Elad Gil Explains Why 90% of Startups Fail in Every Tech Cycle

Elad Gil https://x.com/eladgil, of Gil Capital and Gil & Co, aka one of Silicon Valley’s most influential investors, sits down with Molly O’Shea to unpack it all.. from the 1999 IPO boom to the AI bubble, and what it really takes to build durable companies across cycles. Elad has backed more than 200 companies, from Stripe, Airbnb, and Coinbase to next-gen AI leaders like Perplexity, Harvey, and Decagon. Today, he runs Gil Capital, a multi-stage investment firm potentially managing “billions,” and continues to be one of the most insightful voices on technology, capital, and company-building. Did we mention he’s a bit of an enigma? Portfolio includes: Abridge, Airbnb, Airtable, Anduril, Applied Intuition, Braintrust, Brain Co, Brex, Character, Checkr, Coinbase, dbt Labs, Deel, Decagon, Figma, Flexport, Gitlab, Gusto, Harvey, Instacart, Mistral, Navan, Notion, Opendoor, PagerDuty, Perplexity, Pika, Pinterest, Retool, Rippling, Samsara, Saronic, Square, Stripe etc. In this conversation, Elad reflects on: Lessons from the 1999 IPO Boom: 2,000 internet companies went public, and only a handful remain. What does that mean for today’s AI rush?​AI Bubble? Why Elad believes technology waves are always “overhyped and underhyped” at the same time.​How to Spot the Next Amazon or Stripe in AI: What separates short-term success from long-term durability.​When Founders Should Sell: The four real reasons companies get acquired, and why sometimes it’s the smartest move.​Bottlenecks in AI: Why energy and regulation may shape where global training hubs are built.​Forever Private Companies: Stripe, SpaceX, and the new trend of companies that never go public.​Building a Trillion-Dollar Company: The markets, founder traits, and timing needed to reach that scale. Elad Gil: https://x.com/eladgil Molly O’Shea: ⁠https://x.com/MollySOShea⁠ Sourcery: ⁠https://x.com/sourceryvc⁠ Brought to you by: • Brex https://brex.com/sourcery—The modern finance platform, combining the world’s smartest corporate card with integrated expense management, banking, bill pay, & travel. As a Sourcery Listener you get: 75,000 points after spending $3,000 on Brex card(s), white-glove onboarding, $5,000 in AWS credits, $2,500 in OpenAI credits, & access to $180k+ in SaaS discounts. On top of $500 toward Brex travel, $300 in cashback, plus exclusive perks (like billboards..) visit → https://brex.com/sourcery • Turing https://turing.com/sourcery—Turing delivers top-tier talent, data, and tools to help AI labs improve model performance—and enables enterprises to turn those models into powerful, production-ready systems. Visit: https://turing.com/sourcery • Carta https://carta.com/sourcery—Carta connects founders, investors, and limited partners through software purpose-built for private capital. Trusted by 65,000+ companies in 160+ countries, Carta’s platform of software & services lays the groundwork so you can build, invest, and scale with confidence. Visit: https://carta.com/sourcery • Kalshi https://kalshi.com/sourcery—The largest prediction market and the only legal platform in the US where people can trade directly on the outcomes of future events: https://kalshi.com/sourcery Follow Sourcery for the latest updates! https://www.sourcery.vc/ Chapters: (00:00) Elad Gil (02:00) Elad’s career arc: Google, Twitter, Color, angel investing (02:30) Technology as a force for good (03:20) Why Elad is building monuments (09:09) From angel investing to a potentially multi-billion-dollar fund (11:03) Inside Gil Capital’s small but focused team (11:45) Thesis-driven vs. opportunistic investing (12:43) Backing 200+ companies across waves (15:15) Stripe, Airbnb, Figma, Instacart & investing across competition (17:08) Why every software company is now an AI company (18:47) AI’s biggest bottleneck: energy & geopolitics (22:10) Policy, crypto regulation, and AI’s political risks (26:14) Lessons from the 1999 IPO boom & what it means for AI (34:29) What it takes to build a trillion-dollar company

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Speaker 00

  I was looking a little bit at the data from the 90s because I thought we're in this weird moment in time in AI. What did the internet version look like?

• 0:06Speaker 01

  It spans the globe like a super highway and is called internet. There's a whole group of networks.

• 0:12Speaker 00

  Do you know how many companies went public in 99?

• 0:14Molly O'Shea?

  200.

• 0:15Speaker 00

  It was close, 450. And then if you look at the first part of 2000, another 450 or so companies went public. And then before that, I'm guessing another thousand internet companies went public. How many of those are relevant now or even exist? I think one interesting shift that's happened is there may be a subset of companies that are forever private because it can afford to be. That may be Stripe, that may be SpaceX, that may be some other companies. I don't know if you remember the direct listing wave of like 2020 or whenever all these companies started doing DLs. And it was considered the next big thing. And then like five companies did it, right? I think it's quite possible that this forever private thing will be similar. And then I think there's a bunch of companies that will just do the normal IPO thing and all the rest of it. And it's interesting to ask like, what percentage of future market cap will be these perpetually private companies versus not? And will they eventually go public?

• 0:59Molly O'Shea?

  The HubSpot CEO tweeted out, if you want to sell your company, sell it now.

