natural reason to put the company somewhere else, I wouldn't say ask you one of those two. Sorry, whoever you are. As a founder, how do you navigate maintaining legal control of your company after accepting VC money? As a founder, how do you navigate legal control of your company after accepting VC money? So there's been a good development here, which is that VCs have learned that they make the most money."}, {"start": 2440.75, "end": 2467.46, "text": "When their founders, when founders run the companies. And so 10 years ago, I'd say 15 years ago, it was very unfashionable to let founders have absolute control over a company. After Google, after Facebook, it is very unfashionable for VCs to insist on legal control of a company because they've learned that for whatever reason, they can't help screen it up when they have control. You know, at some point, companies really struggle."}, {"start": 2467.46, "end": 2494.21, "text": "And you want the founder to have absolute control when it happens. And I've experienced this as an investor many times where in my head I've said, man, this company is not very good or the CEO is not very good. And luckily for me and my own financial interests, I couldn't do anything about it and had to just sit there and wait. So the good news is if you're raising money in today's climate, certainly in the Bay Area, but I think really even in Boston, you will never be asked to give up control."}, {"start": 2494.21, "end": 2524.66, "text": "In the extreme case, the investor will let you vote their shares forever. But at a minimum, they will never ask for board control or voting control over the company. What's the best way for business or non-technical founders to look for technical co-founders? What is the best way for non-technical founders to find technical founders? Going to school together, working together are the two best. I think co-founder meetups are definitely the worst."}, {"start": 2524.66, "end": 2560.82, "text": "And then there's everything in between those on that scale, like friends of friends and people who are doing an interesting project that you like and you email to me. But I would certainly try, I would encourage you to learn at least a little bit about science and technology relative to what you're interested in doing. You'll, I think, be way better off and technical founders will respect that effort. Any advice for how to balance working in your hard tech startup with being a good parent or partner?"}, {"start": 2560.91, "end": 2591.18, "text": "It's really hard. How to balance, I'd say any startup, with being a good parent or partner. Honestly, startups are not the best choice for work-life balance. So it's going to be a balancing act. Things are going to be a little bit out of whack. I think, you know, it's really, I don't have kids, but with a partner, I think it's really important to say, here are the things that I will..."}, {"start": 2591.18, "end": 2622.62, "text": "Not compromise on unless my company is about to die. And so, you know, if that's like a date night once a week or weekend trips every month or every other month, I think you just have to have some things that you don't compromise on. I imagine it's the same with kids, but it's really tough. And I think you just go into it realizing that it's not going to be easy. You keep talking about the next $10 billion company, but obviously not all markets are that big. What's your opinion on startups that just can't get that big? I guess it'd be called like lifestyle startups."}, {"start": 2622.62, "end": 2648.43, "text": "Yeah, I think they are perfectly good. They are not what YC is designed for. The economic model of YC works in that we lose our money most of the time, but when we hit, we try to really hit big. And so I think if you are starting a lifestyle business, the economics don't work for the kind of investor that YC is or that some venture firms are."}, {"start": 2648.43, "end": 2692.94, "text": "But there are a whole bunch of other ways to fund those, and they're perfectly good businesses, and they're underdone. I've got one more question. So the last YC you started I worked at, they told me that diversity hiring was a luxury after you got bigger, and said that if they prioritized it too much at the beginning, it would hinder their growth. So I was wondering how, I've noticed that,"}, {"start": 2692.94, "end": 2717.5, "text": "More recently, YC has started speaking about this. How do you address this with Finders Now? And is that actually, is that true? Or is it something that you, even if recruiting women or other groups slows you down with recruitment, do you still think it's worth it? Yeah, I'm sorry to hear that."}, {"start": 2717.5, "end": 2746.19, "text": "You've heard that. That's certainly not our advice, never been our advice, but we don't control the startups. They run their own company. There's a broad range of what people think about this. I think some of our best companies are ones that have taken the stance that a diverse company is really important from the earliest stages. And as other companies see, I think Stripe, for example, is a company that did an excellent job with that, that many people think will be one of the most valuable companies we funded. And so as people see examples like that, I think..."}, {"start": 2746.19, "end": 2776.03, "text": "Other people are like, oh, maybe actually this is important and this does work on it. I mean, there are the extreme cases where if you have the perfect three candidates in a row, no matter sort of where they fit on the diversity spectrum, you should probably hire them. But there's the other extreme case of where you spent no effort and you don't understand the value that diversity brings. And so I think most of our startups understand the value there. Some for sure don't."}, {"start": 2776.03, "end": 2803.79, "text": "And even some recent ones we find we don't. And one of the frustrations as a passive shareholder is that we don't get to tell people what to do. There's a good and bad in that. But one of the bad is if someone doesn't want to take this seriously, we can't make them. But I think by and large, the most successful companies tend to care about this pretty early on. And I think other people see that and say, OK, maybe there's something to this. All right. Thank you all very much. I'm going to stick around for a little bit. If you have more questions, I would love to talk to you."}]}