The end of Venture Capital? (VC Roundtable) | E2285

We may be at the end of the venture capital industry.

5 US firms captured 73.1% of all LP commits in the first quarter of this year.

This is the era of consensus capital. Basically the end of a 50, 60-year run of an industry or a craft business. And that won't be the first time that a craft business has gone extinct.

I don't think that the demand for medium-sized firms is going to disappear. It probably will increase. It's difficult to make returns work at a very high level. And then it's also, I think, difficult for founders to, you know, work with enormous behemoths of venture firms.

The math may just cease working in the venture capital business.

This Week in Startups is brought to you by Grasshopper Bank. Time is money. Don't waste either. Go to grasshopper.bank/twist and get an exclusive $500 cash bonus just for opening an account. PaperOS. Building an empire? PaperOS offers the largest library of AI-driven workflows for both founders and fund managers. Whether you're raising capital, launching a fund, or wading through diligence, PaperOS unlocks simplicity and scale for your ever-growing empire. Claim your $10,000 credit at paperos.com/twist. And LinkedIn Jobs. Hire right the first time. Post your first job and get $100 off towards your job post at linkedin.com/twist. /twist. Hello and welcome back to TWIST. My name is Alex. Today is a venture capital roundtable and I have brought 3 of the smartest minds in the entire world of VC right here. But today is Wednesday, May 6th, or as Jason says, A-Oh-100. So we're now 100 days after we all got Claw-pilled. Now before I introduce our guests, I do wanna say that we here at TWIST are absolutely obsessed with our Applaud pins. We've talked about it on the show ad nauseam for the last couple of months because we can't stop using them. If you are someone who takes notes, talks to themselves, goes through a lot of meetings, does a lot of phone calls, I cannot say this enough, you should try out Plaud. They're fantastic. And if you go to plaud.ai/twist, use the code TWIST, save 10%, get yourself a Plaud NotePin S. Jason's obsessed and thus the entire launch team uses them. All right, thanks for that. Guys, glad to have you here. In one corner, we have Michael Eisenberg from Aleph VC. Michael, how you doing?

Good, Alex, how are you?

I'm good, man. We also have Mike Granoff from Maneve. How you doing?

I'm great. Thanks for having me, Alex.

And we sent you that mic and it made it all the way from, you said LA to Israel to London. Well, why are you taking our equipment around the world, man?

Well, it's hard to send things into Israel on a timely basis and I happened to go to LA, so you sent it to me there and then I went home and now I'm in London. So I've enjoyed traveling with your microphone.

I'm very glad. Please treat it well. And then last but definitely not least, we have Larry Covert from Oxcart Ventures. Larry, you are in Austin, Texas. How are things down south?

They're, they're great. They're great, Alex. Thanks for asking. Uh, we got a storm moving in and, uh, I think we're going to be okay though.

All right, well, if you lose power in the middle of this, we'll just pretend to know what you were going to say and say it for you. Listen, I want to start with an amazing statistic that blew my mind when I saw it. Uh, 5 U.S. firms captured 73.1% of all LP commits. In the first quarter of this year. For, for context, everybody, it was 12 firms and 75% back in 2025. So we're seeing from an already concentrated world of venture capital, an increasing concentration even of what we had before. The next 10 firms raised 15.4%, and then every other fund in the US in the first quarter raised 11.5%. I'm curious what impact this amount of venture concentration has on founders themselves and how they should react to it. Each one of you is behind a fund that has 9 figures of AUM. And so I would say traditional size venture capital funds, not the Andreasons of the world. So, Mike, starting with you, one, is this bad for the venture capital industry? And then how is it impacting founders in your portfolio?

I mean, in aggregate, I think it's not unlike most industries that you get into waves of concentration and then the people see an opportunity to establish niche operations that have special purposes, and then you see proliferation of more. And then I think these are just cycles that most industries go through, and venture is no exception in that regard.

So it's kind of the old riff that all technology is just bundling and unbundling. That also applies to the money behind technology, I suppose?

I think that's true.

When does the wave turn around then? Because if it moves in cycles, we're going to see the other side of this at some point in time. So Is this a 5-year cycle, a 10-year cycle? I'm just curious about how much money you guys can put together in the next, you know, couple of years and if you can compete with these funds that seem less price conscious.

Oh, we don't have to compete with them in size. We, in our case, you know, we, we have a very particular specialty in industry and we look for companies that are around those families of industries that we specialize in. And it's not just about the companies we invest in, but our LP base is more than half strategic investors in automotive and tier 1 suppliers, fleet management, energy, infrastructure, insurance, and so forth. So, you know, we look for partners both on the LP side and on the company side that fit our niche. And I'm very comfortable with where we are. We do not aspire to be in the world of Andreessen in terms of AUM.

So you're not really excited about building a startup index fund, I take it? Not that I'm being— when in doubt. Michael, same question over to you about how this impacts both your firm and your portfolio companies. Now, I think you're a little bit less industry-based on the LP side, so a slightly different kind of mix of capital on your, on your cap table. But yeah, venture concentration, totally nuts. How is that changing your life?

A couple of weeks ago, I tweeted and said at our annual meeting last week that somebody had wrote on Twitter that this is the era of consensus capital. I think that's true on the venture level. It's true on the company level as well. And at Aleph, we're the opposite of consensus. It's an equal partnership like Benchmark, where I was before. We debated it out heartily, and we're not investing where everyone else is. You won't find us in GenAI, or by the way, even though we're in Israel, even in cyber, we've never done cyber. And we think that venture capital as a whole should be a craft business. Now, To directly answer your question, I think we may be in one of two moments right now. Uh, we may be at the end of the venture capital industry and we may be witnessing with this concentration of the capital, basically the end of a, of a 50, 60-year run of an industry or a craft business. And that won't be the first time that a craft business has gone extinct to scale. And If that happens, it will be sad because there'll be less money over time to, to fund way out their innovation because it'll just converge around the consensus. I think that's one option that we see. I think the second option is that we're in this wave of massive consensus capital. And I've been around long enough that I remember the bubble, the dot-com bubble. Jay Cal, who's not here, also remembers that we were around at the same time when he was at Silicon Alley and I was at starting venture capitalist. And one of the things you learned back then is all sorts of funny monkey business went on. And Bill Gurley's highlighted this, which is, you know, venture capital funding customer contracts, funding, you know, down the road. And we see this, you know, both around, you know, purchasing of AI, you know, of chips, of GPUs, like we saw purchasing ads before. I invest money in your startup business, you buy ads from AOL, and they take that and invest equity in your company. And we're seeing the same thing. And that allows kind of the flywheel of money to spend for a very long time. We had a massive boom, then kind of cleaned out to Mike's point earlier. I don't know which of those two is going to happen, but I think one of them will. And the last thing I'll say is we also are now finally investing in businesses and technologies that require more startup and upfront capital, which is some of these hardware businesses or defense tech businesses. But we're like engorging them with capital at the same time. And that always leads to bad behavior. And we're big fans of capital efficiency and not building too much preference on the cap stack.

All right, well, that's been a great show. I think Michael covered it all, so we can just wrap up here and bounce. Larry, we're gonna get to you in a second. I really don't mean to put you at the last of the line here, but Michael, you said that venture capital as we know it might be dead. I think I know what you mean by that, but can you unpack that for us? 'Cause I think there's a lot inside that that we're gonna wanna touch on. So give me the one paragraph version of that idea.

Venture capital works because of asymmetric upside to a small number of companies after you've taken a pile of bets. When the bets get too big upfront— and bets is not a gambling term, it's like you buy asymmetric options. Sure. Uh, on extraordinary human beings and individuals who develop extraordinary innovation. And if you start to engorge too many of these companies, the losses become too big and the big gain can't cover them. The more important part of this— let's go back to the question you asked in the pre-show about the IPOs coming, which maybe we'll touch on— is the ownership levels have sunk in so many of these businesses. In our firm, we learn this from LPs, we're at more than double the ownership of most of these, you know, engorged venture capital firms and engorged companies. Ownership matters in asymmetric businesses. If you can't maintain the ownership, it's going to be a problem. You pile up the cap stack, you take dilution, and then oops, the IPO doesn't deliver. It goes down after the IPO, if there is an IPO. And it becomes, it becomes trouble. So I think the math may just cease working in the venture capital business because of the capital engorgement and the capital concentration, the consensus nature of it. And then the outliers, you know, they just out over here where probably Larry's playing and Mike's playing. But for most of them, it won't.