• 1:02Speaker 00

  I think that's always true.

• 1:15Molly O'Shea?

  Elad, Elad.

• 1:17Speaker 00

  Molly, Molly.

• 1:18Molly O'Shea?

  Welcome to Sorcery.

• 1:19Speaker 00

  Oh, thanks for having me.

• 1:20Molly O'Shea?

  Okay, so I have like two sets of questions. One is the fun route and one is the serious route. Which one do you choose?

• 1:28Speaker 00

  I don't know. Maybe the fun route. We could do the serious route. It's up to you. I mean, it's your, it's your podcast.

• 1:35Molly O'Shea?

  It's your choice.

• 1:36Speaker 00

  Direction. It feels like a very loaded choice. Like if you choose the fun route, but then you're not actually that fun, you're having much fun, people will be like, oh, that person isn't fun. So the serious route is actually the safer route, but it sounds less fun.

• 1:48Molly O'Shea?

  What if I told you I only prepared one?

• 1:51Speaker 00

  Let's do the serious route. That sounds amazing. Let's go.

• 1:54Molly O'Shea?

  Okay, so doing some deep research on you, you have had a very interesting, unexpected career. Google, Twitter, color genomics, you then wrote a book.

• 2:10Molly O'Shea?

  Then you were one of the first AI investors and you have a podcast now. You're doing incubations. You also have a very successful fund. So how did you do all this?

• 2:25Speaker 00

  Yeah, I mean, I guess I've just been around for a while. So these things accumulate with time.

• 2:32Speaker 00

  I don't know if there's much more than that. I think the honest answer is, you know, I have sort of a broader set of things that guide me in terms of I view technology as a force for good. I think it has helped raise, elevate hundreds of millions of people out of poverty, out of famine, out of all sorts of things. And I mean, technology in a Broadway, right? Technology is the agricultural revolution. It's new medicines. It's biotech. It's running water, right? It's all sorts of things. And so when I look at the impact that's had on humanity, it's been very positive. And obviously, there's always the negative things that can happen all the rest. But I think it's been very net positive. And so throughout my career, I've kind of asked myself, where do I think technology can be most valuable, most useful? How can I help transfer it in the world?

• 3:14Speaker 00

  And so that's driven a lot of the choices of where to be and what to work on and, you know, how to get involved with things.

• 3:20Molly O'Shea?

  Not to mention monuments.

• 3:22Speaker 00

  And monuments. And so, yeah, I'm working on a project now to try and build a series of new monuments. So if you look at every society at its apex, ultimately, has built large public art to inspire the next generation to move forward for progress. Maybe it captures its ethical values. Maybe it captures its ambition. And I think the last time we built a large scale monument was Mount Rushmore, which was initiated in the 20s and was built up until the early 40s. And then since that time, we really haven't done anything societally, at least in the US. And maybe you could argue the sphere in Las Vegas is almost like a monument of some sort, right? But we used to do these really big symbols and the Statue of Liberty, which was obviously a gift,

• 4:04Speaker 00

  is a great example of that, right? You get off of Ellis Island as an immigrant and you'd be inspired by this giant statue and this symbolic nature of like liberty and freedom and this new land that you've come to. And so we're working on doing monuments or public art across four or five key cities in the US.

• 4:25Speaker 00

  And we've been talking with different, both artists and artisans around different materials. We've experimented with different bronze works. We've experimented with marble. We're actually looking at carbon fiber, which nobody's really used, which you think is really interesting for all sorts of reasons. And then we've provided proposals to a few places already in terms of some of the public art that we could build. And I think the smaller scale of what we want to do is probably 15 to 20 feet tall, which is kind of the scale of the Atlas in front of Rockefeller Center, which I think is a beautiful work. And then the hope is to eventually do something, at least one, that's, you know, sort of Statue of Liberty or Eiffel Tower scale.

• 4:59Molly O'Shea?

  Sorcery is brought to you by Brex, the financial stack trusted by more than 30,000 companies, including one in three venture-backed startups in the US. Nearly 40% of startups fail because they run out of cash. Brex is literally built to help founders avoid that. Unlike traditional banks that let your money sit idle, shipping away at it with fees, Brex's designs help you spend smarter and move faster. Their all-in-one solution combines checking, treasury, and FDIC protection into one powerful account. You can send and receive money globally at lightning speeds, get 20 times the standard FDIC coverage through their partner banks, and even high yield from day one. With same-day and even same-hour liquidity, access your funds

• 5:43Molly O'Shea?

  anytime. Companies like Scale AI, DoorDash, ServiceTitan, Hims, Anthropic, Flexport, Robinhood, and Plaid. Trust and use Brex.

• 5:53Molly O'Shea?

  Start today at brex.com/sorcery. That's B-R-E-X.com/sorcery. How do you make something not so polarizing in today's world?

• 6:06Speaker 00

  Oh, I don't know. I mean, in yesteryear's world, these things were polarizing. So the Eiffel Tower was considered very ugly when it was built, right? It was a giant steelwork, and Eiffel was actually a bridge maker, right? And he had all these divots and pieces of steel. And he's like, I'm gonna make a big thing for the World Fair in the late 1800s. And it was actually meant to show off French steelmaking, right? It was an ode to technology progress. And apparently, a big chunk of the city thought it was ugly and terrible, and they were looking forward to it getting torn down. And so I think these things that we now view as beautiful and iconic and inspiring and wonderful at the time were not viewed that way. And one thing that I thought was really cool, I went to see the Eiffel Tower actually recently as sort of a refresher over the summer.

