USV新设9亿美元基金：Fred Wilson、Rebecca Kaden与Mike Mignano

1. USV's biggest fund ever: $900M

We raised $900 million in new funds for USV, which is by far the most USV has ever raised in a new fund cycle.

I don’t think a trillion is what we need to underwrite at USV. I think if our biggest winners are in the single-digit billions, given our fund size and our ownership, we can produce really, really healthy returns for our partners. There have been some outliers at $50 billion or $100 billion, but a lot of our biggest winners are in the $10 billion range, and that’s produced enormous returns for us.

It’s also a bigger number than we’ve raised before, somewhat significantly, and marks a new strategy for us. The market has evolved, the outcomes are bigger, and the swings are bigger.

I was a little surprised how many of them said, “Are you serious?” We are here at Union Square Ventures, USV, in New York City with Mike Mignano, and we’re about to go on a bit of an office tour. After that, there’s more. We’re going to sit down with Fred Wilson, Rebecca Kaden, and Mike to go over what we’re about to announce.

We just raised a new fund. We raised $900 million in new funds for USV, which is, I would say, by far the most USV has ever raised in a new fund cycle. But there’s a reason for that, and I think we’ll get into that in a little bit.

So why not $1 billion?

We’ve always been extremely intentional about fund strategy—not just fund size, but fund composition: how many companies, roughly; what size checks we plan to write; stages; and the number of companies. The $900 million really comes from that strategy and that model. We don’t raise funds as vanity metrics. It’s not like we’ve got to hit $1 billion or we’ve got to hit $2 billion.

“I want to be a unicorn fund.”

The $900 million really comes from the strategy, bottoms up, which I think we’ll probably get into.

Yeah. Okay. So maybe we can do what you do best and walk backwards to hide all the equipment.

Okay.

And then we can pick it up around the corner.

2. Inside USV's New York office

Sure, let’s do it.

All right. So we are here at Union Square. We’re going to walk into the kitchen. These are some of the main areas. It’s very active today.

Lots of people here today. This is a bit of a seating and living room area. Obviously, we’ve got the kitchen. My favorite feature of this kitchen is this xBloom pour-over machine. Have you seen this thing?

No. What is this?

It’s basically a pour-over robot. If you like pour-over coffee, this is a robot that makes—

An investment?

I don’t know. No, it’s not. But it’s really cool. I’ve seen these in a few offices—a couple of startup offices and a couple of VC offices—so it’s not like USV is the only place that has this. But if you like coffee, it’s pretty good.

It’s cool. I’ve never seen that.

Yeah. We’ve got our little dining room table here. We’ve got a board from one of our portfolio companies. Do you know what it is?

No. What is this?

Gameboard is the world’s first tabletop gaming console.

Founded by Brynn Putnam, who founded Mirror.

This is an amazing product. Every game comes with its own set of unique pieces that the screen reacts to. They also have a platform to make it really, really easy for developers to build games for this. It’s an awesome product. I would definitely recommend it.

Do you play chess on this?

They don’t have chess.

They don’t? They have all their own games that developers are writing. I guess you could do chess if you shaved off the edges or something, but it’s awesome. Highly recommend.

This is what we call the pit.

The pit.

The pit is where we all work and hang out. This is my desk. It’s very cluttered, so I hope you don’t get a shot of it.

There’s nothing bad on the screen.

We talk about deals. We talk about whatever might be going on on X that day. There’s also a group chat for the pit, so the conversation that happens here very actively every day just continues right when we leave here.

So no one has an office, or do they?

No, no one has an office except—well, actually, Tyranny, Carrie, and Christine have offices.

This is special ops.

They don’t want to deal with the madness of the pit, which I don’t blame them. This is where a lot of board meetings happen, as well as some of our full partnership meetings. We also have another room in here that we use for these types of meetings. One of my favorite things about this office is that pretty much every room has really epic views.

Yeah.

You’ve got great views all the way to the Hudson River over here, and views uptown over here. There’s a really, really cool view from here where you can see straight up Park Avenue, which I really love. Actually, at my desk in the pit, if I turn to my left, it’s literally a perfect vertical view of Park Avenue, which I think is really cool.

I mean, this represents New York pretty well, if you can see all of it, if you’re coming as a founder from—I don’t know.

Totally. I’ll take you up to the roof in a minute, where we have literally 360-degree panoramic views of pretty much all of Manhattan, which is really cool.

So you guys have invested in 263 companies? That number sounds right.

Yes.

And how many do you expect to invest in with this fund?

3. Why great founders still fail

I think we try not to be too strict and specific about the number, but generally speaking, probably somewhere around 30-ish, maybe 30 to 40. It’ll be a mixture of both seeds and Series A’s. I think we’ll do probably a lot more seeds than we’ve typically done in the past. We can get into a little bit of the fund strategy later with Rebecca and Fred, but going earlier is becoming more and more important, given how many startups are emerging as a result of AI and how competitive the market is.

Yeah.

This is the library. This room is great. It’s beautiful, and we use it for a variety of reasons. We host events here with our founders and different people within the community. As you can see, we also have these desks where we often host portfolio companies or companies that aren’t in the portfolio that need a place to work. It also is a library, with lots of books on the shelves that many of us have read.

Do you have a favorite?

Let’s see. One of my favorites is The Messy Middle. Shout-out to USV product advisory partner Scott Belsky. This is his book, and it’s about the middle of the startup journey. There are many books written about the early days and the formation. There are also lots of books written about the success stories and what happens at the end, but I believe Scott’s vision with The Messy Middle was really about what happens in the middle, which is the hardest part, as most startups are not straight lines. That’s a great book. Let’s see what else. Shoe Dog is obviously a classic. There are a lot of good books on this bookshelf.

4. An AI agent that runs your portfolio

Speaking to your point about going earlier with this fund, you guys are also incubating companies. You just released one that you incubated yourselves, SuperTakes.

Yes. One of the things we’ve been talking about is that it’s getting easier and easier to build products, get them to market quickly, and test them quickly. One of the things we’ve been talking about as a partnership for a long time now is the idea that, as a result of AI, it’s never been easier for people to take control of their own financial independence.

We’re coming out of a world in which huge companies and a huge market have been built on paying people for financial expertise and financial services. With SuperTakes, we believe that regular consumers should be able to invest in their ideas without having to know anything about the stock market, picking stocks, building a portfolio, rebalancing, or actively managing a portfolio.

With SuperTakes, it literally does all that for you through an agent. You just give it an idea or belief you have about the world, and SuperTakes will go and invest in that idea directly through your Robinhood or Coinbase accounts or other brokerages. Then it will actively manage it for you. It will tell you how it’s doing, check the news for you regularly, and, if something’s broken in the portfolio, automatically rebalance it for you.

It really speaks to the types of companies that we look for and to the theme of “obliterate, don’t automate,” which we could talk about when we discuss our thesis. We think AI companies are not only making it easier and faster to achieve existing processes in the market, but are literally restructuring markets completely, because humans now have access to intelligence and expertise that previously they had to pay a lot of money for. SuperTakes is a great example of that and something that we were able to build ourselves, at least for the initial product.

That’s awesome. Did you build it in this room?

We did not build it in this room. We built it in lots of other rooms, though.

If it’s becoming so much easier to start companies, and you can do it with just one person, why do we need venture capital?

It’s still expensive for a startup to operate and scale a business. Obviously, things like inference are becoming more and more of a line item for startups. Great products are getting built, but to subsidize those products and make sure that customers can actually use them, maybe without having to pay themselves at first...