Adding a new member to your team is a crucial decision, and you don't want to rush into a hiring situation that you will regret. But if you're a busy founder, you also don't wanna spend a ton of time in the trenches looking for that perfect candidate. Instead, you need a partner and you need a trusted partner. And LinkedIn Hiring Pro is that partner. How do I know this? Because I use LinkedIn Hiring Pro. You want a real world testimony? All right, here you go. We recently hired a new customer success manager. They handle all the advertising accounts here on This Week in Startups, and they ensure that we're keeping all of our wonderful partners happy. LinkedIn helped us connect with an amazing candidate right here in post in and he had exactly the combination of experience we were looking for. It's magic, it's alchemy, it's everything for me to find a great team member. So hire right the first time. Get started by posting your job for free at linkedin.com/twist. Terms and conditions apply. That's linkedin.com/twist.

Okay, so Larry, I was gonna ask you about venture concentration, but first respond to what Michael just said because I don't think I disagree with anything of it, But I can't quite get myself to the same level of polite pessimism. So I'm curious what you think.

Polite pessimism. Well, essentially, I believe it's become two asset classes. One would be consensus VC, CVC, and then TVC, traditional venture capital. And we're certainly more focused on the traditional venture capital in the spirit of Xerox PARC, you know, the old days when you made that bet, as you noted. On an N of 1 founder doing something that was truly fronteering and pioneering in the form of a step function. And we believe that is the true heart of venture capital. There certainly is a place for this CVC, this consensus venture capital in the United States across the institutional investor cohort who never wants to have to apologize for taking a chance on a new manager. They're happy investing in the A16s and the larger firms that have a legacy behind them and may not have the returns of a traditional VC, but they also won't have to apologize for making that bet. So what we ended up doing was not even visiting the larger cities. We just went to what we call the Southern Smile. We went from Southern California to South Carolina, visiting very large institutional-sized families who don't normally get access to true venture capital. And, and that, that proved to be a rather speedy and worthwhile effort. And we were happy to bring a normally inaccessible asset class to these, these investors. It's been so much fun.

Are those institutional-like family offices more risk tolerant than the endowments and the pension funds, or are they less risk interested? I'm just curious about how that might impact how you make wagers at Oxcart?

Yeah. Um, what we found is they typically don't have much, if any at all, of a venture capital allocation. So this is new for them and, and they know that it's a higher risk place to put the money with the asymmetric return capabilities that the rest of their portfolio doesn't provide.

All right, Mike. Uh, I know that CVC usually means corporate venture capital, but now we're going to call it something else. Consensus VC, I believe Larry said. Your take on the death of venture or perhaps its evolution into something that is just a little bit boring, like vanilla ice cream.

You know, everything evolves all the time. And yeah, I like the way Larry put it there, that really there may be this divergence. I don't think that the demand for medium-sized firms is going to disappear. In fact, it probably will increase as, you know, it's difficult to make returns work at a very high level. And then it's also, I think, difficult for founders to, you know, work with enormous behemoths of venture firms. You know, we're on a WhatsApp basis with our founders, like on an hourly basis. And I don't know if you get that kind of access when you're, you know, in such a— So I think there's, on both sides of the equation, it's going to be a demand for something more modest than, than the large-scale VC. And even though I'm sure they will continue to raise money and thrive, I think those of us in the more modest world will as well.

Yeah. Michael, Aleph really stresses, you know, hands-on, high founder-touch services, kind of what I might call like the old-school venture capital value prop. Like, it's not just capital. We're going to help you build, we're going to connect you and so forth. Does that actually get people to take capital from you versus one of these major funds that is a little bit less hands-on? A data point for everybody. I forget where I read this, but I think the Andreessen crypto team alone has like 80 people doing what? But the point is like, you're probably not going to be on a WhatsApp basis with your partner in that case. So how does Aleph use essentially high touch to win deals? And do founders care about that today?

We have the fortune of being in Israel and away from Silicon Valley. right now. So although Andreessen is flying in, I, I just literally read 2 hours ago, they're flying in in early June with, I don't know, 30 founders or 30 investors. I'm not sure what, uh, to Tel Aviv. Uh, we don't, we don't compete exactly on that same, on that same basis. But, um, to your point, uh, I think if you have a, uh, founder in a foreign country, they need access. And access is a high-level game. It is not dialing for dollars or dialing or cold calling for connections, although cold calling's got a place. And so, you need people with access. And I think, you know, what we provide at Aleph, what I try to provide is a network of people. In fact, one of our largest investment of our management fee at Aleph is we built this ridiculous database. Mike is nodding because he knows about it. That before people were talking about AI, we were using earlier versions of it to traverse multiple networks and connect our founders to do that. But ultimately, the call or the email has got to come from me. Or one of my partners. And we are, of course, a partner-only firm. You know, like, we're out— and I learned to benchmark when I was there. But I think the bigger point is the following, and Mike Granoff kind of mentioned that a second ago, which is, you know, Thrive. And I have incredible respect for Josh Kushner. He's somebody I love dearly and have incredible respect for. And I think he's done an incredible job. Now, last week he bought— not but— he bought a piece of the San Francisco Giants. In this Legacy Capital, and that's an outgrowth of a venture fund. And, you know, we're looking at venture deals together. We are. But if you can deploy— I'm making up the number, I have no idea what it is— $200 million into— sure, into the New York Giants.

Um, San Francisco.

San Francisco. Sorry, the New York Giants used to be around, you know, once.

Some of us—

that's how old I am. Football, Tampa.

Oh, it's true. Yeah, that's all I am. But, uh, the, uh, you know, that, that's just a different business. And, uh, he's made a hell of a business out of it and he's doing an incredible job. I'm just not sure that's the same business of venture capital that we're all, you know, Larry's point, uh, what, what we all grew up on.

I was always told that your fund size is your strategy in terms of how you invest, how you conserve capital, how you keep track of dry powder, think about exit timings and so forth. And so, you know, maybe we should just have essentially a fund cap size for VC. Like maybe it's $500 million and below, and everything above that is, you know, to Larry's point, CVC, or essentially just like pre-IPO money with less information rights. Like, I, I don't know, man, call it what you want. Larry, what would you call that, that back half if you couldn't use CVC as the moniker?

Now we have to come up with something clever better than CVC since corporate venture capital's adopted it, but it's definitely consensus VC versus traditional VC in my opinion. And traditional VCs, you know, what I've been doing for 30 years, and it's the most fun thing. It's my seat on the bus. I love it. And it's that touch with the founder, that very handsy, helping kind of approach that makes it the most fun of all. We found a few guys in Grand Rapids, Michigan who had an edge AI technology completely ignored by Silicon Valley. And we engaged at $70 million valuation. And we've gotten them partners, customers, helped them hire. Now they're here in Austin, Texas with their name on one of the tallest buildings in the city at a $2.5 billion valuation, uh, led by Time Ventures, uh, in January. So we're just so excited for these guys and, and, uh, to build something meaningful with great people and go from 5 people to 300 in just a few years. And to me, that's traditional venture capital and nothing against that larger asset class. There's absolutely a place for it. It's just not my seat on the bus.

Hey, here's my proposal, Alex. You ready? Yeah. CCC. 4 Cs.

4 Cs stands for?

Consensus Colossal Collaborative Capital.

Consensus Colossal— wait, do it again.

Consensus Colossal Collaborative Capital.

Good. I'll get that tattooed by next week and we'll see how it does in the market and see how it tests. Maybe we can get a billboard, you know? No, I do think that as we've pushed the IPO window back, to the teenage years of companies. We've extended the venture capital time period as well, but I don't think that venture capital fits for anything today, Series C and beyond. I mean, if you're raising a quarter billion dollars, great, you went public without listing your stock. I, but I don't wanna be a hater and say that we can't change things, but it has been a different flavor. Did you guys see the news that AngelList is putting that USVC, fund together to allow retail investors to buy into certain venture allocations through their own network. Did you guys see that?

I didn't see that.

Oh, really? Basically, they're putting together an unlisted closed-end fund that anyone in the US can buy into for $500 minimum. And the idea is to get retail investors to come in. And what this could do, I think, because part of USVC strategy is to take that money and to deploy it to emerging managers, perhaps like Oxcart, and that it may increase dollars flowing into smaller and mid-sized funds. And I was going to ask, could that help your firms and other firms of similar size have access to more capital? Larry, is that a problem that you need to solve, or am I trying to fix something you don't currently have a struggle with?