• 6:51Speaker 00

  And he had his office on one of the top floors. And it was a power move for him to invite people to his office and they would take the pneumatic lift. I don't know if you've been in it. It's this very quiet elevator that was built 150 years ago, using modern technology right of the day. And so the elevator takes you up there and you get in and you go into his little office and they still have the desk and everything set up. And it was him showing off, like what he built, which I thought was kind of neat.

• 7:15Molly O'Shea?

  I'm curious what happens. Because in San Francisco, do you remember that statue? Is that still up?

• 7:23Speaker 00

  Which one?

• 7:24Molly O'Shea?

  The woman one in front of the Embarcadero?

• 7:26Speaker 00

  Oh, I'm not sure if it's still up. They also used to do a lot of Burning Man art, like in Hayes Valley. And if you remember, they'd bring art from Burning Man or right before they take it to the burn and they put that up there. But there's a lot of temporary art still. And so there's some in San Francisco, there's some in New York periodically where they'll sort of do some public works. There's a little bit in Miami sometimes. But I'd really like to move to like big, permanent installations that have meaning, that hopefully can transcend a really long time period, hundreds of years. You know, you want the thing to be relevant 200 years from now. And so then you ask the question of, okay, what are societally relevant themes over long periods of time? And what is actually forms of beauty that we will agree 200 years from now is still beautiful? Because if you look at a lot of post-modernism and a lot of architecture, I don't think it's

• 8:10Speaker 00

  that attractive anymore, right? And there are things that are still beautiful, right? Gothic periods and Art Deco and, you know, there's all sorts of, people still love seeing Roman ruins because they're so beautiful, right? Even though they're ruins. So I think there's some things that feel like they're universal and there's some things that seem like they're a moment in time from an art or architecture perspective.

• 8:30Molly O'Shea?

  What do you think of, is it Bezos that has the clocks sculpture?

• 8:34Speaker 00

  Yeah, oh yeah. He has the, basically, he tried to build a clock that would last for millennia. And I think it's very cool conceptually. I think it's really neat that he did that and he was thinking big. I think the hard part is it's kind of isolated, you kind of have to go out to get to it. And so what I'm interested in is can you do these societally inspiring things, but in major cities, in major thoroughfares? Like, how can you be central? To where people are spending time because if people don't see it, it kind of doesn't matter. You know?

• 9:02Molly O'Shea?

  As a multi-dimensional person, you have culture, you have investing, you have technology and science. On the investing standpoint, you started off with angel investing.

• 9:13Speaker 00

  Sure.

• 9:13Molly O'Shea?

  Apparently you're going to have a big fund soon.

• 9:15Speaker 00

  Yeah.

• 9:16Molly O'Shea?

  How did we get here? Like, so what was the arc from angel investing to this fund? Because the strategy is insane. Like you're doing seed and pre-seed and incubations, but you're also leading $100 million Series Ds.

• 9:33Speaker 00

  Yeah, I think basically we've ourselves investing from incubation or pre-seed through pre-IPO. And at this point have done every stage and have done very large investments. So I think we've, we've now led a few rounds that are a half billion plus each. And, um, you know, ultimately the, the starting point to your original question was I was just investing my own money as an angel. And I ran out of money, like literally, I just, I ran out of money because I put all my money into startups.

• 10:02Speaker 00

  And so I started raising SPVs, which are these single investment funds, right? You raise money to invest in a specific company. And that was really tough for a couple of years, like nobody had heard of me. It was very unusual for a single person to go and raise funds.

• 10:17Speaker 00

  And I was, I was involved with a lot of the key companies who kept giving me bigger and bigger allocations like a Stripe or other companies. And I just couldn't, I just didn't have the money. And so I scraped together the first few of those and then the LPs in those eventually asked me to, um, if I wanted to do a standalone sort of fund. And I kept angel investing alongside that. And now I kind of do a mix of, I can still do personal investments. I can do things through the funds. I can do incubations, you know, it's, it's kind of multifaceted. And to your point, we even do things now like AI driven rollups, right? So we'll help actually. Fund entrepreneur, founder who wants to go and consolidate an industry or buy into different traditional businesses and then, you know, modernize them with AI.

• 10:57Molly O'Shea?

  How do you think about the categories? Because you're investing in many different things. As a solo GP, by the way.

• 11:03Speaker 00

  Yeah, well, I'm gonna have a small team.

• 11:05Molly O'Shea?

  How big is the team?

