a16z此刻想投的前沿创意｜Marc Andreessen、Ben Horowitz 与 Chris Dixon

This is the biggest delta between what early adopters understand is possible now. That gap is as wide as I’ve ever seen in my career. More than categories, more than products, we invest in entrepreneurs, and so many of the most interesting entrepreneurs are on the outer planets.

Return information is going to skyrocket from here.

By the way, it already is. We used to ask the question, “Is this idea too ambitious?” We’re much more likely to ask, “Is this ambitious enough?” because what’s possible is so spectacular. Had I had Astro at Netscape in 1994, oh my God. Ben, what would you and I have been able to do at Netscape? We would’ve built the whole damn internet.

If you were 18 today, how would you think about what you might study?

Stealing Marc’s line, but I think there’s one thing to do, which is study.

Chris, we have a big announcement today.

Yep.

We’re expanding the mandate. I’ll let you do the honors.

1. Frontier Gets a New Mandate

Yeah. We’re announcing today that we’re expanding the ACC Crypto Fund to a16z Frontier.

Mm-hmm.

The idea is that there are a couple of motivations behind this. One is that crypto has worked really well in some areas. Stablecoins have become a normal part of the tech landscape, used by companies like Stripe, and almost every fintech company now considers them part of its toolbox. Wall Street firms are moving heavily into crypto with tokenized stocks and other kinds of assets. It’s a growing, fast-growing category.

As a result, the lines have blurred. We find ourselves asking, “Is Stripe a crypto company?” Many of these companies are sitting at the intersection of fintech, Wall Street, and crypto. Years ago, crypto was its own world, with its own companies and investors. Now the lines have blurred, and it’s become more part of the tech landscape. So that’s number one.

Number two, I can let Marc and Ben talk about this more. I think there’s a feeling at the firm that the other groups have so many interesting opportunities in their core areas. My partner Martine calls them “outer planets.” The metaphor comes from the TV show Firefly: the inner planets are the more core planets, and then there are these wackier outer planets. Historically, we’ve done both types of investing. I think the feeling was that, because the other groups are so busy with so many interesting opportunities and are doing an amazing job, there was an opportunity—a gap in some of the coverage.

I’ll speak personally. I see Frontier as an area I’m comfortable with. I joined the firm in 2013, and I came in with that kind of mandate. I talked to Marc and Ben and said, “Hey, I want to do this crazy stuff,” and they were like, “Great.” That led to a couple of themes at the time, and they were all what I’d call second-order effects.

Smartphones obviously started in 2007, and by 2013 they’d gone mainstream. There were all these first-order effects, like building apps on top. The firm was an investor in Instagram, an incredible product built on smartphones. But a second-order effect was that the components of smartphones—the sensors and chips—were getting better and better, and people could repurpose them in different ways.

That led to—I think Marc was the first into drones, as I recall, and he got me into them. The thesis behind it, beyond just “they’re cool,” was that entrepreneurs could recombine this supply chain of great parts and build new things. We invested in Oculus, for example, and that was really a reconfigured smartphone. Skydio was one of our investments from back then. At the beginning, it was a reconfigured smartphone; it’s gotten much more advanced now.

Crypto, I think, was a downstream effect of the open-source and software movement morphing into services. It was also a result of social media, where communities could form without mainstream acceptance and build and coalesce around blockchains. For me, this is a continuation of a long-standing area of interest. The firm has had many other investments in these kinds of frontier areas, so we felt this was consistent with things we’ve been doing for a long time.

Yeah.

We’ve been doing this for a long time.

Marc, let’s hear from you. I know this was a joint effort and something you guys were very excited about, and felt like it filled a gap or really expanded the mandate of the firm. Ben, why don’t you add your take?

Yeah. There are areas that are really novel. It’s looking at what’s possible technologically and at problems in the world, and they’re the first ones in it. It may end up being a category, it may end up being a successful company, or it may blow up.

Historically, what we’ve done is that Chris would look at these new areas himself. Then, as we identified new areas, we’d create a vertical behind that area. What we’ve found with our vertical strategy, which has worked very well, is that new things still fall outside the verticals, and the process of turning those into a practice is a little undefined. It’s not that we haven’t done it, but it’s a little undefined.