More and more startups are subsidizing inference on behalf of customers. The other thing I would say is, as it is becoming easier and easier to start a company, there are so many more companies, and therefore it’s probably harder than ever to actually stand out and break out from the pack. We’re seeing startups getting more and more aggressive about their go-to-market and marketing strategies and making sure that they can really break through the noise. I think startups are always going to be expensive to maintain, operate, build, and break out. It’s just building the actual product, I would say, that is arguably easier than ever before.

So, we just walked through the entire floor. This is the main floor, right?

Yeah. Yeah, this is the main floor.

What are your favorite parts about this office? You’ve worked at different offices not too far away. So, what’s your favorite vibe about this space?

I’ll tell you, one of the things I really love about the USV office is that there are lots of different vibes in the office. You come in, and it’s very comfortable. It’s very homey, with beautiful furniture, chairs, and couches. As you get further in, it’s more of a coworking space, with people working and talking. As you go in the back, you have a little bit more of an introspective place in the library, where you can get some quiet. Then, as we’ll see in a bit, the roof offers a whole different element. Whereas I feel like most New York City offices, especially small offices, just feel like you’re in a room, right?

There really are actual different personalities to this office, and we have light.

Yeah. As I said earlier, pretty much every corner of this office gives you a different view of New York City, which I think is really cool because it’s my favorite.

Do you have a good window cleaner? What’s the secret?

It’s a good question. I’m not sure who cleans the windows, but they clearly do a great job.

We’ve made it to the rooftop. What goes on up here?

This is just a really great place to work when the weather’s nice. We’ll come up here, and I’ll take some calls. People do lunch up here. We also occasionally have different get-togethers and meetups with founders. There’s a party that we often have here every fall that people really like.

Much like downstairs, as I mentioned, the nice thing about being upstairs on the roof is that you have an amazing view from pretty much everywhere. Right through this archway here, you can see a perfect view straight up Park Avenue, which I think is incredible. You can see the Empire State Building. If you walk over this way, you can see all the way across the Hudson River to New Jersey. On really nice days, the sunsets on this side of the roof are really spectacular. So, it’s a great roof, it’s a great feature of the office, and it’s something that I think we all really appreciate.

5. Agents won't watch Netflix for you

We’re experiencing some good fall weather today. Okay, what is your hottest take right now?

My hottest take, I would say, is that the user interface is not going away. I think one of the things we’re seeing all over X is that people love to say that the user interface is dead, and everything is being collapsed and folded into a chatbot. I don’t believe that to be true. I think the user interface is where people build trust with a product, where they gain comfort and know how to navigate different experiences. I think the user interface is how products convey different concepts and ideas visually.

I have a hard time imagining that we’re just going to be delegating all of our computing away to agents. Yes, I think agents will do many things for us, especially the things we don’t want to do. They might go shopping for us. They might file our taxes for us. But they’re not going to watch Netflix for us. They’re not going to watch YouTube for us. They might not even scroll X or Instagram for us, right? I think we’re going to want to do those things. So, I have a hard time imagining that the user interface is just going to turn into a chatbot everywhere.

Okay, pretty good hot take.

Thanks.

Thank you. Thanks for having us.

Thank you. Thanks for having us, Molly.

6. USV's new strategy for the AI era

So, we have Fred, Rebecca, and Mike. How are you guys doing?

We’re great. Yeah.

Feeling really good.

All right. So, we have a huge announcement we need to discuss. Who wants to take it away?

I think you should take it.

All right. Well, we’re excited. We closed our most recent set of funds and raised $900 million. In many ways, it’s a continuation of what USV has now been doing for many decades: a thesis-driven, great partnership with founders, backing businesses that we think are going to be leaders of their markets over the next decade.

But it’s also a bigger number than we’ve raised before, somewhat significantly, and marks a new strategy for us. The market has evolved. We think the outcomes are bigger, the swings are bigger, and the opportunity to leverage the new technology in the market and the applications that will be built on top is super significant. The rounds are bigger, more expensive, and faster, and we are evolving our strategy to keep up with that and really be able to attack that market.

Amazing. So, maybe we go back to the 2004 LP letter. What were you talking about then, and how is it different today?

Well, that was early on in the internet era. Commercial internet had arrived in ’95, and by 2003, a lot of the underlying infrastructure had been built. It’s not very different from where we are with AI. A lot of the AI infrastructure has been built, but we’re still waiting to see what the big applications will be that get built on top of it.

That’s where we were in 2003. The infrastructure was there. We had Google, Amazon, Yahoo, and eBay, but we didn’t yet have Facebook, Twitter, YouTube, or the iPhone. We imagined that there were going to be all these new applications built on top of what we now think of as the web. I think the moment we’re in right now reminds me a lot of that, except now it’s AI.

Couldn’t agree more. I think we feel, like Fred said, that the AI infrastructure has been built. Obviously, it’s still getting built, and there’s going to be more and more opportunity, but we’re at this moment where we’re starting to see the beginnings of what feels like an explosion of the application layer of AI.

Obviously, there are products that we’ve all now been using for the past couple of years, like Granola and Suno and things like that. But just over these past weeks, everyone’s been talking about a whole new type of application layer with Instinct and Muse and Grockbot. It feels like we’re entering a new phase where agents will be powering a lot of new experiences, both for enterprises and consumers. That’s really exciting for us.

7. The real cost of building in AI

So, I’m curious: how do you justify the round-size increases if models are—if it’s preferred now to go to open source and it’s cheaper? Where does all that capital go?

I think something we talk about often is how it is getting dramatically easier to build a product. We recently announced that we incubated a product that we built very quickly and very cheaply, and lots of other builders are doing the same thing. But what that means is that there are so many more startups that you have to stand out from and break through the noise from.

One area that a lot of the capital is going toward is doing just that: breaking out, marketing, messaging, and positioning. I think we’re seeing a lot more investment, especially in consumer, for startups trying to break through the noise.

We’re excited about open-source models, and we see the trend moving in that direction. There’s a lot of promise there. If you look at the fundamental cost that companies are spending on compute, it’s in a totally different hemisphere than it’s ever been. Even if you take something like Instinct, the numbers they’re talking about on what they’re going to have to spend on compute to service their needs are pretty wild. So, that’s a really different cost space.

There are a bunch of big questions out there. How much does that go down? I think we’re in this time, and we feel like we’ve been in this time, where we don’t know the asymptotes of things. We don’t know when models stall. We don’t know how low compute goes. One of the challenging and maybe exciting things is that you’re playing a little bit against an unknown field, but there’s a fundamental cost of compute that is part of the conversation and kind of a cost block for companies that really wasn’t there before.

8. Who protects you from your own AI agent?

The other thing going on is that the types of companies we are investing in and that are scaling in the market are different and more varied than probably at any other time in USV history. We have software applications that are emerging that we believe are going to leverage this new technology to totally transform how we live, how we work, how we are productive, and all these kinds of things.

But we’ve also had a robust energy portfolio, and we believe that programmable, cheaper, faster energy is the horizontal undercurrent of everything being built on top.

And the cost structures of those energy companies are totally different, as are our investments in physical AI and the physical world, right? We're investing in things that are really full-stack robotics companies that are building their own models, as well as the deployment on top, and sensor businesses. And so they have different cost structures that are requiring both different levels of capital and sometimes different types of capital than previous eras have justified.

One thing on inference: It actually reminds me of an earlier era of technology, maybe the mobile era or even the post-internet era, where you had these great consumer products and businesses subsidized usage to make sure these products could get adopted as far and as wide as possible before customers were forced to pay. I think right now we're coming out of a period where AI applications in the enterprise and consumer have been more paid for by customers. We're coming out of a lot of subscription businesses, but now products like agents—to your point about inference—we're seeing that startups actually want to subsidize the inference and pay for it on behalf of customers to reach that critical mass of adoption.