• 11:06Speaker 00

  On the investment side, I have four people who are helping me. And then I have about another six, seven people helping with back office and finance. And, you know, I tend to have always thought that it's really important to ensure that you're properly focused on compliance and accounting. And, you know, I'm very paranoid about doing things properly. And I'm always paranoid about fraud or other things that can really be existential to a business. And so I view that in the context of what I do as well, like, is really important to get right. So I have a whole team just dedicated to, you know, accounting and finance. And then I have other people helping with a few other areas, events, and, you know, things like that. You know, it's interesting, the, the, there's sort of two approaches to investing. One approach to investing is to say that you're going to be thesis driven,

• 11:49Speaker 00

  and you're going to look for specific things. And then the other approach is you're just opportunistic, and you wait for the best things to show up. And the argument for the first one is every once in a while, there's a sea change or a shift in technology that's so important that it makes sense to just go looking. And so two examples of that in my career have been crypto, where in 2016, 2017, I just started doing a lot of crypto investing because it was such a clear sea change. And I wish I'd obviously started earlier. But that's when I invested in Coinbase and other companies in that, in that area. And then I think more recently, with the generative AI wave, I ended up getting involved with that quite early, 2021, 2022. And that was, you know, I played with GPT-2, and then GPT-3, and it was a huge step function up. And all the scaling papers were out, and

• 12:32Speaker 00

  you could just read them and see that all this stuff was coming. And so I just started looking for people working in the area. And so that's when I funded perplexity and Harvey, and eventually Decagon and a bridge and a bunch of other companies. And so that was a little bit thesis driven, like, hey, I think this big change is happening. So let me just see who's doing what there. And very few people were looking. But there's a lot of companies I invested in simply because I met the founders, and I thought it was a really great company. You know, that was Airbnb, and that was Stripe, and that was Figma, and that was Instacart, and, you know, Notion, and all. And so I think some of these things in hindsight feel like trends. You know, oh, there was a fintech wave. You know, in fintech, I did Stripe, Square, Affirm, Brax. But I wasn't like, I'm a fintech investor. I just thought it was like an important set of companies.

• 13:16Speaker 00

  And then in hindsight, oh yeah, of course there was this wave that was happening. So, you know, I think in general, there's thousands of great founders out there and you could think of them as a distributed search function, right? You have thousands of people exploring all the space of entrepreneurship. And so how can any one person or small group of people come up with better ideas than thousands of people in a distributed manner? You know, it's just impossible. But every once in a while, there's this technology shift. And if you notice it, you can, you can really get involved with important things.

• 13:41Molly O'Shea?

  How many companies have you invested in?

• 13:42Speaker 00

  Um, I don't know the exact number. So I've invested in, say, I'm making up the number 10 to 15 companies a year, but I've been doing it for 15 years. So that's between 150 to 225, if you use that as an estimate over like a really long time span. So many of those are now sold or public or dead or, you know, all the various things that can happen to a company.

• 14:02Molly O'Shea?

  Can you name all of them?

• 14:04Speaker 00

  Oh yeah, of course. I can actually do it alphabetically and then reverse alphabetically.

• 14:07Molly O'Shea?

  Are you serious?

• 14:08Speaker 00

  No.

• 14:08Molly O'Shea?

  No?

• 14:09Speaker 00

  Yeah, no.

• 14:09Molly O'Shea?

  Oh, come on.

• 14:10Speaker 00

  That'd be tough, yeah.

• 14:12Molly O'Shea?

  Wow, I was really excited there.

• 14:14Speaker 00

  Yeah, it would have been a very long episode.

• 14:16Molly O'Shea?

  Would have been great.

• 14:16Speaker 00

  People would start to get bored. They'd be like, oh, he's on the S's now. It's always the worst letter, so many S's.

• 14:21Molly O'Shea?

  It would make a good clip.

• 14:24Speaker 00

  He just released that on Twitter. Yeah, that'd be really...

• 14:26Molly O'Shea?

  The lead's still going.

• 14:27Speaker 00

  That's so viral. They'll be like, oh my God, the Ts, the Ts in this one are amazing, so.

• 14:32Speaker 04

  In today's high-speed business world, staying ahead means using the smartest tools possible, including the powerful capabilities of artificial intelligence. Meet Turing Intelligence. Turing builds customizable AI systems designed to solve your mission-critical challenges, no matter your industry. From expert guidance to tailored projects, Turing helps top companies realize AI that's more capable, more adaptable, and more effective. With Turing, discover how AI can accelerate your business growth. To learn more, visit turing.com/sourcery, spelt S-O-U-R-C-E-R-Y. That's turing.com/sourcery.

• 15:09Molly O'Shea?

  So, given the fact that you've invested into so many companies through different waves, how do you think about investing in competitive companies? I'm sure there's been some crossover across.

• 15:21Speaker 00

  I try to be really cautious about that because, you know, I've started two companies myself. And you know, I want to be very respectful of founders and what their goals are and their wishes are and everything else. And so I often ask both sides. I say, hey, this thing has come up. Do you view it as a conflict? Is it okay if I proceed, etc.? And so I usually wait for that, okay. I think that there's a number of circumstances where people think something's, I'd say 90% of the time, founders think that they're going to compete and they don't. They tend to go in very different directions, especially if it's too early stage companies.

• 15:59Speaker 00

  And then maybe a few percent of the time, things actually converge. And sometimes they actually converge on the companies that you don't expect. And so I've seen more conflicts and things that I didn't think would converge than things would. And I'll give you an example. I remember I was an investor in Square and the Stripe founders pinged me and asked if I wanted to invest. And so I pinged one of the key executives at Square and said, hey, do you think this is a conflict? And he said, absolutely, this is a conflict. You shouldn't do it. These things are going to collide. And so then I texted Jack and I'm like, hey, is this a conflict? He's like, no, go ahead. And so I went ahead and invested. So I'm glad I did that extra ask. And then in hindsight, they didn't really end up competing. So.