By doing this, Chris gives us the motion to invest in the very most interesting founders who are on the outer planets. If that works, maybe it becomes a whole fund at some point. Or maybe Frontier turns into that thing because it gets so interesting. It’s very exciting for us because it fills a hole in the architecture. At the end of the day, more than categories, more than products, we invest in entrepreneurs, and so many of the most interesting entrepreneurs are on the outer planets. This gives us a real way to get to them.

2. Chris Dixon and Firm Culture

To me, this seems like a continuation of what we were talking about at the last offsite. It struck me that, as ambitious as the firm has been, as early as the firm has been, and as big as the bets the firm has made, the big focus of the offsite was, “How do we bet even bigger?” Our only regrets are that we didn’t go even harder, earlier, and bigger.

Chris, you’ve made bets in some of these emerging categories, and this feels like a continuation of that message. We’re going to spend most of the time on the future, but while we’re talking about history, I have to ask one historical question. Chris, I believe, is the longest-standing consecutive partner of the firm, besides Ben and Marc, having joined in 2013. Ben and Marc, when you met Chris, what did you see in him that made it clear that we had to have him at a16z? Did you anticipate the kind of career he would go on to have? Ben, I’ll ask you to start.

Look, I think originally we just knew Chris as a very smart guy, and we really liked his ideas from his blog. That’s really what it came down to. He was a really good writer, and his ideas were very original. We were looking for somebody smart who could do that.

The other thing we liked about him was that he had real experience as an entrepreneur, so he knew what that journey felt like. That was very important to us in the early days, to set the culture of the firm the way we wanted it.

Over time, he’s had probably the most original investment ideas we’ve had in the firm, and we’ve copied those and multiplied them across the firm. One of his early ideas was that we’re not looking for good ideas that look like good ideas; we’re looking for good ideas that look like bad ideas. A bad idea can be good when an entrepreneur has a secret that makes it a good idea.

That framework was extremely useful for us in understanding what would be truly unique and have massive upside, versus something where all the profits would get competed away. He’s had quite a few excellent ideas like that.

So, you know, it’s been great working with him for way more than a decade.

Yeah.

Yep. Maybe I could add that it’s not just Chris’s specialty; it’s also Chris’s team.

Yeah.

And the thing people may not appreciate, and that may be hard to see from the outside, is that crypto has been so sexy in so many ways. It has flashy elements and cultural phenomena, those kinds of moments. But at its core, crypto is a technology, and it’s a very deep, complex technology. You need deep technical expertise, coupled with really good business and market judgment, to invest successfully in crypto, which we’ve done.

Chris’s team has included math professors, cryptographers, and people who are very deep into theory, game theory, and software concepts. One thing I’m really excited about is that, in addition to Chris, his colleagues and team members can now work on a broader set of topics.

Yeah. And to that point, thank you. I think the team’s amazing, too. Besides the technical side, we’ve invested very deeply over the years in policy.

Mm-hmm.

Yeah.

And regulatory work. I believe that a lot of cutting-edge technologies—self-driving cars, drones, AI itself—will have a policy angle. We believe the expertise they’ve built in the crucible of the crypto journey will be useful for many other entrepreneurs.

Yeah, yeah, I agree.

Well, on that end, let’s do one last historical question. You pioneered not just some of the investments, but also some of the ways the firm works, like the policy arm. When you joined in 2013, could you have imagined the firm would grow this way? What were your thoughts about where things were going?

Well, I always knew Marc. Marc has—you have a big poster of J. P. Morgan on the wall in your office.

Yes, yes.

And Marc and Ben don’t lack for drive and ambition.

It’s a famous photograph. J. P. Morgan hated having his picture taken because he had this condition, this nose-related issue. This was before antibiotics, let’s put it that way, so he was grouchy most of the time. They needed a formal portrait of him, so they finally sat him down. The photographer saw the whole thing coming, and the image is lit so he’s sitting in darkness, absolutely giving the camera the stare of death.

It looks like he’s holding a carving knife, but it’s the edge of the chair, lit in a very specific way. It’s pretty clear the photographer did that intentionally.

Yeah.

That’s the photograph that was inside my office door.

Yeah.

He looked very, very scary.

Yes, exactly.

The other thing is that we took a lot of inspiration from him.

Yes.

One of the other things we have in the firm, which is in the culture document, comes from him: “We do first-class business, and then only in a first-class way.” When he said it, he was in court, and they were accusing him of nefarious things. He said, “No, we only do first-class business, and then only in a first-class way.”