9. Betting on markets that don't exist yet

I do think it's actually going to be pretty expensive to build large-scale consumer businesses if you don't want your customers subscribing. OpenAI and Anthropic are still subsidizing core chat offerings, right? I mean, Claude will ask you to start paying at some point, but ChatGPT has been free for how long? Gemini is still free, right? A startup can't come in and charge for something that their competitor is giving away for free. And so I think you're right that inference, or compute, is going to be an expensive subsidy for fast-growing apps for a while.

10. The fight over monetizing Twitter

Yeah. I have a question. In the early days of Twitter or Facebook, they offered something for free to the market. And I think one of the things that USV believed in those days is that giving away a lot of value for free in order to back into an opportunity to create a network was a very good strategy. Did it feel like they needed more money than other companies had ever needed because they were doing that?

They did need a lot of money. I mean, Twitter's original financing was $5 million, and then a year later it was $25 million, and then a year later it was $100 million. At that time, those were big numbers, relatively speaking, right? We could multiply by 10 and you'd get to $50 million, $250 million, and $1 billion.

That's kind of what we're seeing now, right? And that was also 2007, so that's 20 years ago. Just pure inflation is going to give you a multiple too. I think what's also interesting about that era compared to this one is that a lot of these products, as people have pointed out, are not that retentive, because there aren't things like network effects. But back then, with Twitter and Facebook, I have to imagine that you all were seeing these and saying, "We need to scale these things as quickly as possible, because once we do, we're going to have this incredible network effect, and there's going to be this incredible lock-in." It feels like it—

What was different then, which I don't think is true anymore, was that not every VC saw that—

The network effect.

Yeah. In the early days of Facebook and Twitter, there were venture capitalists who were arguing, "We need to monetize this. We need to have a subscription so we can have a revenue line that will allow us to finance the company." And there were others—we were in the other camp—who were arguing that, no, if we can get 10 million, 50 million, 100 million users, we will be able to finance the company. We ended up winning that argument, and we were right about that. And that's really how that model emerged that companies used, even to this day, to get to network effects.

I wonder if one analogous thing is that there's the belief right now that if you look at these emerging consumer agents, the answer is scale, right? The functionality is not all that different among them, but if you hit breaking points of scale—or maybe distribution—you can win markets, and so it's a race to see if you can have the right compute and the right distribution to get there. There's another strategy, which is one that we've been talking a lot about, which is maybe more analogous to the Twitter era: You actually may still need a network effect in order to maintain longevity and have retentive behavior. And that network effect is going to come with who can really be the first and best at multiplayer behavior, taking these new single-player behaviors and figuring out how to make them truly multiplayer. It's not that we don't need network effects anymore; it's just that we need to get there, and we haven't seen a new way. We haven't really seen them yet.

11. Why Amazon is shutting out agents

I want to push back on the agent economy so much because it's kind of obvious every single company is going to have an agent of some sort. You'll probably have more secure sandboxing within those. I mean, we were joking about this earlier, but it's kind of true: You input all of your sensitive data, and it kind of roams free. But it's clear with Facebook and Muse. I mean, Deel even launched theirs for most of its management work and that kind of thing. Every company will have an agent of some sort. Obviously, it's one of the most cost-intensive things because it's running so many different programs at once, and nodes and that kind of thing, but I just have to push back on that. So, do you think that's a viable long-term category? How does that compare with years ago?

We do. We do. We think that Amazon will have an agent. So I'll have my Amazon agent, and I'll have my Shopify agent, and I'll have my Etsy agent. But I think I'll also want to have my shopping agent, or maybe just my agent that will talk to my Amazon agent, my Etsy agent, and my Shopify agent, so that I don't actually have to be managing conversations with each and every one of those agents. I can just have my agent go off and do a task.

Today, what's happening is the agents, as you point out, are logging into these services and kind of pretending to be me. Amazon doesn't like that. I think last week they said to Manus, "We're not going to let Manus access Amazon," right? We'll see more of that. But I think eventually what will happen is there will be some standard or some way of agents connecting to the agents that each service offers you and making it easier. So I think we're in the early innings of how this plays out, but we do believe that people will want their own agents that represent them, as opposed to Amazon only having my Amazon agent.

This episode is brought to you by Brex, my favorite. You become what you spend on. And I refuse to spend my time on work that shouldn't exist. Expense reports, receipt chasing, and manual closes. The companies building what's next from Versel, OpenAI, Anthropic, Granola, and Deepgram Graham all made the same call. They all run on Brex. Rex is the intelligent finance platform that combines cards, expenses, and banking into a single stack with a Gentic Finance built in AI agents that handle expenses automatically, enforce policy before spend happens, and close your books in minutes. That's why Sorcery runs on Brex so I can spend time on building and not busy work. It's time to get Brex af. Learn more at brex.com/sourcy. That's b re x.com/s o r c e r y. Bye.

Turing is training the next generation of AI with tasks that require real expertise and realworld judgment. That's why companies like Nvidia, Anthropic, Salesforce, and Gemini partner with Turing. Turing builds realistic reinforcement learning environments and data systems based on real operational traces. The kind of infrastructure frontier labs need to train super intelligence. Visit turring.com/souc. AI needs more than chips. It needs power, land, and infrastructure. Zone develops next generation data center campuses partnering with AI companies, site developers, and technology leaders to bring compute online faster and at scale. Zone is building the foundation of the AI frontier. Visit zonfrontier.com to learn more. That's zonfrontier.com to learn more.

Before we go too far, let's talk about USV's strategic positioning: where you guys are investing and what your typical check size is. And what was the pitch to LPs? Was it an easy one?

Yeah. So USV has always, for its history, really been focused on the early stage, mainly Series A, sometimes some seed, sometimes a little bit more than that. I think we're going to continue to play the same game, invest in the same stages, for the most part. But to Rebecca's point from earlier, we're going to be doing it with a slightly larger capital base—not so that we can do many more deals or fewer deals, but more because, as we talked about earlier, the shape of the market and the shape of the deals has changed.

As it's gotten easier to build, there's also much, much more demand flowing in in the form of capital, and startups are raising more. So for a fund like ours to stick to our strategy, to stick to our various ownership targets, and make the fund math work, we need to write slightly larger checks. It doesn't necessarily change the way we think about deals or the way we think about companies or our thesis-driven investing. It's more of a reflection of where the market is today.

I think USV has a long history with a group of LPs that have been very, very loyal to USV, and I think USV has been loyal and good to them as well. So there's a lot of trust established there.

We added some new ones. What was interesting to me was that LPs see the same thing we're seeing. They see it too.

We're not living in different worlds. They see it too. Maybe a few years ago, this would have been an edgier push than it was right now because the market has pretty clearly moved in a direction.

You cannot like things about the market, but you're unlikely to change the things about the market. We believe that it's a pretty unique time to be on the field investing, and we want to be on the field in the right way.

We have 4 areas that we talk a lot about in conversations like this, on our website, with our LPs, and with founders: physical AI, AI applications, AI infrastructure, and energy. Those are the areas that we're investing in, and we're consistent about that everywhere. Our LPs understand that.

There are other areas out there. Founders who are working in those areas have lots of places to go to, but they don't come to us because we're not loud about those things. The things that we are loud about, founders find their way to us.

I think that's what we mean when we say the strategy is kind of the same as it's always been at USV. We're thesis-driven. There are things we like, and there are things we don't like.