• 16:38Molly O'Shea?

  Is it harder on the earlier stage side?

• 16:42Speaker 00

  There's definitely more potential for conflicts, but I think most of the time it's less likely. I actually think on the later stage side is usually when it's pretty clear. Or suddenly the industry shifts and then it sort of crystallizes and you're like, oh, wow, like cognition is now competing with cursor, they pop windsurf. You know, it's a more direct thing than maybe it was even six months ago.

• 17:01Molly O'Shea?

  Do you think because we're in this wave of AI, you have to be an AI company in order to survive or even be relevant?

• 17:08Speaker 00

  Yeah, I mean, arguably, anything software is to some extent AI now. And so I can't think of a modern new software company that isn't using AI. Maybe if you're truly doing something basic in fintech or some other area, although that fintech wave is, for chunks of it are kind of over, right? There's still a lot of innovation in different areas and new things to do. And there's a lot to do in crypto, but fundamentally, you know, traditional fintech isn't as active as it used to be.

• 17:35Speaker 00

  And so there may be some pockets where if you started a company today, the AI wouldn't matter that much. And you could argue it two ways. You could say, well, if you're in the middle of one of the most important technology waves of all times, maybe you should be focused on that thing, right? And in general, early in technology waves, that's when you have all the low hanging fruit. And so one of my pieces of advice to founders is often who are doing really hard things right now in AI, I'm like, why are you doing the really hard thing? Just go do the dumb, easy thing that you think is non-defensible and build defensibility, because the really hard, complex thing is something you should do in three, four years when you've run out of the easy stuff, right? And so it's almost like you're going for a walk and there's a giant mountain and there's a nice little path around it. And you're like, I'm gonna go up the mountain. You're like, well, you know, here's just take the nice path. And so I think there's

• 18:19Speaker 00

  that sort of overlay right now, there's easy stuff in AI, it's a good time to go do it. You know, the flip side of it is, there's lots of other areas that people are working in, although again, often AI plays a role, like defense tech is largely drones and intelligence and other things that are, you know, really important now in the AI world. But there may be other verticals where it's just a little bit less important, maybe aspects of energy, you know, like what base energy is doing with battery storage, that isn't necessarily AI-enabled, right? But it's a very different type of sector or industry.

• 18:47Molly O'Shea?

  How do you think about the bottlenecks in AI, one being energy?

• 18:51Speaker 00

  Yeah, I mean, it's an interesting question, because energy is one of those areas where we have all sorts of solutions, we just choose not to use them, right? Nuclear is a great solution for all sorts of things in energy. And if you look at nuclear as an example, public opinion on nuclear flipped in the mid 70s, right? It went from something everybody was for to something many people were against. And that was probably a mixture of the environmental lobby and big oil effectively wanting to kill the industry. The US is still 18% nuclear despite not having opened a nuclear power plant in 50 years. France is 70%, Japan is 25%. So we know that we have a clean, abundant, very safe energy source, actually. I don't know if you've ever read the Wikipedia page on nuclear. That's how I spend most of my free time is actually on that page.

• 19:34Molly O'Shea?

  Wikipedia?

• 19:35Speaker 00

  Yeah, no, the nuclear page specifically.

• 19:36Molly O'Shea?

  Specifically?

• 19:37Speaker 00

  Yeah.

• 19:37Molly O'Shea?

  Why?

• 19:38Speaker 00

  I'm joking, I don't really spend much.

• 19:39Molly O'Shea?

  What?

• 19:39Speaker 00

  Yeah, that'd be awful.

• 19:40Molly O'Shea?

  You're pulling jokes on me. This is unexpected.

• 19:43Speaker 00

  I'll stop. So if you read that page, you basically see that a lot of the direct evidence on deaths from nuclear are very small. And in some cases, it's like in Fukushima, there was two or three deaths, and they think it was elderly people who had heart attacks, right? It wasn't necessarily from the radiation, although there may have been one death there. And so most of the deaths from nuclear that are clearly provably traced back to the nuclear accident are either Russian submarines in the 80s, which for all sorts of reasons you can imagine, and then radiology lab accidents where they misdose people. And if you compare that to stats of people dying installing solar by falling off of roofs, more people per year die from solar installs.

• 20:28Speaker 00

  And so it's very safe, right? And so we do have these things we could do societally. We choose not to, which is a different topic. But if we want to do something, we can. And the question is, are we going to constrain ourselves? And the best example of that is probably Europe, right? Europe's energy policy is really hurting their ability to win new data centers for training. And obviously, they have extra regulation for it, too, that makes it worse and all these things. But fundamentally, energy costs in Europe keep going up. They shut down nuclear power plants in Germany that are working perfectly fine. They're dependent on Russian oil. There's all sorts of very bad policy decisions. But what that's doing is it's having this weird side effect of hurting their AI industry because you're not going to build new training data centers there because it's too expensive

• 21:13Speaker 00

  on a relative basis to run them. You may do inference because of GDPR forcing you to do it, but you're not going to do training. So all the training will probably go to places that have the cheapest, densest energy and then a proper fiber optic cable in for the data transfer. And so that's probably parts of the US. That's probably the Gulf, like the Gulf. I never, five years ago, I never would expect it, but I think they're really going to end up with some of these data center build outs potentially. The Trump administration, I think, in their AI plan actually had the ability to sort of extend into UAE is effectively US land from the perspective of training. And so it's interesting to ask these questions of like, how do global energy policies impact

• 21:53Speaker 00

  costs and therefore impact where AI is going to be really important as an industry, right? Because if you look at these data center build outs, they've added or they've accelerated some aspects of GDP in the US, right? It's actually a lot of money is being spent. And so Europe is shutting itself off from some of that money right now.