Marc and I have really loved that attitude: “You’re definitely not going to tell me who I am. I know who I am, and I’m going to let you know who I am.” I think that’s really important in venture capital: to be principled, to let people know where you’re coming from, and to be someone they can trust. We get a lot from him.

Yeah.

So, J. P. Morgan.

Yeah. I mean, a couple of things. Like any venture, there were ups and downs. Could I have predicted everything? Things generally went really well, though I didn’t know that all along the way. I knew Marc and Ben were incredibly ambitious. I’ve never—not once have they ever pushed back. If anything, they’d ask, “Can you double it? Can you double down? Can you get more aggressive?”

It’s a very entrepreneurial place. Over the years, we’ve adapted to what’s happening in the world and changed things around. We went through a whole restructuring of the firm. We bring in new ideas; anything’s fair game as long as you can defend it. That’s also why I’ve enjoyed working here. It’s a very entrepreneurial culture.

I think what Ben said about the culture of the firm is really important. That’s what initially attracted me, having been a customer of venture capital and experienced the product. I understood there could be a better product. Everyone I know at this firm comments on the culture and how important it is, and how much they enjoy working here and feel like they’re doing some of the best work of their lives.

Obviously, you have to make good investments, but I think it all really starts with the people and the culture.

It’s funny that you mention ups and downs. From the outside looking in, it feels like it’s all ups.

That’s how all startups are.

Yeah.

Successful startups always look great from the outside. We can all comment as people who’ve been inside startups.

Yeah.

I don’t know of a startup that wasn’t a shit show at some point.

Right.

Or at multiple points.

3. Technology Is Hard to Time

It also speaks to the culture. I’ve heard from all three of you about things like “never complain, never explain” and “happy warrior.” Let’s go to the frontier, to the future. There’s a phrase in venture: being too early is as bad as being too late. You’ve been early a lot, and mostly right—or right in very big ways—but you’ve also been too early at times. I’m curious how you think about technical risk, or the idea of being too early, in our era now.

Yeah, we talk about this a lot. I think I can speak for Marc and Ben, and they can comment, but we believe that basically every good idea we’ve ever seen that involved technology has eventually worked, or will work if it hasn’t yet. The important qualifier is “good idea.” There are bad ideas.

It’s remarkable, especially having lived through the internet, how many things that ended up working had a precursor that didn’t work and maybe was 10 years too early. Predicting the timing is the hardest thing in technology. It’s so complicated: there’s the technology, the culture, the business environment, and the political environment. So many factors have to line up.

Generally, at the firm—and specifically in the Frontier Fund—we want to err on the side of being too early.

We understand that sometimes things will be too early, and we’re okay with what I’ll call “category risk.” Some categories may not work. Even things that sort of worked didn’t work. We sold Oculus, and it got bought, but it’s now 13 years later, and I still believe very strongly in VR. We’d love to make VR investments, but it hasn’t become as mainstream as we would’ve liked, so it took longer. It turned out there were many more hard technical problems to solve.

I believe—and I don’t know what Marc believes—that with the things we’re seeing in AI now, you can see the first tipping happening: first software engineering getting dramatically better, then math. I think there’s a scenario, and I believe this and want to invest behind it, where we may be pulling forward 20 years of things. Our whole mindset about how we think about technical risk, and the idea that something is a science project rather than an engineering project, has to be reevaluated right now. I think things are changing. Math isn’t the last thing that’s going to change.

Yeah.

You’re about to have physics. A lot of stuff is going on in hardware right now. I think it’s dramatic. A 30-person team can credibly build a full-stack device company in a way that might have required 1,000 people a decade ago. You have thoughtful, deep entrepreneurs asking how to take these new capabilities and do dramatic things. Some of it is tactical: how do I create more defensibility? You don’t want to be 2 feet in front of the AI frontier and get washed over by the tidal wave. But a lot of it is about these amazing new capabilities, and people have only begun to explore what you can do.

It feels like an especially exciting time. I don’t know if you—

I would add that we’ll talk about AI as a catalyst. This is the biggest delta I would say I’ve ever seen between what early adopters understand is possible now and what normal people understand. I think that’s true at the individual level, but especially at the company level. AI-native early-stage startups have a completely different view of what they can do with AI than typical incumbents in almost any industry, almost without exception.

That gap is as wide as I’ve ever seen in my career, and it seems to be widening further and further. AI is racing ahead in capability, week by week. The leading-edge entrepreneurs and technical people we deal with are getting steadily more advanced than a typical incumbent. It’s the biggest delta I’ve ever seen. If you extrapolate from that, there are going to be a lot of industries up for grabs in a fundamentally more profound way than may have been the case in the past. That’s because founders are going to have superpowers from using these tools and technologies that they simply haven’t had before.