$900 million may represent a big jump up for us, but relatively speaking, it's still maybe considered small relative to a bunch of other funds. I think that size and this thesis-driven approach are what enable us to play the game that we want to play relative to the rest of the market.

12. Why smaller funds pick better

Well, to that point, I was an LP at one point in my career, and there are very few funds to invest in. You typically know which ones you're going to invest in a year before you invest. It's all dated and very organized.

Were they waiting for you to raise larger? Did you have pressure for that over the years? One thing that I think USV has always done quite well—and I think Fred and Brad said this—is that we're very communicative with our LPs. We talk to them a lot. We don't take their support for granted.

We do quarterly calls. We update them on our strategy and our portfolio. We tell them what's working, what's not working, what we're thinking, and where our heads are. By the time we come and say, “Okay, we're going to shift the fund size from X to Y,” they've known it's coming. This is part of a conversation that's been going on for a long time.

I was a little surprised by how many of them said, “Are you sure you shouldn't raise more?”

Really?

Yeah. Again, we're not living in a bubble, right? They're seeing the same things in the world that we are, and they're seeing the landscape of what's advantaged.

Why have a small fund? I think there are a bunch of reasons. One, you can return it more easily, get into carry, and have the dynamics of the fund work the way you want them to. Two, you can be pickier, right?

We always want to manage a fund in a bottom-up way. When I say that, what I mean is there's a universe of companies. We don't want to say we need to be in X% of them for this math to work. We want to say we have strong hypotheses and ideas about what we want to do, and if we're right enough about the ones we do that are aligned with us, this fund is going to be a big success.

We don't want to get bigger than that strategy allows for. But that number has shifted, and so that was an ongoing conversation with the LPs.

It also aligns interests, right? When the funds are small, it focuses the results on the performance of the portfolio companies, and everyone rallies around that. When the funds are really big, it potentially distorts the incentives a bit. That's one thing that I've really appreciated about USV's small-fund strategy.

When you started USV, did you want it to stay a small fund, or did you ever want it to be a really big fund?

I like the earlier stage of investing. When we started, we were thinking of writing a $3 million Series A check. Obviously, that number has changed a lot over the years, but that's where we wanted to be, and that's where we still want to be.

13. What Series A looks like now

What is a typical Series A today?

I'd say it starts with 10.

I think 10 is now on the small side.

That's why I say it starts with 10.

Yes, 10. You're obviously seeing some Series A rounds that are $100 million or—

A billion.

A billion, right? We also know what's not in our zone. That's not in our zone.

Overall, we want to lead or co-lead investments. We want to make real bets in things and stand behind them. There might be a time when we break that and write a strategic check into something in a different way, but overall, that's really our strategy.

The Series A checks that we believe in this fund will probably be 10 to 30. They don't have to be the whole round, so the rounds could be bigger than that, but that's probably what we're looking at. Then we'll do a lot of seeds, and we'll even do some earlier than seeds. Maybe Michael can talk a little bit about that, too.

One of the things that we've been thinking a lot about at the seed stage is what new and different strategies we can pursue as a partnership that USV hasn't traditionally pursued to get real ownership at the seed stage.

One thing could be taking more pure founder bets. As a thesis-driven firm, we haven't done a ton of that in the past. Another thing could be that, earlier, we talked about how much easier it is to build something than ever before—how much cheaper and how much faster.

We have a lot of strong opinions about different categories and theses, and in some cases, we feel like it might make sense to actually build something inside the partnership as a bit of an incubation.

There's another slant on that. We have a couple venture partners, Jared Hec and Scott Bellski, who are really accomplished entrepreneurs, product thinkers, and builders. Working with them and Michael, who is also very much in that camp, we can do things that aren't really an incubation.

We have an idea, we go find somebody who maybe wants to start something, and we partner with them to get it off the ground. We're not the founder per se, but it's a situation where we have full ball control, if you will, and we'll provide maybe the first 2 checks into it before it goes out to market.

This is all a continuum between real incubation, founder bets, and what I would call proprietary seeds. I think those will make up as much as a third to a half of all the seed investments we make.

As Michael said, it's getting easier and easier to prototype something, take it to a small group of users, get real feedback, and see if you've got something or not.

That's a good point. So how has the way in which you diligence these companies, and the milestones and success metrics, changed over time? What do you look for now, or what do you overlook now, that you might have looked at before?

One of the things that I've been really excited by, coming in as USV's newest partner, is how much this team cares about products. It feels like the deals that we've dug in on sort of always require—not a hard requirement, but it feels like there needs to be somebody on the team who's really excited about the product.

Obviously, that can't always work, especially in today's market where, as we just mentioned, in some cases we want to go earlier and earlier and just bet on a founder—maybe somebody that's in our network that we really trust—but—

Or we're betting on micro nuclear reactors, you know.

Right, exactly. So it can't always be—

That's a little different. In the energy side of our business, it's different. But I would say in anything that touches the consumer, it's a combination of what Michael said, which is using the product and getting an instinctive feel for whether the product is good or not.

When you combine that with retention data and cohort data, the magic is when the thing you feel is the thing you see in the data. Either the retention that you feel and see, or maybe you're like, “I might churn off of this because of X, Y, Z,” and you see that in the data. Then you're like, “Oh, they're going to have a churn problem.”

It's a combination of using the product and a product-native mindset, combined with a lot of data and really looking at cohorts and understanding what real users are doing.

I think the other thing, though, is that team is, and always has been, the most important thing. No great company can be built or started without an A-plus-plus team.

We actually debate that sometimes because there are examples of crazy runaway network-effect businesses where the momentum of the network can carry it far enough that maybe team matters less. But I think in today's world, team always plays an important role. I would argue it's playing an even more important role than ever, for a bunch of reasons.

One is that I think great hiring is harder than it's ever been.

There's massive, massive talent competition, and so the ability to pull people away who are excellent from really, really strong other opportunities—sometimes where they're getting paid a lot of money by large labs or incumbents or whatever it is, or startups that have raised billions of dollars—you need the right founders who can really mobilize a world around that.

The other reason is that I think narratives and market narratives are aggregating faster than they ever have. The market is designating winners, right, and telling these stories in a way where value aggregates, often in a way that way, way, way precedes where the business is. So the ability to be the founder who drives that story, gathers that attention, breaks through the noise, and designates themselves as the owner of a market very early is super important, both for investors and raising money, for teams, for customers. That narrative piece and storytelling, I think, we are paying way more attention to now than we probably have in other markets.

14. Is ARR a fake metric?

Also, if it's easier than ever to build a product, we want to back founders that we believe can really go the distance, right? And that's, I think, more rare than ever.

I know this is—we're talking early stage—but as the companies mature, you guys are sophisticated investors. You are actually looking at performance metrics. It's clear which ones end up becoming scams because they're really good at marketing themselves, and which ones actually have performance, revenue, and that kind of traction.

So I'm trying to get below the layer of, "Yes, it's a great product; yes, it's a great team; yes, it's a great founder." What do you want that team to be focused on? Is it retention? Is it revenue? What do you care about?

So, last time I was on Sourcery, there's a clip now that I see on X all the time from that interview where I guess I said ARR is a fake metric. I said something like that. I see this.

I think ARR has become a metric that a lot of investors pay attention to, given how fast the revenue rate has ramped for so many of these companies. But I think one thing that we care a lot about is clicking a layer deeper than that.

So many of these products now are subscription-based. We talked a little bit earlier about how so many products are subscription-based. I think it's easier than ever before to get somebody to pay for a product via subscription without knowing if they actually like that product. A lot of the subscription plans are annual. A lot of the time, you have to pay just to try the thing.