• 22:10Molly O'Shea?

  How do you think about the new administration's approach to technology? We're going through many different waves of lots of support that, to your point, nuclear AI. How do you view that and then the action afterwards? Because things come up and they're like trendy, they're cool, you get a lot of support. And then your 15 minutes of fame like dies out.

• 22:32Speaker 00

  Yeah, I think it's really sector dependent. So for example, I think in crypto, the main thing the crypto industry wanted for years was just certainty. Is this something that SEC regulates, the commodities folks, like where, who governs what? What are the rules and how can we just follow the rules? And so I think a lot of the crypto industry just wanted clear guidance on what to do. And that was absent for many years. And then I think this administration has started to provide a lot of that clarity. To your point, you then need to codify it with legislation because if you don't, it's just going to be dependent on who's in power. And so you want to remove that dependence. And so I think a lot of what the crypto industry, my sense is, is thinking about is how do we pass certain bills or help certain bills progress so that we can end up in a situation where these things are codified in law?

• 23:17Speaker 00

  And they don't just become executive orders that can be rescinded with the next administration. For AI, I think similarly, we have to be very thoughtful about it. And the dangerous part in AI was their state level initiatives that have been very negative or could have been very negative for AI. I think Scott Wiener proposed one in California. I think there's others that have been proposed in other states that you could really curtail overall US progress in AI because you set the standards in one state where it's important to function, where most of the researchers are, right? Or most of the, where some of the data centers are, et cetera. So, you know, there's a national level of it and then there's a state level. And I think both are important for some of these topics.

• 23:55Molly O'Shea?

  Yeah, I think about this a lot because I love the American dynamism movement and how that like propelled small tech as the Andreessen Media Empire calls it to now big tech. So as we think about the government's involvement in these different sectors, they're also putting dollars to work. So how are these deals, one, getting structured, whether it's MP Materials or Intel? And are you at all concerned about them and whatever happens with TikTok?

• 24:24Speaker 00

  Yeah, I don't have like deep thoughts on those things. I haven't spent a lot of time looking at government investment in some of these companies. And they very much are companies that I tend not to spend as much time with because they're much later on in life. I mean, Intel was started in the seventies and I mean, it was late sixties. I can't remember at this point. And so I just haven't thought about this very deeply. I think there's a slightly different question, and I think Andreessen Horowitz, for example, has been very good at promoting this concept of small tech, and we need to advocate for that. I do think that one thing the prior administration has done, which this administration is, to some extent, continuing, is effectively curtailing some of the MNA activity you'd normally see big tech participating in. And I actually think that's a negative for the startup ecosystem. I think it's good for companies to be able to exit.

• 25:09Speaker 00

  I think we're going through a massive sea change right now technologically. A lot of things are being innovated on that won't necessarily sustain on their own, and actually having larger partners will help a lot for them. I was looking a little bit at the data from the 90s, because I thought it was really interesting to say, okay, we're in this weird moment in time in AI. What did the internet version look like?

• 25:28Speaker 04

  Sorcery is proudly sponsored by Carta. Carta is transforming the private marketplace, connecting founders, investors, and limited partners through software purpose-built for private capital. Trusted by more than 65,000 companies in over 160 countries, Carta's platform of software and services lays the groundwork so you can build, invest, and scale with confidence. Carta's fund administration platform supports over 9,000 funds and SPVs, representing nearly $185 billion in assets under management, with tools designed to enhance the strategic impact of fund CFOs. For more information, visit carta.com/sorcery. That's C-A-R-T-A dot com slash S-O-U-R-C-E-R-Y.

• 26:13Speaker 00

  Do you know how many companies went public in 99?

• 26:16Speaker 04

  200.

• 26:17Speaker 00

  It was close, 450. And then in the first, oh, you know, it's within an order of magnitude. And the hundreds is close, in my opinion. You didn't say like 10 or, you know, 10,000. And then if you look at the first part of 2000, another 450 or so companies went public. And then before that, I'm guessing, another 1,000 internet companies went public. So you had 2,000 companies go public.

• 26:39Speaker 00

  How many of those are relevant now or even exist, right? There's probably two or three companies that are highly relevant, very important, Amazon and the like. And then there's probably a dozen companies that are still around that are important enough or important ish. And then the other 1,980 companies all died or got bought. Right. And so we had this massive culling of the entire ecosystem over a reasonably short period of time. And so the question is, what's the AI version of that? And how should you think about that as a founder? And how should you think about that as an investor? And there's all sorts of aspects of that. Should you sell your company or not? What are the signs you should? When should you exit? How do you think about that relative to what you're trying to accomplish?

• 27:23Speaker 00

  Both from an impact perspective and then also from a business perspective. And I don't think that many people are really actively thinking about that, but it is an interesting cautionary tale.