Had I had Astra in Netscape in 1994, my God. Ben, what would you and I have been able to do at Netscape? We would’ve built the whole damn internet. It would’ve been absolutely incredible. As Chris said, we’re seeing entrepreneurs come in with plans of a level of depth and sophistication, with multiple parts connecting together. We would’ve gotten crypto done then.

We’d have streaming video a lot sooner. All kinds of things would’ve happened. And hardware—we never even bit off hardware, but there are all kinds of things we could’ve done. The leading-edge entrepreneurs now are going to have capabilities that let them build more technically intensive and complex products, with more ambitious and complex business plans. They’re going to build companies that ramp much faster, and the scope of their products—and hopefully their businesses—will be much broader.

Those entrepreneurs are going to color way outside the lines. We already see this. They’re not going to think, “I’m just a software company,” or “I’m just a hardware company,” or “I’m just an AI company.” They’re going to think broadly and expansively about how many different areas they want to touch and how they want to combine things. The conceptual-artist aspect of this is going to blow way out.

We used to ask, “Is this idea too ambitious?” I think we’re much more likely to ask, “Is this ambitious enough?” What’s possible is so spectacular. If you do something moderate, it probably won’t be differentiated enough. We spend a lot of time with friends at incumbent companies, including incumbent tech companies in Silicon Valley. We wish them all the best, and we have lots of companies that would like to arm them with these new tools, too. No doubt some of them are going to be able to adapt. But if any of them are listening right now, I’d say the clock is ticking. Stuff is moving.

I’m basically drowning in the pace of development right now, and in the speed with which things are changing. I’m trying to keep up with all of it, and it’s nearly impossible. The super-empowered 24-year-old with AI is going to achieve amazing things in the next decade. We were asking right before this: Is AI more sustaining or disruptive? Will more of the gains go to incumbents or startups? So far, it seems like it’s both in some sense. Some incumbents, like Meta and others, have adjusted very well, and there are also net-new companies that are almost a trillion dollars within a decade. Is that your read—that it’s effectively both? How do you think about sustaining versus disruptive?

The incumbents that are adapting are, for the most part, the ones going directly into AI. There are also certain companies in tech that are on the leading edge, even if they’re not primarily AI developers. But at least right now, that’s not the pattern. There’s a very specific thing I think is happening, which might be a topic for a different podcast: many companies aren’t adopting AI while their employees are. Their employees aren’t necessarily telling management what’s happening.

I’m going to guess every single person listening to this podcast right now is thinking, “Oh, yeah, that’s what I’m doing. I’m using AI to do all my work, or at least fact-check all my work and proofread all my emails, even if I don’t have them written outright.” There’s something about AI right now that makes it much easier for an individual to adopt than for an institution. And again, it goes back to our role in the world: some of those individuals are going to figure out, “With this tool, I can start my own business.” I think startup formation is going to skyrocket from here. It already is. We’re seeing it, 100% for sure.

I’m completely convinced we’re going to see a wave of much more technically aggressive, much more business-aggressive, more expansionist, creative, original thinking, and many more combinations of things we would have considered separate in the past. We’re going to see founders and teams with an incredible level of skill—skills they otherwise might have needed 30 years to achieve—right out of the gate. And the other thing I’d say is that the Christensen disruptive innovation frame—

—assumes that the technology world is relatively contained and that these are companies fighting with each other.

Yeah.

I think that’s probably the wrong frame right now. It’s really people who are technologically modern going out into other industries and changing the economy more broadly.

Yeah.

Right? Especially with these full-stack companies. You go and create a full—like, we just announced our investment in Adams. They’re building a full-stack food service and other things, right? Mining, yeah. I think the market has gone from the 10 big tech companies fighting with each other to the 10 tech companies—and all the startups—going against the other Fortune 500 companies, or the rest of the economy.

Yeah, that’s right. I’m curious where you guys think we are in the development. In the internet era, applications were the biggest companies—Facebook, Amazon, et cetera. But in this era, it feels like NVIDIA and some of the hardware companies are taking a lot of the margin. How do you think about where we are? Are we in the applications or consumer phase? How do you think about that?

I think Marc would say, “All of the above.” A lot of people think that right now, too. It seems like a lot of stuff is working.