And so we try to look often more at the true usage retention of the product. How often are people actually coming back to this product and using it? Are they coming back every day? Are they coming back every week? Do they churn out after a month? I think that's probably a more telling thing to look at than just revenue retention.

Yes, great, if the company can hold on to that revenue for a year or 2 years or 3 years. Obviously, that's going to show up in the business. But if the product isn't good, at some point they're going to churn and the business is going to fall apart. So I think people should be paying more attention to that, like we used to, I'd say, in an earlier era of investing.

I think it's user retention and product innovation. I think we're in a moment in the market, on the consumer application side, where trial behavior is at an absolute high. Totally.

And engagement behavior is at an absolute low. And that makes sense, right? We're getting thrown so many new things all the time in the market, and you have to pay to try them.

You have to pay to try them, and we want to. It's on the news. It's the main discourse in the world. Everyone's like, "Are you up to speed? Have you tried Instinct [?] Have you tried Muse? Have you tried Robinhood?" We want to try them all. We're going to stick with very few.

So the trial behavior is a little bit of a fake-out, right? Because we know you can get people to try, but it's only really going to have long-term value if you can get it to integrate into someone's life.

Totally.

And so figuring out how to tell that is, I think, the whole thing. Part of that is watching how someone iterates on product, right? I don't only think it's great product insight. It's what are the little hooks and things they do, and changes they make, that keep that product going and fresh and different?

Because realistically, the idea that we get full insight into retentive product behavior before we get to make a decision is not really true, right? If we really needed that, we'd be having to write way, way bigger checks than we would like to write here. So we're going to have to make an inference about that with less data than maybe we would like, and so I think you're looking for a proxy for that, both on the team and on the data side, that gives you the sense that that's to come.

But a lot of this that we've just been talking about, I think of as a seed, Series A, Series B kind of conversation. One of the hallmarks of USV is that we are very long-term investors. We're still, all of us, working on companies that we funded over a decade ago.

And so, as these companies grow, what you want to see is that the team can launch new products and new lines of business. We had all of our CEOs together last week at our annual CEO summit. One of the stars of our 2016 fund is this company called Abridge, which is an AI-for-healthcare company. Shiv was telling a bunch of us, just in a casual conversation, that they had made all of their money for the last 5 or 6 years on a single product, and he's most proud of the fact that in the last 12 months they've launched 2 new products—really, 2 new lines of business.

So now it's a 3-legged stool instead of a 1-legged stool. It's going to be a much more stable business. And the 3 products are very synergistic. When a customer adopts 1, they're kind of locked in. When they adopt the second one, they're more locked in. When they adopt the third, they're kind of all in on Abridge.

So that, to me, is the evolution of a company that's going from being a great company to being potentially a fantastic company. And the thing about venture capital, as anybody who studies it—you've been an LP, so you know—is that usually 2 or 3, maybe 4 companies per fund end up being the stars of that fund.

And to be 1 of those companies, 1 of 2 or 3 things has to happen. What's happening in Abridge is probably my favorite example of a company that can become something that could be like a 100×.

Well, one of the things I like about that, and I think, is that if you think about Abridge, or a bunch of the winners that have emerged, including Twitter, they're often not straight lines. It's not like they popped immediately and it was obvious, right? Abridge in 2016, you know—

And we're seeing a lot of companies now that feel like they are straight—

Lines. Obviously, it's unlikely that that's the story.

15. 20 years of USV's biggest wins

That's a good point. So, to dig into—I guess this pattern, if there is one, over time—is it obvious to see that in a company? Is it a straight line, to Rebecca's point?

If we go back, USV has probably now had over 10 early-stage funds in the life of USV, and every single one of them has produced 2 or 3 really great companies. If you just sit there and list them, in our first fund we had Twitter, Etsy, and Indeed.com.

Indeed.com—we still see ads on TV. We haven't been an investor in that company in 15 years, but I suspect that company is 50 or 100 times bigger than when we sold it a decade or more ago.

And then, in our second fund, we had MongoDB and Twilio, which were products that developers used. Then we had Stripe, Coinbase, and Duolingo. Now we're getting into the more recent era. Abridge is a 2016 company, and I've left off a few, but there are probably 20 or 25 companies that we've invested in over the last 20 years that I would consider to be extraordinary companies—not just in terms of the investment that we made, but in terms of what they've become.

Indeed is really interesting. We were there at the very beginning. It was Paul and Rony and a couple of engineers, and they had this idea. Today it's probably the largest job-hiring system in the world, and it just came out of nowhere to become that. You watch these things emerge that way, and it's remarkable.

Twitter—now we know what it is. It's one of the most important pieces of media infrastructure in the world. When we invested, it was 5 or 6 people, right? So that's pretty interesting.

And one of the things that we can give founders—and we're not alone in this, by the way—is that there are plenty of people here at USV who have that kind of perspective on what it takes to really build a generational type of company, and the fact that it's not a straight line, right?

Coinbase was a series of little moves and then flat for a long time, then a little move, flat for a long time, with a couple of big breakouts in 2017 and 2019 that led to it becoming a public company. So it's never a straight line to the top, and sometimes there are management changes. Twitter's sort of famous for having a revolving door of CEOs for a long time.

So that's its own set of things that you have to work through sometimes.

But if you take Abridge, for instance, the idea of the product was very similar, but it went down a very different go-to-market road for years. That didn't get them anywhere, and it turned out not to be the go-to-market road that made sense for that product at that time. What was it? Consumer.

Consumer.

The idea was that you would walk into a doctor's office and record the appointment.

And so you think you back a whole business, right? You think you back this idea that patients want to record themselves in their medical appointments for XYZ reasons, and you're backing this team to go after that. What you learn you're backing is not that. You're backing a team, a product sense, and a direction, right?

What you backed was the idea that Shiv could build the team that took this kernel of an idea that was directionally right and kind of wrong in the details to start, and navigate it to the place where the right product—and then, as Fred talked about, eventually the right series of products—met the right go-to-market at the right time. That sometimes happens right away, and sometimes it takes a while. You need teams that are going to be able to go on that journey and fight that out.

It's usually an extraordinary founder, though. Not always, but Shiv's an extraordinary founder, and the other startup that we backed in Pittsburgh, Duolingo, again had an extraordinary founder.

More of a straight line.

No, no. The first product—

What was the first product of Duolingo? It was translating foreign-language documents.

No, it had nothing to do with that.

He's amazing.

Founders, as Rebecca said earlier, who can combine a vision for something that they want to exist in the world with the ability to tell that story and galvanize the right people, the right investors, and the right product—that's what we've all seen. We know when we look at it, we're like, “Oh, that's a great founder.”

But it's easier to see it when they're the CEO of a public company than when they're just walking in our door with an idea. And we get that wrong more than we get it right.

How do you get it wrong?

We think somebody can be a Luis or a Shiv, but it turns out that they can't get out of their own way. They overthink things. They struggle recruiting. They struggle retaining. They're a micromanager.

They struggle making hard decisions to make the big changes early enough, right? To say, “This go-to-market isn't working. Let's try something really different.” It's hard to burn boats as startups.

Mike's been a founder, so he has a lot of empathy for this.

Not a straight line.

Yeah, it's never a straight line. But everything these two are talking about right now is what I love about USV being thesis-driven. All of the examples that they both just cited, if you think about them in the context of when the bets were made, were extremely early.

The Coinbase bet was very, very early in the history of crypto. The Abridge bet was very, very, very early in the history of AI and data capture. And I think Rebecca's framework that she just laid out a few minutes ago about a great team and a direction of an idea is so important and so key. I think that's how USV has been successful: these bets were early.