The world's greatest moderator and the world's greatest angel investor needs the world's greatest solution for managing his funds. So I want to tell you about PaperOS and why we use it here at our own companies, Launch and The Syndicate. PaperOS has helped over 10,000 founders and investors, including me, automate their workflows. Their tools will help take you through handling the operational details I don't want to worry about because I'm busy meeting founders and looking for great companies. PaperOS has made the onboarding process so seamless for limited partners. Their investor intake forms, automated subscriptions, it just takes all the busywork out of managing our syndicate. And it automates capital calls, investor accreditation, tax filings. These are the chores of running a fund or a syndicate. And they have saved me and my team hundreds— actually, now it's probably thousands of hours. Our friends at PaperOS want to give you a complimentary $10,000 credit. Go to paperos.com/twist, and if you prefer price discrimination, you can go to paperos.com/diskrazia.

I mean, we don't have a struggle with it. We have very few investors, and we're just really careful about, you know, who is, who's part of the family. But I want to I imagine it could be quite useful for a firm that wasn't as discriminating about, you know, who was in the LP base for their fund. Yeah, it could be a really powerful tool.

I think it's important to acknowledge a very uncomfortable truth about the venture capital business. And the uncomfortable truth about the venture capital business is most funds don't make money. The index doesn't make money. And so there is a small slice of the venture capital business, the top quartile for argument's sake, that makes money that's worthwhile to invest in this —call it riskier asset class. I'm not sure it's an asset class. It's riskier craft business. And I'm not sure that what you call the mid-market or the small market or traditional VC or whatever we're going to call it, getting money from doctors and dentists puts more top decile or top quartile venture funds into business. It's not clear to me that that's an accurate portrayal. My sense is that guys with craft venture capital firms who are good at this can raise money from institutions. You know, we are, and I'm sure Larry is and Mike is. And so this is not an— this is not an area of investment that benefits from additional capital necessarily. In fact, I think the opposite's true.

So this is a question that I wanted to put you guys later on, but we'll do it now. Are there more breakout startups being built now than in, say, 5 years ago? Because to me, there would be more room, Michael, to deploy capital into investments that could have fund-level returns, venture capital-level returns, if there are more total companies that are that good. But if we're seeing capital increase, but the number of breakout companies stay static, then I feel like we're going to have price inflation, dilution of your stake, as you mentioned earlier, and so forth. So are we seeing actually more breakout companies today And is that enough to absorb the greater amount of capital flowing into venture?

The complicated answer, generically the answer is no, there are not more breakout companies, but okay. The, the, the fine point is the following. The more capital that comes into the system, there's more competition. We learned this in the bubble, you know, in whatever the area that the unique entrepreneur that Larry described earlier is in, and that reduces overall returns in that area. That startup or unique innovation is pursuing. So, actually, the increase in capital will reduce overall returns for the industry. What I do think we have right now, because it doesn't matter if you call it a bubble or not a bubble, everyone, the narrative is way ahead of the results. There are some companies growing at extraordinary rates. And at the same time, there's 50 companies funded in each category because it's so easy to gin up these companies in many categories. So, over time, you will have decreasing outcomes to scale of capital focused on the same area. And I think that's what we see right now because the valuation is so large. Everyone is kind of, even though they're illiquid, everyone is kind of fooled that there are so many more outcomes right now. But I think in reality, what we're building is a lot bigger losses down the road in many of these categories. Look, we've seen a lot of companies go public and their stock dropped 50%, 60%, 70%, 80%. That tells you something. A market where a fund, private fund measures it is a valuation determined by two people. Market with a public val— you know, values it tends to be a true valuation.

Yeah. Larry, can you jump in on this? I'm curious what you think about what Michael just said. Yeah.

Well, I was thinking about the definition of the word breakout and, you know, the rule of numbers applies here, as you noted, Mike. Like the more entrants into a market, the capital is spread out. And it's harder to be that, that big breakout unless you truly are unique, uniquely qualified and uniquely positioned. So I can't disagree with what you said there. We're focused on national security and there's just obviously been way more interest in national security over the last 5 years than probably ever, maybe not since early Silicon Valley, but since then. And they're not all going to win. And what we tried to do is say, what are the non-obvious places to put money related to national security? They don't all have to be sharp, pointy, and explode. And there's many infrastructure opportunities where it's less crowded and less hyped up, where you can still contribute to national security without paying, you know, overpaying rather for a position or competing for a place in the cap table. That's kind of our way to hedge that.

We're going to get to some of your portcos in a second, but I want to get, I want to get Mike in here because ElevenLabs just added a bit more to its Series D. I think it's now a $550 million round, but at the same time, the company announced that it has crossed $500 million ARR, up from $350 at the end of last year. Pretty insane growth at scale. In the old days, this would have been an IPO already. That's why I asked if there's more breakout companies, because it feels like the companies that are truly scaling on the startup side are getting so much bigger, so much more quickly that I wonder if that makes the math math a little bit more. Even if to Michael's point, there aren't a greater number of these companies as they get bigger, maybe they'll have enough returns to make everything settle out.

Well, I think what we've seen there is there are more companies for sure. Deal flow is always increasing. And in terms of breakout, AI may have distorted this metric for a period of time because there are just, you know, so many enormous companies in that space that have received unprecedented financing and are seeing unprecedented growth. But at the end of the day, will that portend a long-term trend for venture? You know, probably not. Probably return to the mean. So, you know, I think we might be in a moment where it's very difficult to assess whether what we're seeing is really a harbinger of things to come, to note one of our portfolio companies, Harbinger. But it, you know, I think I think it's, it's difficult to take a static picture right now and, and make judgments.

Michael, would you buy into ElevenLabs at like a $11 billion valuation today, or is that one of those companies that you think, uh, will come back down to earth in time?

I know nothing about this company. I think Benchmark's invested there, uh, which gives me a lot of, uh, comfort. Um, but I, I know nothing about the company, so I won't say about invest, but I, I will say that revenue is a crappy metric because it's a proxy for gross margins. And the other thing that nobody wants to talk about is that AI company gross margins are not software margins. There is a cost to those tokens. There is a cost to that compute. And I can't tell you how many fast-growing companies I've seen that have mere gross margins. Now, I am of the belief that gross margins are talked about way too much in early-stage startups. And when you grow up and get bigger, you can actually manage those margin problems. And probably the cost of compute will go down, but maybe not. I kind of think we're looking at surge pricing for compute going forward because otherwise you won't be able to manage the grid. And so you think about that and you think about the applications, you have to kind of think two levels deep. Okay, is this application is going to have to be during surge times, which means it's going to have surge pricing, which means it's going to eat in the and the gross margins and you go, I don't know, you know, and we'll see whether Wall Street has a similar view of some of these companies. You know, by the way, I'm old enough to remember also there were a lot of companies in the late '90s that were go public or go bust. And we may have some of those too where, you know, they need the public market capital in order to ensure they don't go out of business. We may have some of those now too. It's just, it's impossible to know.

We haven't seen the financials. Piggyback on what Michael said with regard to energy as a limiting factor. You know, I've been following batteries now for more than 20 years. And it's interesting that the decline in the price of energy storage alongside the decline in the price of PV, of solar, is leading to a situation in which people are going out now, especially in Larry's part of the world, and saying, you know, we don't need the grid. Like, we can create virtual power plants or we can create specialized energy sources for a data center, for a manufacturing facility. Facility off the grid. And you know, I think that's, that's as, as energy becomes the clearly the, the limiting factor in compute and, and therefore in, in growth in a lot of ways, then I think we're gonna see more and more innovation around that and more scaling of these types of systems and really transform the energy world, which is, has profound implications for all industries. Here's a startup truth bomb.

A lot of founders have no idea what's actually going on with their money. If that's true of your company, hey, no judgments. I know you're busy hiring, building your product, go-to-market, all that important stuff, but your company needs a reliable financial partner, not a lifestyle brand, okay? Grasshopper is a real federally chartered digital bank that's not trying to win you over with a rewards program. Instead, they're building deep integrations, treasury products that are gonna actually help you expand your runway, and innovative tools like an MC-based AI connector. Oh man, that's awesome. You can connect it to all of our agents and do reporting. And that will put you in command of your money. As a TWIST listener, you're gonna get $500 cash bonus just for opening an account. Think of that. You open an account, boom, there's $500 in it. So leap on over to grasshopper.bank/twist and use the promo code TWIST. As a TWIST listener, you're going to get a $500 cash bonus just for opening an account. grasshopper.bank/twist.