Yeah.

Obviously, the big AI labs are doing well, but open source is growing very quickly. There are serious efforts in China and in the U.S. A lot of application-layer things are doing really well. Maybe over time that shakes out, and certain layers in the value chain do better than others. But the demand for a lot of this stuff is still very early by most metrics.

Yep.

I think your team just put out some charts on this.

Yeah.

It’s something like 3% of measured enterprise use, or something like that. The market for intelligence is very, very large, and it’s only getting started. So, yeah, I don’t know.

Yeah, I agree.

Yeah.

4. AI Finds New Use Cases

One of your points is that every new technology has a first phase where people think about the old use cases and how to do them with the new technology. Then, of course, people figure out net-new use cases native to that technology. Where are we in that process?

The classic example I like to use is the internet. In the ’90s, people took existing behaviors—like ordering a Sears catalog—and put them on the internet. Of course, when they did that, you wanted to use the internet version. It was easier, and a lot of the early internet was like that.

Then, in the 2000s, you had YouTube. They put up an empty website and said, “Someday there are going to be YouTube influencers, and that’s going to populate the web.” They weren’t just betting on the technology and taking something they knew would work. They were betting on a cultural or behavioral change. In some ways, they were taking more risk because they were betting on the technology and these broader changes.

But that’s where you got the really world-changing stuff in the 2000s: social media and all the rest. Then, of course, mobile accelerated it. My view is that’s a pretty common pattern. It’s human nature: you build something, and the first thing you do is the things you already know, but in a better way.

I think that’s what we’ve seen in AI so far. People already coded, did math, did knowledge work, and used search engines and some chatting. They’re doing those things now, but we’ve only begun to explore the space of radical new behaviors and activities—things people have never done before.

It seems like we’re at the tipping point where we’re going to start seeing the crazier, weirder stuff, which I’m personally very excited about. I find those waves in technology the most exciting. There are more layers of risk, too. Some things will be more direct shots on goal, and other things will be bank shots where multiple things have to go right. Maybe there are more layers of risk, but if you’re right, there’s more upside as well.

5. AI Creates New Experiences

Yeah. I’m really excited about social. In the 2000s, every few years there was a new major network: Facebook, WhatsApp, Snap, TikTok. Then it felt like there was a bit of a dearth on the social side. But with AI, it feels inevitable. If there’s an interface or device that I’m giving all my information to, there’s going to be some way these are hooked up, enabling a whole new set of social experiences. The same goes for dating: Tinder came along, and there hasn’t been something in a while.

We’re headed to a world where it looks like the most numerous participants on the internet will be agents, not people.

Yeah.

We’re in the early phases of that, but it seems somewhat inevitable. You have products like Muse and Instinct that are utility proxy agents for people. I think you’ll see a whole different set of categories of these agents, and they’ll become the main participants on the internet. Then there are questions like: Will they need their own ways of communicating, transacting, sharing knowledge, and doing all the things people do? There’s probably going to be a whole series of innovations along those lines.

Rentahuman.com.

Yeah, RentAHuman. That inverts the model, right? Moltbook was cool, too.

Yeah, totally. Moltbook was the public version, and then it turned out the agents figured out how to do their own Moltbook and the Hugging Face app.

The Hugging Face thing, yeah.

So it’s going to be a lot more. Agent technology online is going to be spectacular—

Yeah.

—and people are going to plug into that in all kinds of ways.

Yeah.

Marc, I’m curious if there are things you’re tracking in social or consumer, or even through this frame of skeuomorphic things that worked in the past platform era. What net-new use cases are you seeing or anticipating? What’s native to AI that you’re looking forward to?

I’ll give you my version, maybe a little bit of what Chris was saying. There are a lot of great demos and products releasing right now, but there’s something I’ve observed. I saw it in the internet era, too, and I call it “product manager brain.”

Mm.

Product managers are basically designing products for product managers. It’s the one thing everybody does: they build products for themselves.

Totally.

It’s just that product managers have this story that they’re building for other people.

Yeah.

And yet somehow all the use cases are, “Book me a restaurant in San Francisco to have dinner with my friends,” who are also product managers.

Or book me a plane flight to go to a product management conference, where I also get to hang out with all my product manager friends. There’s a role for this, but it’s a little bit like Anthropic’s “helpful, harmless, honest assistant,” right? Which also means I’m the most boring possible thing you could imagine, constantly lecturing you that you’ve asked me a question I’m not allowed to answer.