Another one that we've been talking a lot about is Radiant, which Fred is on the board of. That bet was made before it was obvious to everyone in the world that energy was going to need to be so abundant for AI. But this team had a thesis and a point of view about the importance of energy, which is what led to that investment. And now, sure enough, how many years later—

We backed Radiant in 2021, so we're 5 years into that one.

Anyway, it just speaks to the power of thesis-driven investments. But I think, to your point and to your answer, Fred, we don't always get it right. Sometimes not only is the team or the person wrong, but sometimes the thesis is wrong.

Sometimes the benefit of thesis-driven investing is that you can do a few things. One, you can bet on spaces that are more on the come or less obvious because you have a prepared mind and perspective that you're bringing to it—something you believe. You're not just seeing a company and seeing what they believe; you have a belief.

You can also start the conversation with a founder in a different place, right? Because you're saying, “I have a set of hypotheses on something, and you have a set of hypotheses on something. Where do we align?” Which I think is actually a lot more fun and probably sets you up for a relationship in a different way, but also tends to get us to a better outcome than otherwise.

The risk is that there are a lot of ways you can be wrong. You can be wrong about people. You can be wrong on the whole thesis. You can go down a road that isn't the right road to go down. But when you go down the right one, it's worth it.

Today's episode is sponsored by VCX by Fundrise, the public ticker for private tech, allowing investors of all sizes to invest in venture capital. Learn more at getvcx.com. Some of you may not have heard this yet, but our sponsor Public just launched something called generated assets, and it brings AI into investing in a way I've honestly never seen before. Here's how it works. You type in an idea like AI powered supply chain companies with positive free cash flow or defense tech companies growing revenue over 25% year-over-year. Publix AI then dispatches a swarm of agents that scan every single US stock, evaluates them, and instantly builds a custom index around your thesis. What really stands out is how clearly it explains why each stock is included. And before you invest, you can even back test your idea against the S&P 500, so you're making decisions with real context, not just guessing. And beyond generated assets, Public lets you invest in stocks, bonds, options, crypto, all in one place. They'll even give you an uncapped 1% match when you transfer your investments over from another platform. If you want to build a portfolio that actually reflects your thesis, visit public.com/sourcy, paid for by public investing. Full disclosures in the description. Founders scale faster on deal. Set up payroll for any country in minutes. Hire anyone anywhere. Get visas handled fast. And get back to building. Visit deal.comsourcy. That's deeel.comscory.

I guess, to distill this theme down, what are the key behaviors or actions that successful companies make that win each technological shift? We're in a major technological shift right now. How have you seen that throughout time? What are the key behaviors of those teams or of those founders, and maybe the key traits of those types?

It's getting the idea right. What I mean by that is everybody knows that we're all going to have personal agents. But if you look at each of the products that are out there, they're all a little different in terms of how they feel and how they work.

Getting it right is the right set of features stitched together in the right way and delivered with the right business model, and then also with a team that has the ability to access capital and scale against the opportunity if they actually catch fire.

So I think it's not just the ability to raise capital. There are lots of people who are great at raising capital, but they can't build a product that really wins. And there are people who build a product that really wins, but they can't galvanize a team or money, and so they get beat. It's the ones who can do both that ultimately really win.

Was identifying that aspect of fundraising ability as important maybe 10 years ago? Was it as obviously important 10 years ago?

The interesting thing, and Rebecca alluded to this earlier, is that sometimes if a product really catches fire, it can raise money even without the right team around it. And then you, the founding team, and the board have to scramble to figure out how to fix the team and get it right so that all the things can come together.

I would say that was true at Twitter. That was also true at Etsy. So it isn't always the case that the right product emerges from a team that has maybe the right stuff to make it a lasting business.

16. Will Anthropic compete with its customers?

Why is one of the categories called Rebel Alliance?

“Rebel” is just in the sense that there are now the established winners in AI, which would be OpenAI, Anthropic, Grok, and Gemini—I guess all that's Google.

And then there's this open-source world out there of open-source models, open-weight models, open-source harnesses, and all this other infrastructure that you can stitch together to build something that looks like Claude or looks like Cursor or some sort of full-stack product.

We call it the Rebel Alliance because a lot of developers out there don't want to build on a closed stack. They're worried that if they build on top of Claude, for example, just to pick on Anthropic, somehow Anthropic will ultimately compete with them and then they'll get cut out.

And this famously sort of happened in the early days of social media. People built lots of businesses on top of Facebook, and Facebook ultimately decided they wanted to be in those businesses themselves, and they got cut out. So developers out there have this concern.

I need a stack that I can control and trust. We call it the Rebel Alliance because there are people who are kind of rebelling against the leading emerging stacks and trying to construct their own.

It turns out that stack is actually very large, and there are lots of layers—not only for opportunity, but for the different needs of developers. There are things like alignment with the models and making sure that the models are aligned with the needs of the enterprise or the customer. There are things like memory and routing, and there are the models themselves, making sure that they're open-weight so a developer can actually manipulate them the way they need to.

17. Why Coinbase moved off Claude

We feel like not only is there this uprising of developers who don't want to be beholden to one player or ecosystem, but it's also a very large space with lots of layers, which is why it represents an entire thesis for us. We had a moment onstage last week at the USV CEO Summit where I was interviewing the CTO of Coinbase about some of the changes they've made organizationally to become more AI-centric and how they do the work. He casually mentioned that they moved away from Claude and built their own harness—their own coding-agent harness—on an open-source product called Pi.

We're an investor in this company, Arendelle[?], which put Pi out into the market. I didn't even know that had happened. I was just like, "Yes, you know, it's great." But that's true of tons of developers.

Any other areas?

18. Your AI lawyer is the worst it'll ever be

I think energy in general was an area that we staked out early. We basically said, "Look, if you want a simple way to bet on AI and not have to think about all this other stuff, just bet on energy."

We're not just doing that, by the way. It's 20% to 30% of our portfolio, but we feel like it's a very levered bet on AI.

19. An AI doctor that writes prescriptions

It's a good one. We talk a lot about "Obliterate, Don't Automate," which is another phrase for the investments we're making at the application layer of AI. The way we think about that is the idea that a new—

AI is going to kill us all. No, no—

No, no. Opposite.

—that it may obliterate an existing market. There was an era of software over the past decade—SaaS, enterprise software—that was really about automating businesses and markets that already existed. But with AI, businesses and consumers can tap into intelligence that was previously gatekept by really large organizations or institutions, at a fraction of the cost.

We believe that through this, entire markets can literally be restructured through the democratization of intelligence. A great example is healthcare. We have an investment in a company called Doctronic, where they're leveraging AI not just to make the existing healthcare system more efficient or practical, but to literally give consumers a doctor in their pocket in the form of AI—

—that can actually write prescriptions in some states.

Really?

Utah, right now.

What kind of prescriptions?

You name it. What do you mean?

Oh, no.

Probably not.

Probably not controlled substances. It's not controlled right now, really.

But you can see—

—anything you want.

But you can see that if everyone had access to a doctor literally on demand, immediately, for virtually no cost, what that would do to the existing healthcare market in this country.

One of the things that I love about "Obliterate, Don't Automate" is that it's perfectly applicable to this moment in time, but it's probably the best articulation of the thesis USV has always had. Fred actually wrote a blog post in 2014 called "Obliterate, Don't Automate" about the USV thesis.

We've always believed in the opportunity to restructure markets and drive the value back to the end user, away from the centers of power in the middle. You can do that with a whole new tool set right now. The opportunity to restructure healthcare, education, manufacturing, or whatever gives us the opportunity to drive value out of the middle to the end user in a totally different way because of the tool sets at hand.