Fun fact, there's a Turkish boat off the coast of Texas. I think it's the southeast corner of Texas that is an energy-producing boat. Uh, you can go see it.

Is it like a, like a tanker that's a power plant?

Yeah, it's not— it doesn't hold oil. It actually has a power plant on the boat and provides surge power.

I never thought I'd get to the point when humans were so power constrained that we'd be talking about beaming solar from space or making power boats, but I guess this just goes to show that the demand for AI compute is as high as we say it is. On this exact note, we did have, um, BasePower on the show October 8th, 2025, if you want to check that out, listeners. uh, that are with us. And also today, a company called Span, they make smart, um, electrical panels for your home, announced a thing called, uh, XFRA, which is a distributed data center, essentially taking the base power model of having in-home battery storage for off-peak power and then putting GPUs in your house in partnership with NVIDIA to create a scaled compute network. No idea if it's gonna work, but I think it just goes to show how demand is for those products. Uh, Michael, looping back to the AI gross margins point. Did you see that Cursor? I think the information reported they had like negative 23% gross margins for a while and they were growing so incredibly quickly. Is that the, the rare case or is that something that's actually happening more frequently when I see these headline ARR numbers come out from companies?

Look, we, we need to get a look at financials and in order to know this, and each company is a little different, but you know, what we can say is like The hyperscalers are investing money in— look at all the Anthropics, the Cursors, you know, and now Elon's doing a deal with, with Cursor, um, which I think makes a lot of sense for what it's worth, uh, for X. But, um, and, and they're funneling that money in, they're buying chips with it, and they're deploying them at the hyperscalers. Um, it's hard to know what the gross margins are, by the way. By accounting standards, some of that might be negative revenue. I think a lot of venture people have forgotten how to, how to do these financials. And so, uh, it's really, it's just impossible to know. But one thing is clear is that the cost of compute has not yet approached the cost of energy, and the cost of energy has been rising. That's tied to the war in Iran also. But it's just, there's a shortage, and it takes a lot longer to lay copper and pipes and build nuclear and all these things than it takes to, you know, to brew semiconductors. And so this is like where the rubber meets the road. How you connect the pipes, you connect the fiber. It's like This is real physical labor. Do you have enough electricians and enough HVAC guys? It's like hard. And so I expect actually the cost to kind of keep on cooking. And, you know, the point you made earlier about like the GPUs in the homes, this is like so reminiscent. Again, pity Jay Cowell's not here, but back in the '90s, there was a company called— or a product called SETI@Home. I don't know if anyone remembers. Shout out.

Yeah. No, no. It was the screensaver that would number crunch to find signals of extraterrestrial life, right?

Exactly. Yeah, exactly. So here we are. It was like Back to the Future again. That didn't make it after some period of time. I don't remember why, but it's like, okay, now put GPUs in people's homes, let's run SETI at home, we'll search for extraterrestrial life, we'll crunch some numbers, we'll do some matrix multiplication. I don't know.

And so, but on that point, like Intel, Snapdragon, AMD, they're all shipping chips right now inside laptops that do 50 to 80 trillion operations per second on 15 or so watts. Like, it's not like we don't have compute in our fingertips right now. And, you know, all of this is probably going to be the answer. We're going to need all of it. We're absolutely insatiable when it comes to compute demand. And it's going to require, you know, a tremendous amount of innovation at each one of those levels, at the grid, at the management of the grid, at the generation. At the home and it's going to be a lot of fun. There's a lot of opportunity there. The durability and the cost of each are questions for me.

Larry, Michael was talking about how compute prices could be essentially surged when everyone wants to use power at the same time, could stay very expensive. Your firm works a lot in kind of the dual-use space, I might say. And I'm curious on the more hardware-focused side of things, are you seeing your portcos struggle as much with their AI costs as perhaps some of the more software-focused startups are? Or are you as exposed to, I don't know, let's just say Opus 4.7 pricing as anyone else is?

Well, very interesting. So when we first started the firm, our first investment was WebAI, which is edge AI. 75% of a business's generation of data and usage of data is actually on-premise. And so they end up doing a lot of air-gapped work in the public sector, but also a lot of work. Thank you. Also a lot of work in the enterprise, like commercially. So that was our first investment. And now our most recent investment, as of just 3, 4 weeks ago, is a magnesium company. So we went from bits to atoms, and this company uses electrolytic processes to generate magnesium from seawater. So they couldn't be any more different. They're alike in that they're unique, N of 1 companies led by, you know, extraordinary founders, but that's where the commonality starts to diverge.

And you're talking about from bits to atoms. Magrathea, right? Yeah, that's right. Who I recently spoke to. And a little teaser for everyone who's listening, they're on the show on Monday. So if you want to learn more. Yeah. So if you want to learn more about how we're going to yank magnesium out of seawater and not torch the environment while we do it and take down China's metals advantage, well, tune her up because I'm going to pretend to understand physics for about 25 minutes. On Monday. It's gonna be, it's gonna be a good time. Uh, Mike, I do wanna ask you kind of the question that we're talking about AI and margins and spend. Uh, an NVIDIA exec recently said that some people are spending more on tokens now than humans, kind of an inversion. We thought that AI was gonna save money. It can actually be kind of a cost center. How are your portcos handling rising token costs and is it changing how they're investing and deploying AI?

Yeah, you know, I mean, it, it varies. A lot of our, our companies are also, um, early on in their, uh, cycles, but You know, there are— I have seen cases in which, you know, projections for headcount have gone down over time because of the agents. And yes, there's also, you know, new line items around tokens as well. Right now, it's hard for me to say which is the more profound trend, but, you know, All of us are in the progress business and we know from history that these waves of technology ultimately deliver more value and deliver more opportunities for everyone. But there are disruptions in the shorter terms in the transitions and it's very difficult to manage those sometimes and can have a lot of implications for our industries and beyond.

Your firm has done a lot of work with chips for the automotive world. We talked about Halo actually in the pre-show. H-A-I-L-O. Has the, the race to build more chips globally leading to essentially TSMC and other companies having a hard time keeping up with capacity? Has that harmed the ability of people that are working in the automotive space to build new things? I'm just curious if essentially Nvidia stole all the fabs and has left other companies bereft.

Well, it has not slowed down competition in the chip sector. And in fact, aside from Halo in our first fund, we have two other chip companies also based in Israel in our third fund. One of them is in stealth. So, okay. One of them is called Neologic. So, but, but I think this— what it's showing is, is that there's still a lot of room for innovation. And then when it comes to scaling, yeah, obviously the NVIDIA effect is, is is pretty, pretty, pretty having a big impact. But when it comes to automotive in particular, and our chip companies are not specific to automotive, but they, at least in Halo's case, it was autonomous driving that was kind of the template on which they built the original chips for edge computing. And, you know, the auto industry, that's a whole other direction of conversation, but they're, education, I should say, the traditional auto industry in technology and in particularly in semiconductors has caused, you know, lots of disruptions over the last decade. And that's a story that's continued to play out, and particularly when it comes to China.

On the chip front, I want to go off topic here for a little bit, off-piste, if you will. We're seeing all the major companies build their own chips, talking about, you know, the latest TPU generation from Google, Trainium and Inferentium or whatever from Amazon. Meta wants to do chips, OpenAI wants to do chips, uh, Etched, a lot of other startups are doing ASICs and so forth. Do you think that there's going to be something that actually replaces the NVIDIA GPU as kind of the core beating heart of the AI economy in the next couple of years? Or are we talking more about like serious players, but to the side of the main show, which is NVIDIA?