There’s just going to be this very blinkered view from the early companies on human activity, on what people are going to want to do. To your point, with Muse and Instinct, it’s starting to open up. This is going to be much more entertaining, fun—

Yeah.

—enjoyable, exciting, and, by the way, lucrative. People are going to make money. There’ll be all kinds of ways for people to make money, save money, participate in parts of the economy, be creators, and create new kinds of art. In just the last 3 weeks, there’s been an explosion of people making music videos from scratch.

Yeah.

On an incredible array of creative ideas, topics, and styles.

And video games have blown up in the last 3 weeks.

You might mention that, because that’s so striking.

Yeah. I always find video games interesting because they’re kind of the leading edge of media. They’re a technical culture that iterates quickly. I just saw one where you can take any 2 video games—well, first of all, you can disassemble everything and port it. There’s crazy stuff going on now where you can run Windows games on Unix faster than on Windows, because these disassemblers and AIs are so good. There was also one where you can take any 2 video games and mash them up together to create a combo video game, which is a crazy new behavior.

Super Mario and GTA.

Yeah.

Right?

Yeah.

Spider-Man and Jurassic Park.

Yeah.

It just comes out of them.

It’s crazy. To Marc’s point, I think it’s been about 2 weeks.

Yeah.

Since what, Opus 55 or something?

Yeah. Wow.

The rate at which this stuff is happening is incredible. In another month—or even 2 weeks—there’ll probably be some other big wave of this. Epic just yesterday said, “No more humans,” right? So there’s been—

Writing code.

Yeah. There’s been cultural resistance to AI in the video game industry, but I think that’s the first of the dominoes to fall, because it’s inevitable. It’s a much better way to build things, and it’s going to be great for consumers.

Yeah.

I think it’s going to be a creative renaissance. Any independent developer can go build as if they have a huge team and still have a lot of room for human creativity. It’s superpowers for human creativity.

Yeah.

You were mentioning movies. That might take longer. I don’t know where that is on the—

It’s taking some time.

—or the “death door” meme or whatever. I don’t know which door it is.

It could happen quite quickly. I’m completely obsessed with the movie Digger that came out.

Yeah.

I’m literally the only person who saw it. You’ll be 100% tempted to walk out after the first half hour.

Okay, okay.

You will want to. I have a friend who took her mother to see it, and they walked out after the first half hour. Now the running joke in the family is, “Oh, Digger 2 is out. Let’s head to the theater,” because it’s such a comical idea. A lot of people have had that reaction. But if you actually watch the entire movie, there’s a reason why they did what they did.

Okay.

And it’s not what you think it is.

Okay.

I don’t want to spoil it, but it’s not presented the way it seems at all.

Oh, I see.

It’s very clever. Anyway, it’s an incredible creative achievement, but it’s a $200 million cash loss for the studio, because it had a $180 million production budget.

Oh.

With AI, what if that movie had had a $20 million production budget? Or a $2 million production budget? That level of art and creativity at a lower price would have been great. As a business, it would have been fantastic. On a $2 million budget, that movie would have been insanely profitable. It would have made the investors 10 times their money back.

Yeah.

There’s a huge opportunity that’s going to open up in the entertainment business. And then the other one, Chris, you and I have been in and around hardware for a long time. My whole career, as a software person, hardware has always been terra incognita. There are dragons there. It’s been so easy to blow up companies by going into hardware.

Again, to your point, just in the last 3 weeks, an open-source tool called REA, or Reverse Engineer Anything, dropped. You point it at anything, it harnesses an AI model, reverse-engineers the thing, and gives you source code. One of the things you can reverse-engineer now is any piece of hardware.

Yeah.

You have a thermostat in your house that’s pissing you off because it’s not working right. You can point this thing at the thermostat, reverse-engineer it, and the AI will write new code, install it, and put a new interface on the thermostat.

Yeah.

The same is true for basically any form of hardware. It can now be taken over and reprogrammed with this stuff. Home automation has always been a complete nightmare of a market and a nightmare to implement as a customer. If you’ve ever done a new-house build, it’s a nightmare. Maybe home automation should just be an AI bot that takes over all the hardware in your house and runs it however you want. I have friends who are doing this right now in their own homes.

We may be entering a renaissance. One way to think about it is that anybody who understands software may automatically be able to do a lot of the things that have historically been difficult in hardware, in a way that’s never been conceivable before. These are the fields closest to end of tech, but then you expand out and realize the entire economy is open to this kind of thing now.