A simple way to think about this is: don't hire a lawyer; use an AI lawyer. Don't hire an accountant; use an AI accountant. Don't hire a doctor; use an AI doctor.

No human whatsoever in the loop. It's not about selling AI to doctors, lawyers, or accountants. It's like, "I'm not going to have an accountant. I'm literally going to tell my AI to prepare my tax return. I'm going to give it access to my bank, and it'll do the numbers and send the IRS my tax return." I don't need a human, and I don't need to pay a human.

I would say 20% to 30% of the legal work that we do here at USV is no longer touched by a human lawyer. We know what our AI lawyer can do and what our AI lawyer shouldn't do. That's increasing over time, but we're not careless about it. There are certain things, like issuing a term sheet, where we don't need a lawyer to do anything.

No.

How many of those have you issued?

Right. We can train a model on all our previous term sheets and issue them. The thing that's exciting now isn't where everything is; it's the slope. If you just play out that slope, it covers a lot of—

Your AI lawyer is as bad as it's ever going to be.

It will never be as bad as it is today. The company that we recently announced, which we talked about as sort of an incubation startup, is another great example of this. The whole idea of that company is that you don't need expertise around the stock market, and you don't even need to pay a financial manager to come up with a portfolio strategy for you and then deploy it. You can just have the AI do that.

That's another example of breaking down the barriers around gatekept institutional knowledge through AI.

The age-old question would be: Why can't incumbents do that?

They won't do it. They have people. A lawyer is a perfect example. We say to a lawyer, "You can issue the term sheet. We're just not going to pay you to do it." And they're like, "Well, why would I do that? I have people to pay." Fine. Then I'll just use an AI lawyer to do it.

Their business model depends on the people. The incumbents cannot—and I think it's particularly powerful in people-based businesses—offer something for free.

Our partner Brad, I think, always says this best: The best time to bet against an incumbent is when what you want them to do fundamentally threatens their current business. It's very hard for major incumbents to burn their own boats and take that kind of risk.

There is plenty of news about certain law firms hiring engineers and machine-learning talent—

—and JPMorgan is going into AI. Did you see that?

Yes, exactly. They're trying to build the models, and some of them will do it and some will try, but overall it will cannibalize their own business.

Right. Google has done, I think, an admirable job. They know that Gemini is going to cannibalize Google Search, and the numbers are staggering. I saw last week that search is way, way down. It's dramatic.

They are doing it, but they have all these other businesses. They've got YouTube, they've got Waymo, they've got Gmail, and they've got Android. They can literally kill the thing that funded all those other businesses because they now have all the other businesses. If all they had was search and they had never built those other businesses, they'd be in a world of hurt right now.

Yeah. A lot of publicly traded incumbents are going to have a really hard time doing this because they have to answer to their shareholders and keep revenues going up and to the right.

But I think your fundamental question is the one we're thinking about all the time. It's not only why incumbents can't do it; it's why anyone can do it.

Right.

Right. Why can't I build that agent to do my taxes? Why can't I train my own? We trained our own legal model; we didn't buy it. So what is going to create advantage in this moment, when building is faster and more efficient than ever?

That's something we're thinking about. It goes back to network effects, speed, team, and product decisions that are going to make these abstractions the ones people rely on, as well as categories like physical AI, data at the edge, energy, and things that operate in a somewhat different way than software.

So, an area where all of this is directing us—it’s part of the Rebel Alliance—is trust and security. How are you thinking about cybersecurity and all of those types of things as we obliterate and automate everything?

It’s definitely an area that I think is about to change. A lot of cyber focus over the past few decades has been focused on the enterprise. It’s probably going to undergo a lot of change at the consumer layer now, just because of how many consumers are going to be, I think you said earlier, handing over everything to their agent, whether it’s Instinct or Grockbot or Muse.

One of the things I was thinking about just this morning was the concept of password reset. That’s going to have to change. If you give your agent access to your email, which is where all your password-reset emails go, it will automatically be able to access literally every account you own. So, it’s definitely an area we’re thinking about. We haven’t traditionally made a ton of investments there, especially at the enterprise level, as we mentioned.

But it’s an area that’s going to change dramatically.

We have a portfolio company. They call themselves Common Fabric now.

They just announced, right?

Yes, Common Fabric. It’s a sandbox that’s focused on this very issue: How do you give users the security that they want, but also the power that they need? They have a particular approach. I think it will be interesting to see whether that approach or a different approach emerges, but we have a lot of conviction that users are going to need some new help with security around this stuff.

People get spooked by those AI videos all the time.

All the time. We’ve always been interested in this idea of security, privacy, and control with the user versus the platform. We’ve done that with more and less success over time, in terms of how much the end user actually cares about it.

There have been companies like DuckDuckGo that had enough of an audience that really cared about privacy and search, and that created a really, really strong business around it. There have been plenty of companies where we may want users to care about privacy and security, but there’s not enough of an incentive system for the end user to really care about it. That may be changing.

Right. It may be part of the conversation in a big enough way where things like on-device and private models, and these kinds of personal-security and more consumer-oriented products, may have a moment that they didn’t have in previous iterations because of the awareness, the scale, and all those kinds of things.

I also think things like biometrics may have a different kind of moment in terms of proving humanity. I think we may be entering a time where things that previously struggled to get attention have this moment to grow.

Molly and I saw a company this weekend that’s doing local models on your phone.

So, we’re there.

20. Fred Wilson's been hacked 4 times

Yeah, I think we’re there. Wow, surprise. Well, Fred, you’d know of all people, though, that it doesn’t just stop at that. Even in crypto, you could click a link and your whole wallet will be erased.

I’ve done it.

You’ve done it? Oh, wow. Do you have anything left?

No. No.

I should be worried.

So, what stops your agent from clicking the link? I have owned crypto assets for 15 years, and I’ve been hacked 4 times, but I still have about 99% of all my crypto.

Oh, not bad.

But I’ve been stupid and done things I shouldn’t have done, and not had my crypto as secure as it needed to be. I’ve learned the hard way, as most people in crypto have. If you talk to most people who have been in crypto for as long as I’ve been, they will tell you something similar to that. Hopefully, the ratio of losses to what they’ve kept is the same as mine, because I feel good about that.

There are some stories out there that are horrible, where people have lost everything. But I think most people got hacked, and it scared them enough to take the proper measures.

I almost lost all my crypto once. I got very close to being scammed. I was on the phone with a scammer for about 15 minutes, thinking they were legitimate.

Yeah, it was bad. My wife was like—

“You need to hang up the phone right now.” Sure enough, she was right. But I came very close to losing all of my crypto.

21. Coinbase: A security company in disguise

I mean, Coinbase—a company that I’m still on the board of, and one of our really great investments over the years here at USV—that’s a big part of what they do. They have cold storage, where they store assets super securely for their customers. There are 2-factor authentication, delayed withdrawals, and all these things that they do to protect those assets and literally protect their customers from their own stupidity, which I’m guilty of, by the way.

Yeah. In many ways, it’s a security company more than anything else, right?

Yeah. I mean, that’s an interesting analogy because people wanted to get into crypto and were interested in it, but it felt dangerous because you could get scammed all the time, and there was a big privacy risk. Coinbase came along and said, “Just give it to us, and we’re going to handle that part of it for you”—a trusted brand.

No offense.

Exactly.

Marketing.

Yeah, and most—

They’re the most trusted brand in the sketchiest industry ever.

That should be their tagline.

No, they make it literally seem so normal. It’s awesome. I’ve talked to them about it. Even Kate—they do a great job marketing.

And most people don’t know exactly how they’re keeping it safe, right?