A couple of things. I actually just released a podcast on my podcast today with the CEO of this company, Next Silicon, Elad Raz, who says that NVIDIA chips from generation to generation are scaling compute and power linearly. And so somebody will need to break it with exponentiality, and they intend to do that. And they're already showing those results. They're in Gen 3 of their chip, which is coming out, which is coming out now. So, you know, these things never last forever, but I think kind of chip leadership lasts for a very long time. A very, very long time. And, you know, Intel is even making an incredible comeback right now. You look at that stock, it might be the best performing stock in the world over the last 6 months. It's unbelievable because chip design and chip manufacturing is really, really, really hard. Yeah, really, really, really hard. And, you know, by the way, interestingly, one of the things I've said a few times is there's like 6 countries in the world which have the full stack of AI. Uh, the United States is one. That's like semiconductor design, semiconductor checking equipment, and, you know, semiconductor manufacturing, uh, you know, tokens and call it, uh, AI applications. So that's like the United States, uh, China, obviously Japan, uh, South Korea, um, and Israel. And a little-known fact, a lot of the CPUs and TPUs that you describe right now, that are at Amazon and Google and Apple in particular, are designed and built in Israel. A significant number. And Nvidia, by the way, you may not know this, but Nvidia has like 3,000 or 4,000 people in Israel, growing to 6,000 now. The acquisition that enabled Nvidia to scale GPUs to data centers, which is what we have right now, is a company called Mellanox, purchased in Israel. Started by a guy named Eyal Waldman, who unfortunately his daughter was killed on October 7th. In the terror attack. And this is what's going on in the semi space is unreal right now. And I think we're going to see ever more competition. Why? Because a lot of money there. And so I think CPUs we're going to find on the edge, more GPUs at the core. And I think going back to the point Larry made earlier, which I think is spot on, is the AI model companies that are going to be successful or kind of small model companies Um, we'll figure out how to make access, uh, make use of the CPUs, uh, at the edge in your laptop or on your phone, as the case may be, uh, more than the GPUs, uh, at the data center.

My own bet, by the way, is very bullish on Apple because of that, because they have some of the best consumer CPUs in the market today with the M series chips.

I'm reading, uh, and they're close to the data on the edge, the personal data on the edge.

As long as the models can get small enough and, and strong enough, do you think distillation will allow people to get models small enough and good enough to actually pull that off? Because I think today's smaller models, even in the more recent Gemma 4 family, I believe, are fine, but they're a pretty material step back from the edge.

Michael, I think they— you, you'll see more. It always happens this way, right? You've got these kind of general models, and then you'll get the specificity. And so I think the models will get smaller, virtual specificity. The company Mike and I are co-investing in, this company called Nexart, released a model called Badass, Badass 2.0. It beats, it beats Cosmos by NVIDIA by like 2.5, 3x. And it's, I think, an order of magnitude smaller in model size because it's trained on real-world data. Training on simulated data or training on digital data has been highly inefficient. And I think we're going to find more efficient ways to do this. And by the way, these guys, Nexar, started on the edge. They literally started by putting dashcams next to cell phones on people's dashboards. And badass is a badass. And, and it's kicked the crap out of Nvidia's Cosmos model, which I'm very pleased to say right here on your show.

I'm glad you put Jensen in his place. You know, let's, let's bring that guy down a peg.

No, no, he's the greatest, that guy. He's the—

anyone who wears a leather jacket in the sun is suspect. He's— I'm not—

what a mention of star that guy is.

He's great. It's great. I was gonna say 50 years ago, 5 megabytes would fit in the back of a tractor trailer, right? And so I think you're gonna see the same thing as Mike, you noted, it's a forcing function that this constriction with GPUs and cost of GPUs is a forcing function to push models down to more widely available and frankly much more widely distributed compute devices and use cases. It's a good thing, but again, we'll need all of it for sure. But I think the vast majority of the models will end up being in your hands, so to speak. And for AI to truly— Jensen actually said this— you have AI, then you have AGI, and you will not achieve ASI until we accumulate the billions of models, these nano models, if you will, into a super layer of synthesis. And then you can actually trust AI. Right now, yes, you can use it to help plan your week or go to the grocery store or write a PowerPoint to 80 to 90% trust. And then you got to go make it judgment-worthy. We can't use it to make any true, meaningful, high-stakes judgment calls yet. And we won't be able to do that, I believe, until we synthesize these trusted models, which are personal models. And once everyone has the ability, particularly experts in whatever chosen field, can make their model at 100%, we're just never going to be able to trust it for high-stakes judgment calls. And that would be ASI. It is ahead, though.

Larry brings up an interesting point, which is like public trust in AI. And I think we're seeing that actually begin to play out in real time in a in a way that people can really understand with their own eyes, which is in the world of autonomous vehicles. And if you look at the statistics, for example, on Waymo and the safety statistics, the conclusion I think you have to come to is, you know, at what point should government simply mandate autonomous vehicles? Because it is the most— it's proven to save lives better than almost any medication that's ever come around. Preach. Preach. Yeah. And, but I, but I think that will kind of be, you know, there's an Israeli professor, Dan Ariely, wrote a famous book years ago called Predictive Irrationality. And I've always said that the most predictably irrational consumer is the mobility consumer. And, and, and, and it has a serious edge when it comes to the world of AVs because, you know, I, when I was in LA last week and I had Dinner with a cousin I hadn't seen in a long time. I said, you know, how do you, how do you like Waymo? And she says, oh, I'm afraid to get in. I said, you're afraid? It's like statistically I can show you that you're like 10 times safer at least in there than you are in any human-driven Uber. And, you know, I think obviously the, we don't know what the end is that you have to get to until the public has comfort in it. But it is going to be an interesting test case, I think, for AI in general.

Yeah, it's going to be a cultural conversation, I think, because so many people in the United States associate their self-worth with the car they drive that I think we're going to have a hard time breaking into those people to get them to step out of, you know, actually driving their own vehicles. But is it the same vibe in Israel, or are people there perhaps more open to self-driving than some people in the United States?

I'm just curious if there's a cultural difference. Great question. I think next week will be a very interesting week in that regard because Elon is supposed to come over for Israel's big mobility conference. And, uh, okay, and we'll see what that leads. Uh, Mobileye has obviously been one of the world leaders in autonomy and is right there in Jerusalem, um, and they've been testing cars, but there hasn't been any consumer AVs, uh, so far in Israel. And even the, uh, even the Teslas have not, uh, been permitted to have the FSD engaged. Um, so we'll see if that, if that changes next week. But, uh, is that a safety security thing? I I think, as Michael can talk about better than I can, Israel's kind of divided along this line of the incredible high-tech sector and then the socialist founding ethos of the state, which makes a lot of the rules and is sometimes very slow to come around to innovation. So it's kind of an interesting dichotomy to live in.

I'll say, I've always been a car guy, and Alex, you brought up people associate their net worth with cars, but I just like a great driving car. And it's a lot of fun, but my wife and I both got Teslas and we use the full self-driving all the time. And recently we were driving out to what we call the Hill Country out here, just outside of Austin. And it was about 1 in the morning getting to our Airbnb, cruising at maybe 75 miles an hour on full self-driving and the car slams on the brakes. I said a few cuss words, but then a deer, you know, lollygagged across the road, not one but two deer. Then another deer came by and it saved our life. And I believe they're at 10 billion miles worldwide with Teslas now. So, I mean, that's an extraordinary amount of data. I would imagine you can get very good at full self-driving and we feel very safe in it. But you're right. I think some sort of— I don't know about a mandate, but man, I wish it were greater adopted and we would save a lot of lives.

Yeah. Which is the point.

Well, no, I'm not sure you're right that it's the point. And this is, I think, the important thing. Here's an insight. Full self-driving cars don't vote in elections. And in democratic countries, therefore, it's very difficult to mandate these things. You know, Mike made the point about, you know, the kind of socialist origins that still kind of occupy some of the organs of state in Israel. We don't have Uber in Israel. You know why we don't have Uber in Israel? Because of the taxi lobby that they vote for the current ruling party. Yeah. Now, so we're going to have to like leapfrog Uber, self-driving, you know, we got to get all the way out there.

But it was like 12 years ago that Netanyahu went to Davos and met Travis and came back and said, we're going to do this here. And then like his party came to us.

No, we're not. Yeah. They vote. And so we underestimate as a democratic society the power of certain interest groups to prevent technological leaps. We also underestimate the power of the media to develop narratives. You just showed irrefutable data about the safety of self-driving. Yet you rarely see a news article or a TV broadcast about a car accident that, you know, one of Larry's friends was in. And you'll never see an article about the FSD and his Tesla saving his life. You'll see a ton of articles on the one time that the Tesla ran into a truck because there was a glare, because its dog didn't eat man, its man ate dog. And so the press will continue to be against this because they too are afraid of losing their jobs.