• 27:33Molly O'Shea?

  What do you think is going to happen? All those companies that were supposed to go public that are just waiting.

• 27:37Speaker 00

  Oh, so the backlog from five, 10 years ago.

• 27:39Molly O'Shea?

  The backlog.

• 27:40Speaker 00

  Yeah, I mean, I think a subset of them will go out. I think one interesting shift that's happened is there may be a subset of companies that are forever private because it can afford to be. That may be Stripe, that may be SpaceX, that may be some other companies. And I think it's only, I don't know if you remember the direct listing wave of like 2020 or whenever all these companies started doing DLs. And it was considered the next big thing. And then like five companies did it, right? I think it's quite possible that this forever private thing will be similar. And there's a handful of companies that are murky brands. They'll always have access to capital. And the founders have a specific way that they want to do things and they'll just be private. And then I think there's a bunch of companies that will just do the normal IPO thing and all the rest of it.

• 28:21Speaker 00

  And it's kind of interesting to ask like, what percentage of future market cap will be these perpetually private companies versus not? And will they eventually go public? And there's all sorts of dynamics that then kick in around capital markets and other aspects of these things.

• 28:33Molly O'Shea?

  How do you view that as an investor in getting returns?

• 28:37Speaker 00

  Like I've never sold a share of Stripe and I hope to never, you know, do so. Or I shouldn't say never, but like for a very long time, you know, I'm a big believer in the company.

• 28:47Speaker 00

  But it does impact how you think about, you know, do they have regular tenders? Can you get liquidity? And, you know, for most companies that are doing well, you can always find some form of liquidity. The question is, do you want to take it? What's the discount? And also just, is there pressure? And what sort of pressure? And can you avoid that pressure over time? And so how do you even structure? How you approach investments if you're an investor, if you're a founder, similarly, there may be things that you do or don't want to do relative to the structure. So.

• 29:13Molly O'Shea?

  Do you coach any of your founders on the different exit paths?

• 29:16Speaker 00

  Yeah, people call me quite a bit on that stuff. And I think it's always just context dependent. I think the only good generic startup advice is that there's no good generic startup advice. So I think it's all contextual.

• 29:26Molly O'Shea?

  The HubSpot CEO tweeted out, if you want to sell your company, sell it now. What do you think about that?

• 29:33Speaker 00

  I think that's always true. So if you want to sell your company, you know, you probably should sell your company. I think there's usually four reasons people sell, right? They're really tired. You're burnt out. You've been working nonstop. You're not getting any joy out of it. You're fighting with your co-founder, whatever it is. So one is you're just unhappy or you're tired. Two is a company is paying so far ahead that it makes sense for you. You do the math and you're like, there's no way I'm going to hit that in any reasonable timeframe. And some people keep going then, but rationally, it's kind of like, maybe I really should get out. And the hard part is to know, is that correct or not? And I think that maybe the way I would put it is, there's a handful of companies that should never sell or should

• 30:15Speaker 00

  never have sold, but in some cases they did. And maybe they did the math wrong, or maybe they misunderstood their own position or something else, right? But if you're in one of those companies, you should never sell. You should just keep going. And then there's, and then most other companies probably should sell. And then she just figured out the right price because most companies just don't get that big. And most companies fail. I think one thing we forgot in this era of sort of perpetual funding is that the traditional venture portfolio, you know, 50 to 80% of companies would fail outright. And even at the series B stage, it was like 40, 50% of every company fails. It's a lot of companies go out of business. And I remember there was one company, I think it was Friendster,

• 30:57Speaker 00

  they got offered, what was it, like a percent or two of Google at the time, as they were raising their series A. And it's possible they could have turned into Facebook and that was the right decision. But, you know, you see these companies where they really could have been these massive outsized outcomes if they'd sold. And, you know, in the end, it didn't necessarily work out as well. And then there's a third reason, which is competitive dynamics. Like, you know, there's again, a handful of reasons to actually sell.

• 31:22Molly O'Shea?

  This is, this is an interesting stat because.

• 31:26Molly O'Shea?

  The Series A, there's like such a wide gap there and a wide gap in valuation. Those companies that are actually AI companies are getting a 30% premium. Do you think that's like, you think that's a mark of a hype cycle? Do you think that's relative?

• 31:41Speaker 00

  I think every time you go through a big technology shift, things are simultaneously over and underhyped. And they're

• 31:49Speaker 00

  underhyped because it's such a massive wave of transformation. So say you look at Azure earnings a quarter or two ago, they did, I think it was like 28 billion in revenue-ish. And 10 to 15% of that was AI. So it's a few billion dollars a quarter on Azure alone for AI spend. That's enormous, right? If you extrapolate to the other clouds and everything else, that's real money. So real money is being spent on AI today and there's very little actual impact, which to me means there's even more room to go. Right? In other words, we're just in the very earliest innings of all this stuff from adoption and usage and everything else. And there's massive adoption. You look at ChatGPT, you look at Cursor, you look at cognition, you look at all these things. So we already know that it's important that it works and it can really scale up.