Well, that’s on purpose.

Totally. Probably.

Yeah. Also, they don’t really care. As long as you trust that it’s safe, you don’t really need to know. I wonder if we’re at that tipping point or moment in AI as well, where everyone’s interested and we know it’s going to provide a lot of utility, but we’re looking for these trusted brands and platforms that can give us the comfort that things are in good hands.

I went to their last product launch in New York, just a couple of blocks away, and that was one of the questions that I asked both Kate and Brian: “You guys built the most trusted brand in one of the toughest markets. What do you think all these AI companies are getting wrong?”

What did they say?

Well, they’re telling people they’re going to lose their jobs and die, so that doesn’t help.

That doesn’t help.

It definitely doesn’t help.

It’s a good way to destroy trust.

Yeah.

I hope you don’t mind—I went over a bit because this dynamic just works really well. It’s awesome. We might not have that much time for solo questions, but I figured this is just awesome.

Maybe we just do the quick office thing. The 3 of us can just do that. You have to go in a few minutes. What time is it?

9:52.

Yeah, I’ve got 8 more minutes.

Okay, cool. I figure just 2 or 3 more questions. This is awesome.

Great. So, then maybe we’re—

You guys are so fun to talk to.

The vibe and the dynamic are great.

This is the vibe. What you’re seeing is the vibe among our partnership and, frankly, the entire team. I don’t mean to make it just about the partners. It’s a very conversational kind of back-and-forth, wanting to hear what the other person thinks and collectively coming to ideas together. That’s how we work.

Okay, so I’m going to ask you the hard question now.

Sure.

Do you expect every company to become a trillion-dollar company?

No, of course not. If you expected every company to become a trillion-dollar company, you would not be taking enough risk, right? Venture is based on an outlier mechanic, but we don’t invest in companies where we don’t see the opportunity ahead of us.

Do you think it’s realistic to think that these companies right now will become—I mean, obviously, Anthropic is not that old. We have ChatGPT, which is a little bit older—OpenAI—but I maybe have 2 takes on the question. One part of your question is: Does every market have to be a trillion dollars, right? Can you get into a market that you don’t see today?

I think one of the things that USV has done quite well historically is be willing to bet on markets that aren’t yet there but are on the come. Crypto was maybe the best example of it, and energy right now as well, along with education—these markets where, if you were underwriting exactly the dollars spent in crypto at the moment of a Coinbase investment, you would never get close to being able to underwrite the value of what that became, nor what that market became.

So, you need to bet on markets that you’re excited about today, but even more excited about where they’re going. We’re certainly doing that right now.

We've been talking about things like robotics and what can happen because of sensor proliferation. None of that is particularly interesting today in terms of what is live in the market. But when we look at 2030, 2040, or 2050, do we think those can be trillion-dollar markets? Absolutely.

Radiant Nuclear, right? The market for nuclear energy today is almost zero. It doesn't really exist. But we believe it's going to be one of the most essential energy sources out there. So we have an interest in and ability to write forward on markets—on what we think markets can become—versus only what they are today, which I think is really important to our strategy.

I think there was a time when we could underwrite businesses of different scale. Part of it was fund size, and part of it was market dynamics. There was a time when number 3, 4, or 5 in the market could still accrue a lot of value. I am not as sure that's where we are today.

Others may disagree, but I actually think value is aggregating to the top players faster than it ever has before. Interest isn't really there in the longer tail. So we're less interested in saying, “Well, we're not in the number 1 or 2 player, but maybe this number 3 or 4 player can get enough value that it's a great investment.” I think there was a time when that worked. I'm not sure that time is now.

In many ways, every USV investment has to be one where you almost believe it's a TAM expansion opportunity. If you're evaluating the size of the TAM as a fixed TAM when you make the investment, it's probably not going to make sense.

But I also think that if there was a time when we felt that if a company could be worth $500 million, we could make an investment, then I think the number grew to $1 billion. I don't think a trillion is what we need to underwrite to at USV. I think if our biggest winners are in the single-digit billions, given our fund size and our ownership, we can produce really healthy returns for our partners.

And that's where it's been. There have been some outliers at $50 billion or $100 billion, but a lot of our biggest winners are in the $10 billion range, and that's produced enormous returns for us.

I asked this also because I just came from Bending Spoons, and they're obviously buying up a lot of companies. They're doing it in the low-$1 billion range. They did some earlier ones and that kind of thing, but it was remarkably like the ZIRP era of funded companies that were at $11 billion or $15 billion and got written down to around $1 billion or $2 billion.

But how do you think about that valuation sensitivity today, and where does it get irrational?

If we get into companies early enough, at the earliest stages and with the right ownership size, as Fred just said, even exits in the single-digit billions can still be enormous for us, given the small fund size of our strategy.

This will always happen. There will be some companies that overshoot what they're really worth, and they'll come back to Earth. Not all companies—some companies are just rocket ships. But some companies have a moment where the world thinks they're worth $5 billion or $10 billion, but they really are worth $1 billion.

And that's perfect for Bending Spoons, right? Their model is really interesting. They can pick things up when the venture journey's over and grow those businesses with a different kind of capital structure.

22. From first-time founder to Suno

So, as we close out, this is my partner question, but my partner, my favorite partner, I don't know where my camera is. There's like 500 cameras here, but Sorcery is sponsored by Brex, and so they think a lot about performance, spending smart, or moving faster. I think about this, you've gotten this question before in terms of who you surround yourselves with, and performance is a measure of maybe the 5 closest people to you. Without pumping your book too much, I will try to get you to think outward of that. It could be an author, a historical figure, or just a longtime mentor of yours. Who are the people who really keep you inspired or keep you challenging yourself?

Founders. I mean, it really is founders. At our CEO summit last week, we all went around the room and said who inspires us the most. Our partner Brad got up and said, “Entrepreneurs.” And I thought to myself, “Damn, he did it again.”

That's why it's great to be a venture capitalist. Every day, someone walks in and you're like, “Oh, wow. That person is thinking about things in a completely different way and changes the way you think about something.” So I have to say, most venture capitalists will probably tell you the same thing: founders are the real inspiration for us.

I love when you see an outlier founder start to chug and hire that talent.

Are there any particular names that you guys have?

Yeah, I've been thinking a lot about this. I have a founder I work with named Zach Kanter. He's the founder of a company called Stedi. He sent an update yesterday. He's been at this for 10 years, which is a long time.

The business is cranking. That really happened in the last 3 years. He fought it out for 7 years, and then he got there and knew exactly how to take advantage of the moment when the stars aligned for him. Some of it was the external market. Some of it was a really high bar for the team, even through really hard times. Some of it was just getting the product to exactly the right place so that when it was useful, it was there.

That kind of intensity and lining up the puzzle pieces is so hard. And I think about the haves and have-nots in the portfolios I work with. A lot of it is that: Can you make these things line up so you're really ready at exactly the right time, and keep that intensity at full charge even when the wind's not at your back? I find that really inspiring to work with.

A thing I always say about the best founders is that, at a certain point, they reach this level where it's almost like they see the Matrix and can do anything. A great example—you and I know this, and we've talked in the past about Suno and Mikey Shulman. Suno is a USV portfolio company.

When they just started out, they were obviously so excited about the opportunity ahead of them in music, but they had never founded a company before, and they had never operated in music before, which is such a challenging market to operate in. After going through the paces for a couple of years and having a tremendous amount of success, Mikey is operating at an insane level.

If you met him when he had just started, it's just a world of difference. We learn a lot from him and from all of our founders.

Thank you very much, Fred, Rebecca, and Mike. I really appreciate it, and congratulations.