You know, you know, Michael, disagree with you there. Everyone's already been fired from the world of media. There's like 7 people left. So I think that that's already kind of run its course. The reason why I don't agree with this being the media's fault. It's because I think you're looking at a human condition and then looking at how the media represents it and putting the blame there. But when I talk to normies in my life about these very issues that we're talking about here, and I think we're all agreed that self-driving cars safer, better, going to help the world, uh, the average person feels much less confident letting a machine drive them than their friend who's an idiot. And I don't think you can put that bit of human stupidity on the shoulders of the media. When that's how the people who vote think. I think it's more reflection than points of derivation.

I'll take the other side of that. Now we have a robust debate, which is helpful, which is the media writ large creates narratives that open permission windows. That's the way it works. That's how the Overton window works. Yep. And these are permission architectures. And so you may be right that the average, what you called normie, um, is, is, feels more comfortable with a human driver than Larry's very safe FSD. But if we amplified the Larry story of not running into the goat instead of amplifying the Tesla car going under the white semi-trailer, we would create the permission architecture in which were more, as you call them, normies would feel more comfortable with this. Here's the crazy thing. We see a pilot go into the cockpit of an airline. They don't fly the damn plane. It's all flown on autopilot. Yes. Permission architecture is the following. I'm in the back and there's a giant wall or a curtain between me and the pilots. I got no idea and no one's ever told me. And so, all we need to create is the permission architecture for people to feel more comfortable.

It's interesting that we're talking about this because actually, the way that I think about self-driving here in the States at least is that it's been pretty open. I mean, It seems that Waymo's been able to go to most places they want to. And yes, I think Boston—

not New York. Not New York. Now that's the first place that— yeah, well, them. And that's, I think, also a harbinger of things to come.

Well, Mike, we got to 500,000 paid Waymo rides a week. Yeah.

Going to a million by the end of this year. It's—

oh, it's going to be one of the best companies of the last 50 years. I'm such a believer in them, but I've been really encouraged, actually. By how cities have been relatively okay with it.

I agree with you. I was just making the point of New York as the outlier here, which, you know, I think is in many respects the case.

Yeah. Well, I mean, going back to the point about lobbies in Israel for the taxis, I mean, New York City has a lot of entrenched interests, you might say, that all have their opinion. But, you know, let's broaden this one notch because people talk about, you know, AI doomerism and worry about job loss and, you know, worry about, you know, it going rogue and what's your P-Doom? And despite all of that, I think the last stat that I saw was that more than half of Americans use AI at least once a week at work, which is pretty quick adoption for a population of the US's size. So I guess I don't see as much of a problem, Michael, in the media narratives with the Overton Window as you're describing it, because technology seems to be advancing and being adopted at a pace that surprises me, frankly, in a positive way. So maybe I'm just more, maybe I'm just, uh, more of an optimist than you.

No, no. I'm a venture capital investor. I must be an optimist. I was kidding. But I'll ask you the following question. How many people will die in car accidents this year? Too many. If we had a federal mandate to greenlight FSD for everyone or insist on FSD for everyone, how many people's lives would be saved this year?

Many. Well, presuming you can get the technology rolled out in time and hardware and all that.

But why isn't there a national campaign in the media to do that?

You know, uh, I don't think that's the media's job. I mean, when, when my— when Waymo goes to Miami, there is national and local coverage of that and, and, and so forth. So I, I think maybe you're ascribing more motive to media coverage than I think there is with my experience in news.

I think Michael and I are on the same page here because you have a very different perspective on global media when you sit in Israel and, you know, come to London, as I did today for a board meeting, and have chats with folks about, you know, what they're seeing on BBC and so forth and compare it to my own reality. And I see their jaws drop when I describe the variations between the truth and what they are consuming. And so, yeah, I think that the media bias takes lots of forms, including oversaturation. And including undercoverage of truly enormous changes that could be for the better, like the safety of self-driving. Yeah, I enjoy that.

Half my friends think that The New York Times is a communist rag taking us all the way down the rabbit hole to the left, and the other half of my friends think that it's a reactionary conservative publication out to crush the nascent labor movement. So let's put media criticism aside. I want to talk about a couple more things.

Well, I'll just say Uh, in AI you have something called attention steering, uh, to make models more efficient. And yes, and that's what media is as well. It's the same, same deal.

Well, you, you do have humans watching and then deciding what to write down. Yeah. I, uh, I wish we had better funding for more media so we could have more voices talking about things in a more even keeled format, you might say.

But, uh, I would, I would just point out that there are some glimmers on the horizon in, in the media world. Like, for example, the Free Press, which I think is doing extraordinary journalism. And there's a new platform which I've become a little bit affiliated with called Two Way that has now 7 or 8 programs on, on, on YouTube and on other platforms and is taking a different approach to media that is interactive. So I— some of those things give me hope and I hope that they continue to proliferate.

So people can look it up later. What's the Two Way URL, just in case they want to find it?

The number two, Wei.tv.

We've talked about a lot of countries in the last couple of minutes, and one that comes up a lot in the AI realm is China. And a couple of people in Congress are basically haranguing some US companies for using open, open source Chinese models. Think about Cursor, which used KIMI-K 2.5 to form the base for their Composer 2 model. And then of course, Alibaba is well known for using some models from Alibaba's Qin team. And we're seeing other companies like Pinterest kind of work in some open models to their stack. I'm curious, let's go with Larry on this one. I'm curious if any of your portcos are using open models from China and if you think that's a concern or simply a great way to get quite a lot of intelligence at a relatively low price.

It's certainly a way to get intelligence at a low price for sure, and try to infer their own inference capabilities. No pun intended, pun intended, but they try not to, they try not to. I know one of our portfolio companies does for the research side and then testing and benchmarking and seeing where we are at adversarially from a leading edge perspective. We used to say of the top 60 critical categories of technology, we owned 90+% of those as far as having the leadership position globally. And now China has that. So a lot of our companies are well aware of that as part of their mission, and they're executing and they're doing well. But specific to Chinese models, I mean, you know, they have the ability to tell an entire village or town or city, for that matter, to just go do one thing and just focus on that. And then, and some of that's manufacturing that, You know, it was hard to beat them on speed and cost. Maybe you can beat them on vector, the right vector, but hard to beat them on speed and cost in that regard. And as far as AI, there's a lot of, a lot of theft, a lot of theft, and frankly, a lot of infiltration into some of our largest, most powerful companies in the United States. So it is a high-stakes environment.

I would say one of the most important books recently is by Patrick McGee on Apple in China. And really to tell the story of how one company's endeavor to find cheaper labor in China changed the entire trajectory of the global economy in a sense. And I— Tim Cook, I mean, frankly. Right. And I think we're— a friend of mine, Michael Dunn, is writing what I would call the sequel to Patrick McGee's book about Chinese cars, which just to give you a sense, in the US, when I talk about Chinese cars, I mean, you know, there's a bias because you don't see them. And so you don't think that it's such a big deal, even though it is pretty fairly well covered in the media. But in Israel, the first Chinese cars came in in 2022. And okay, we're now at a point where it's almost one out of every new two— one out of two new cars sold is a Chinese EV in Israel. And then just goes to the fact that the quality is there. And the pricing is just unbelievable, completely unbeatable for lots of different reasons. And, you know, next week, in addition to Elon coming to Israel, Trump is going to China. And I frankly wouldn't be surprised if he comes back with a deal on, on bringing Chinese manufacturers to the US. If it doesn't happen next week, I think it's going to happen within the next couple of years for lots of reasons, but not the least of which is that, you know, the algorithm, the same algorithms that promote some propaganda in certain ways to young people, are also now pushing the incredible technology of Chinese cars into the phones of young Americans. And they're starting to say, hey, where can I buy one of these things? And so, no, you can't. And I don't think that's a sustainable situation. So it'll be—

as a free trade guy myself, I'm absolutely on board with you. And I think that it would put a lot of fire underneath the company that keeps making larger F-150s as their only form of innovation to maybe do something slightly slightly different for once. Well, and I say that as someone who learned to drive a 250.

Ford is very cognizant of this. He drove a Chinese car for a year and he talked about how much he loved it. And he's— I think he's doing what he can, but it's very difficult given the advantages that these Chinese companies have.

Defense, Iran, drones, lots is going on here. Larry, I'm thinking about companies like Sironiq, Vatten Systems, a lot of people are making for drones, both for the air, for the, uh, on the water and below water. What impact has the Iran war had on your portfolio companies? And I would say interest in the government in dual-use technologies in general.