• 32:33Speaker 00

  And so it's underhyped from that perspective. And then the average company is probably overhyped back to this point of 2,000 IPOs and like a dozen relevant companies. Right? Definitionally, most startups are going to fail. And this wave is going to be just like every other one. And the place where every founder and investor think they are is in the right companies. Right? And the reality is most people aren't, right? 90% of people are going to be in something that doesn't work. And so then the question is, can you actually tell what is good and what isn't? And what's durable versus not? And there's different forms of durability, the short-term durability and long-term durability. The short-term durability was Braintree, right? Braintree was a great payments company. When I funded Stripe, everybody thought, why would you fund Stripe? Braintree already does it, literally.

• 33:18Speaker 00

  That's what people would say. And then Braintree eventually got bought by PayPal for, I think it was 800 million. And Stripe is now a $100 billion market cap company. And so it would have been great to be involved with Brain Tree, it would have been even better to be involved with Stripe. And so when I look at the set of AI companies today, part of the question I ask is, is this Brain Tree or is this Stripe, even if it's working, right? Because because just the fact that it's working now doesn't mean that five years from now, it's going to be the important company in the market, given how fast things are changing and the competitive dynamics and all the rest of it.

• 33:46Molly O'Shea?

  Can you justify OpenAI's $500 billion secondary valuation?

• 33:51Speaker 00

  I don't know all the details of their revenue and things like that. My sense is that for many of these foundation lab companies, if you just do a revenue multiple and you look at the growth rate, it's actually pretty reasonable.

• 34:04Speaker 00

  That's true of OpenAI, that's true of Anthropic. I think it's actually surprising. You look at it and you're like, oh, my gosh, they have a lot of revenue and it's growing really fast. And obviously you look at margin structure and other things, but it actually seems like it's in the ballpark of reasonable for many of these things. I don't know, again, specifically OpenAI's latest numbers, so I can't say specifically about them. But in general, as I've looked at these things over time, it's actually been pretty reasonable.

• 34:26Molly O'Shea?

  What do you think it takes to become a trillion dollar company?

• 34:29Speaker 00

  TAM, just a big market. That's really the answer. Google's a giant market. AI chat or API is a giant market. Cloud businesses are giant markets. Selling productivity software plus the world's most important OS or one of the most important OS is a giant market. Selling phones to a billion people and one of the world's most important OS is a giant business. These are just giant businesses. Selling chips for everything for AI is a giant business.

• 34:56Molly O'Shea?

  What do you think on the founder level? Like, what do you think it takes for a founder to get to that level?

• 35:02Speaker 00

  Yeah, I think I think ultimately there's usually two or three characteristics. One is you have to find a giant TAM and build a product people care about in it. And the question is, is that luck or is that skill?

• 35:17Speaker 00

  And I think for some people, it's I used to do this thing. I don't really do it anymore. When I'd meet people, I'd sometimes almost think of their life as like a Monte Carlo simulation. If I re-ran their life, which everybody does, right? It's like a standard thing people do. If you ran their life a billion times, what's the expected outcome, right? Or what's the average expected outcome? And for some people you meet them and you're like, oh, my gosh, this person is amazing. Like their average outcome is going to be up here. And for other people, you're like, wow, this person got lucky. You know, they're not that good. And they hit that thing and oh, my gosh, like what a lucky, you know, I'm not going to cuss, but what a lucky person.

• 35:52Speaker 00

  And so usually the people running these companies are really good. And you know, is really good finding the time or is really good exploiting it? And maybe it's both, I'm not sure. But the best founders tend to be very smart, very driven, very aggressive, wanna win, relentless work ethic, you know, it's a bunch of stuff that for a couple of years, people tried to say wasn't important, but it's extremely important. They tend to be very smart, very strategic, able to motivate people extremely well. I don't think, and you know, people are kind of on this weird mimetic thing about autism. They're not autistic, you know, they read people very well.

• 36:33Speaker 00

  Whether they care about how the person's feeling in the moment may matter less or more, but they can read people very well and they know how to motivate them and poke them, you know, so I think a lot of that stuff gets really misunderstood. Some of them are deeply loyal and friendly people, you know, so I think there's these interesting almost misconceptions in startup lore about like what makes for somebody who's very effective at these highest levels.

• 36:55Molly O'Shea?

  Who's an entrepreneur that you admire?

• 36:58Speaker 00

  I mean, Elon Musk has just accomplished amazing things.

• 37:01Molly O'Shea?

  Why?

• 37:02Speaker 00

  Why has he accomplished them or why is that, why are they amazing? Well, I mean, sending rockets into space, building a next gen electrification company slash cars for the first time in the US in a long time.

• 37:17Speaker 00

  You know, co-founding OpenAI, like that's pretty, that's pretty good.

• 37:22Molly O'Shea?

  It's not bad.

• 37:22Speaker 00

  Yeah, it's not bad, you know?

• 37:23Molly O'Shea?

  He's doing okay.

• 37:24Speaker 00

  He's doing fine, you know, I could have done better, but I'm just saying, you know, he's not bad, so.

• 37:29Molly O'Shea?

  Hey, it's Molly. If you enjoy our interviews, check out our newsletter, sorcery.vc, where we deliver a once a week top deals and tech headlines email and also go deeper on our podcast interviews. Subscribe to Sorcery today and don't forget to subscribe to the podcast on YouTube, Spotify, Apple, or wherever you listen. Link in description to sign up.