I mean, as long as someone else has an offense, we're going to have to have defense. That's for sure. So, so the, the industry has been around a long time. It will continue to be around. The, the most recent, let's go from, you know, Ukraine, that was absolutely a new epoch in defense investing and defense innovation, primarily on the kind of drone sector, if you will. They've been extraordinarily innovating, fronteering in Ukraine, using that environment as a theater of war of prototyping and proving out. Iran, a little different. A little different there. We're tending to use more of the legacy primes in that environment than we would be in more of the Ukraine environment. But both of them are lifting the sea, if you will. And what we're seeing is you've got legacy primes out there valued between $100 to $400 billion. And now all these new primes, these new entrants, you know, Anduril, you could argue Saronik at reaching $9.5 billion valuation in 2.5 years is quite a burgeoning firm. They aren't an N of 1 completely, but they are an N of 1 when it comes competitively at the moment. And what Dino's done is remarkable. So he has a capital moat, he's got speed, he's got credibility, and I'm really proud of what they've done. It's not a portfolio company, but they are local and he's a friend. So I got to give them a shout out. Really proud of those guys. I think you'll see a lot of zeros being added to valuations, much like these AI companies who haven't proven that ARR is different than CARR. And some of that's going to go away. There will be some, you know, a cemetery of defense companies out there. There was something that happened during the Middle Eastern wars when we were fighting in like austere sandy places and a lot of IEDs were blowing up our, our troops and our equipment. Uh, there were about 200 firms that came into the marketplace to develop bombproof or bomb-resistant type vehicles to save lives. And for a couple of years, BAE and Oshkosh just sat back and watched all these people come and go and then took the best in class, reading the green, so to speak. And then they came out with MRAP. And got all the contracts and still do. So I think you're seeing that in the drone space. We've tried to stay away from drones just because there are at least 2,000 drone companies out there. I'm not saying they're bad, it's just they're not all going to win. And to the point earlier, the more interest you have into a market, the harder it is. So we've stayed away from that. I think that Andrew effect is real. Therefore, you got to be extraordinarily careful about the valuations that you're coming in at these companies. And I do believe To close up my defense soliloquy, I do believe you're going to see a lot of assimilation, a lot of M&A go on to absorb some of this innovation versus letting them fall. All right.

Larry, sticking with you for one more second. I'm trying to keep tabs about what's going to happen in the broader Middle East regarding the Iran conflict. And as you can tell from this poly market chart, it's all over the place. People don't quite know what's going on. What are you telling your portfolios about disruptions to supply chains, how long the situation could last, and when things might kind of get back to quote-unquote normal? I'm curious what your timeline is and what you're telling your portfolio friends.

My personal opinion is summer. I would love to see a bow put on this. It is extraordinarily dynamic. There are dynamic leaders involved on both sides and kind of hard-to-predict leaders on both sides. Both of the leadership have a bombastic way. I'm not taking an opinion either side. I think America should be a leader, and sometimes it's required that we take that more bombastic approach. At the same time, Iran is very unpredictable, and they're not Venezuela. This isn't a drop-in, get something done, come back home and celebrate. So I would like to see it happen sometime in the summer, get a bow on it and move on. The Strait is extraordinarily important. Supply chain is very important to us being able to accomplish what we need to accomplish. All of our companies are paying close attention to supply chain, not just from this most nascent activity, but the last several years have proven that supply chain is critical. And if you look at any theater of war and in human history, supply chain is kind of the crux.

Yeah, just see The Bronze Age Collapse for more about that. But Mike, I think you were gonna jump in.

No, I thought it was important to note, Larry's absolutely right, especially when he pointed out about the American troops that were blown up by IEDs that happened in Iraq and it was done by Iran. And that's an important thing to remember. But it's also important, I think, to understand that there's an upside here. You know, as Michael well knows, you know, Tel Aviv stock The Nasdaq exchange has kind of outperformed every other index over the last year and change. And part of the reason for that, I think, is because investors in that fund are understanding that there has been for decades this Iran overhang on the Israeli economy, which actually is on the global economy. When you think about a place like Iran with the talent and the resources that they have, if they were actually to become a democratic country, if they were actually to be a place where we could go look at innovation and portfolio companies, you know, then you have— I've, for years, I've spoken of Iran as the single biggest binary in the world because, you know, at least up till now, they have been the most disruptive force in the world, not just in our region. And, you know, the nuclear threat has been an apocalyptic one. And now, on the other hand, if the regime goes away, and I believe it will go away, whether it comes from kinetic action in this next period of time or whether it comes from the people. 'Cause remember last year when we had the war in June, it took 6 months till people came out in the streets and then were slaughtered. And that's when Trump said, "We're standing up and help is on the way." And so I do believe the people will win and they will turn on government. And then think about the resource that they have. Think about that oil coming onto the global market and what that does for oil prices, for the economy, for energy. And then the innovation that'll ultimately come from the talented Iranian people that have been suppressed for half a century.

Michael, you mentioned October 7th earlier, and I think there's been quite a lot of pain in the Middle East in the last couple of years. How is, how has the situation impacted the Tel Aviv technology scene and founders and building inside of Israel versus building maybe with some Israeli roots but having a headquarters somewhere else?

We've had an unbelievable capital inflow, not just to the public markets, as Mike said before, but also the venture ecosystem. You know, I'm just— the hardware, as I mentioned before, has turned up. You know, Accel Sequoia has opened an office since the war started. Just the inflow of capital is extraordinary. And what we've learned is that Israeli founders are just incredibly resilient and they're innovating in real time and they want to do something meaningful. You know, as this war has really affected everyone in the country. And part of that meaningfulness is defense and some of it is around AI and health and other areas. But you see these founders just like, aspiring to do something world-beating that, that, that, that really matters. Um, but I actually want to take that, with your permission, to the previous question and connect it. And so, um, there's a, there's a meta question here we need to, we need to answer, which is, has the global supply chain changed forever due to bipolar great power competition between the US and China, the war between Russia and Ukraine, and the war currently in the Middle East, which has also been exported to other places. Has it changed because NATO is on probably on the verge of collapse, right? Because President Trump has decided he's not going to support it anymore. And that'll change the entire Europe while we're at it. And by the way, you know, Europe's contending with its own problems, the rising energy costs and the loss of an ability to compete in so many areas, inflation and kind of overhung balance sheets. It's— the world is changing and the world is not going back to the way it was before Russia invaded Ukraine. Like, it's over. We need to acknowledge that. And so, you know, in our annual letter this year, I talked about sovereign countries, sovereign companies, and sovereign individuals. And we're going to a world, I believe, where you're going to have to have a sovereign allied supply chain. You will not be able necessarily to rely on adversaries for key components of supply chain. That will certainly be true in defense. It's obviously true in semiconductors, but it's taken a long time to get there, um, and will take an even longer time because these are, these are atoms. And so you need to kind of figure out how the globalist supply chain changes to an allied supply chain if you're going to want to compete. So that's point one. Point two is October 7th and the Iran War, as Mike was saying before, you know, the collaboration between the United States and Israel is literally unprecedented in the history of global warfare. It is unprecedented. The intelligence sharing, the use of AI, the use of modern technologies without getting into too much detail, the collaboration on the battlefield, the refueling to allow jets to go endless, endless thousands of kilometers. And The tight ability to connect intelligence services to operational efficiencies is stunning. They're going to write books about this for the next 100 years. It also says who are the allies of the United States, not just when it comes to warfare, but when it comes to supply chain. The UAE, for example, has turned up in spades and that is going to be— it is an incredible country. It's going to be even more incredible country in the race for AI because they can be trusted. And that's why they're building a 5-gigawatt data center there that US companies are allowed to go to and Nvidia chips can go to. So the world is going to rearrange in entirely new alliances as NATO fades, as globalist supply chains kind of realign to sovereign allied supply chains, as Europe gets weaker, significantly weaker, uh, over the coming decades, and parts of the Middle East and Southeast Asia rise— India, for example, and Indonesia. We're going to see the whole world reordering itself over the next decades. Companies need to get into that, uh, uh, jet stream in order to be successful for the coming decades.

So, so does that change the, the prominence of Silicon Valley and let's just say Shanghai as global hubs? Does it, does it create more or does it shift where the center of gravity is in the world of technology somewhere else, Michael?

You know, I think Larry's, uh, comments earlier about the magnesium company in the Grand and the Grand Rapids anecdote is critically important. So, Gen AI is won by Silicon Valley. There's no sense investing in Gen AI anywhere else on planet Earth. They've won. They're 91% of the dollars, 91% of the companies. It's over. Yeah. We incubated beginning like 3 years ago what we call a creationary chemistry company that does 3 things. It makes rocket fuel, makes energetics, and it makes chemicals that purify rare earth minerals. And they have solved manufacturing because you don't need to scale up manufacturing. You can kind of spin up modular manufacturing in Austin or Grand Rapids or Tel Aviv. It doesn't matter. That's the kind of company that has a big moat that can be created in Israel. And you can win, you know, through distribution of that and local manufacturing by doing that. And Larry's point is we're going to see a tremendous amount of innovation, I believe, that's not in kind of the— what do we call it— the 4 Cs, the consensus colossal capital areas, uh, that, that, that's been around Gen AI and a few of these other areas. Um, then they can spin up anywhere. By the way, look at defense tech, which is now in Southern California. People didn't see that coming. Austin, Texas become a center. Miami's kind of maybe on the rise. And, uh, Tel Aviv continues to outperform, uh, at a meaningful level. And Shanghai is Shanghai. You know, China's going to have going to have a center. The next question we need to ask ourselves, though, which I'll just finish with and shut up, is— No, you're good. You know, who's going to finance all this stuff? So NASDAQ and the New York Stock Exchange has raised the threshold so much you can't go public anymore in these places unless you're, you know, go public or go bust at a $10 billion valuation or $100 billion, whatever the number is, or, or you're Anthropic or SpaceX. Those IPOs may dry up a lot of liquidity for other companies very, very quickly. So the question becomes Who's going to be the financier of choice for the companies that Larry was talking about, Mike's talking about, and I'm talking about? Well, Hong Kong is the gateway for these Chinese companies, for Shanghai. That's where you finance those companies. Here's my radical comment: the Tel Aviv Stock Exchange is going to be the NASDAQ of the current decade, and you're going to take Larry's, uh, Grand Rapids, Michigan company, the magnesium company, and my creation of chemistry and cyber companies who are not going to reach you know, not going to hit the $10 billion threshold you need to go public on NASDAQ, the New York Stock Exchange, and the robust financial markets in Tel Aviv that Mike pointed out is the best performing stock index over the last 3 years. That's where you're going to take companies public in the areas that we're investing in over the next couple of decades.

So the NASDAQ's future equivalent of the Shanghai Stock Exchange Star Market, the smaller tech-focused index, will be, in your view, essentially the Tel Aviv stock market. That's correct. Larry, do you agree? Are you going to get Magrathea to list over in Tel Aviv? I haven't been yet. That's why I don't have an opinion.

You're all invited. That's the first.

Come on over. I don't know if it's that toxic anymore. I mean, 5 years ago, like you said, you wouldn't assume that the greatest or some of the greatest future defense primes would be coming out of out of California would say that's absurd. I think the markets will absorb these companies here in the United States blended with your Tel Aviv assumption or positioning, probably some of both. And we've met with several companies, by the way, out of there that are really attractive. We haven't made a bet there yet, but the level of engineering and commitment and tenacity is extraordinarily evident. So I wouldn't say that wouldn't happen in the near term.

I think that's right, Larry. And just to put a finer point also on what Michael said before, our kids' generation is being called Israel's greatest generation. And I firmly believe that they are going to create some of the most incredible startups that the world has seen. And I say this not only because I'm obviously biased towards my own kids, but because we sit, we sit around the table every Shabbat and have our neighbors and their kids and they talk about their experiences and what they've been through the last few years and how it has inspired them to go out and build. And as they talk about it often in very specific and particular ways and are out there getting ready to start companies. So keep an eye out for that, for our ecosystem to have another step up over the next few years.

Well, here's hoping that the CCC firms don't show up and torch everyone's early-stage valuations in Tel Aviv. I would like you guys to have fat returns. Um, we have, we have to wrap, but I want to do one last question for each of you, kind of a lightning round. Uh, what is your current favorite portfolio company, or if you're too cowardly to say, which one has recently performed well that you would like to shout out?

You know, when you have 8 children, you learn not to answer that question. I'll tell you something to pay attention to. Uh, there's a company in our portfolio called Jiga. Which has done an extraordinary amount of work in AI for procurement, particularly for discrete manufacturing. How, if you're an engineer or an engineering team at a manufacturing company, it doesn't matter if it's your defense company, Larry, or you need to produce magnesium or whatever it is, I need to get a discrete part. How do I get that ordered and designed up, et cetera? And this company is explosive, uh, both from a growth perspective but also for how much faster they can turn, you know, the turns on manufacturing. And that matters in the market, and it also matters for what I described earlier, which is a supply chain. They're able to show you how to create a supply chain that's no longer dependent on China.

They do custom CNC machining. Ah, I grew up having non-CNC machining tools, so I've always been like jealous of people that actually had the good ones. Um, but if you want to take a look at that one, it's jiga.io. Thank you, Michael. Larry, favorite child?

Um, you know, there's a company called Divergent that's in our portfolio that does something similar to what you just described. Um, really, really interesting company. They build hypercars under the name Zinger, about $2 million apiece. But, uh, what really got us interested was the fact that they could print, uh, essentially print maritime assets, terrestrial assets on demand in austere places when you need them to. So they're actually a pretty meaningful supplier right now to, let's say, F1 and other automobile manufacturers. But the really powerful thing that attracted us to the company was the fact that they're getting Navy and Army contracts out there doing really, really important work behind the scenes, and they're not out there talking about it. Really, really cool. Our greatest return would be our oldest investment, which was our first one, the WebAI that we engaged at $70 million. We co-led the round at $700 million. Then they did a $2.5B and there's a multiple of that coming out this summer, a new round for them. So we're really proud of them. That's the edge AI company. We also have a company of omniscience called Umbra where they do a tremendous amount of the United States reconnaissance from from space using synthetic aperture radar. We've got a wonderful portfolio. Another one that's crazy is called Volleback, and they do— they've for about 6 years been building things out of really crazy materials. So they're a material science company creating consumer clothing. And I met with the CEO. I said, why aren't you doing things for the DOD or the Ministry of Defense? They're out of London. And now they are. They've hired the head of human performance from NASA. People from the agency, and they're doing really extraordinary things behind the scenes. Not very showy, but really important. Another one, Firehawk, that 3D prints propulsion, 3D prints rocket fuel. It's just— that's my favorite. It's really good from your portfolio.

Larry's got 3 more companies, so it hits my number of kids. He's got—

come on. I was gonna say, he couldn't even get to conviction on picking his favorite. I mean, how is he going to lead a deal if he can't— no, no, he's cut off. Larry, you've made all of us look wishy-washy. Mike. You get the last word here.

Yeah, well, with all the disclaimers that have already been stated, I'd say the company of ours getting the most attention right now is out of LA called Harbinger, harbinger.industries. And, you know, they figured out that, you know, I've been, you know, in the electrification space and especially around transportation for a couple of decades now and seen a lot of ups and downs. But the thing that, that Harbinger figured out was that there are segments of the trucking space, particularly the middle-duty trucking space, where you can today produce electric trucks cheaper than you can produce traditional trucks. So you go out and start the conversation from, from a discount upfront cost, and then you get to appreciate the operating reduction in cost as well. But they've figured something else out also. By the fact that they've brought in an extraordinary team and their manufacturing is led by the same guy actually who fixed the production hell of the Tesla Model 3. And as a result of the team they put together and the innovations they've made around putting these chassis together for mid-duty trucks, they've understood that there's other things that you can do in industrial manufacturing. With today's tools and with AI. And so they're beginning to look at other areas in there as well. They've raised a lot of money and we're in very close touch with those guys in spite of the 10 time zone differences. I was out there last week and so keep a deep eye on them.

If you want to learn more about Harbinger, check out TWIST episode 2154 from July of last year when I spoke to the company. Oh, fantastic. They're making these essentially like EV skateboard chassis for trucks and you can just drop whatever you want on top of them. It's kind of the old coach-built car model, but for the EV world and it's a super cool company. I love what they're doing and anything that gets more electrification is fine by me. Mike, appreciate it. All right, guys, we got to wrap. Thank you all so much. A real treat. I learned a lot. I hope everyone enjoyed this. I guess the gist is AI matters. Compute's expensive. Israel is going to be fine. Technology is good. We need more IPOs and it's all Biden's fault. All right. This has been TWIST. Thanks, everybody.