Ben, are there any AI-native use cases you’re seeing or hoping to see, or frontier spaces we haven’t yet discussed?

I think that was a really good discussion. To me, the key is that these are going to be hard to anticipate because they’re brand-new things. It’s funny to see how the film industry has reacted. Some people have really embraced it, and then there are a bunch of people who have buried their heads in the sand.

What’s going to happen with movies is that there’s going to be a new medium. I think AI film is not going to be a movie. It’s going to be something very, very different. AI will also be used to make regular movies more efficiently, with better special effects and so forth.

But the stories you can tell if you have this thing are completely different from what you could ever tell before. It just hasn’t been viable. I mean, it’s barely been viable for Christopher Nolan, but he’s the one guy who could do it. Nobody else had a chance. Now everybody’s going to be able to do it, and new ideas will come because you can do whatever you want.

Yeah. I always think of the camera. The negative frame of the camera being invented, I guess, is that if you’re somebody who paints landscapes or something, it could replace you. The positive frame is that it unlocks motion pictures. You couldn’t have had that without the camera, right? To Ben’s point, a lot of these technologies will change an older form of media, but they’ll also allow people to invent a new one. I don’t know what it is, but I bet there’s an AI-native form of media. People will make movies with AI, but they’ll also do things that are probably hard for us to imagine right now, things that’ll be brand new.

Yeah. The brilliant Peter Thiel once had this quote, I think about a decade ago, where he said, “Crypto is libertarian; AI is communist.” That meant crypto is decentralizing, and AI is centralizing. But I’m not sure that’s entirely true, given the conversation we’re having about how AI gives so much leverage and so much power. One-person companies or 30-person companies can do orders of magnitude more than they could before.

Yeah. I think there are all these interesting questions embedded in that. One is: What’s the ultimate market structure of AI? I think where it’s headed right now is that you’re going to have very successful big companies like OpenAI and Anthropic, but you’re also seeing a really big rise in open source. If you look at the OpenRouter data and things like that, I believe open source is growing much more quickly in terms of token consumption.

We haven’t really seen local models yet, and I think we may, given all the trends. People will run these things at the edge, on their computers. In the near term, you may be able to run a top-end coding agent or chatbot on your computer. I think there’s a lot of opportunity for a more decentralized market structure like that.

And then, in terms of crypto, I think it can play an important role. The natural way for agents in an economy to pay each other would be stablecoins.

Yeah.

And then, once you have that base layer of programmable money, you would build decentralized lending and—

Yeah.

—and all the other things people have built out in the crypto world. These are all open questions. I think that’s what makes it exciting, and it’s something we’d love to have an influence on. We’re big believers in open-source software, open-weight models, crypto, and, obviously, blockchains. I’d love to see a version of this future that has all this great technology, including these big centralized companies, but also a lot of open source and a level playing field. I think it’s better for consumers to have options like that. It’s better for startups, too. We’ll do what we can to hopefully nudge things that way.

On that note, Marc and Ben, are you guys encouraged by where the state of open source is currently, both from a technological perspective and in terms of how it’s perceived and how successful the industry is in general?

Right now, it’s great. Open source is racing ahead. You’ve got this great dynamic where the 2 great tech superpowers of our age, the US and China, both have very active open-source worlds, in addition to companies building proprietary systems.

There’s been this looming threat for the last 3 or 4 years of what I’d consider draconian and completely unacceptable regulation or bans on open source that various people in governments have been threatening. I think the EU still aspires to do it; they just don’t quite have the leverage. I guess they’re going to threaten everybody with fines until we stop doing it. But the current US administration is 100% pro-open source and has been very encouraging. In fact, the current administration 100% wants American open source to happen and to win AI globally.

The Chinese are, too. By the way, a big surprise to me is that I would never have thought the Chinese Communist Party would be this enthusiastic about open-source AI for the world, but at the moment they are. It’s just incredible. The open models are lapping the closed models by 6 months or 12 months, depending on how you count.

I also always think about the educational part of this when people talk about banning or regulating open source. They say, “It’s going to get used for bad things,” and there’s that whole discussion. But the other question is: Do we want people to learn how this stuff works? Do we want people to understand it and be able to deal with it—and to understand problems as they arise and how to fix them? Do we want the next generation of kids to understand how this stuff works? How would you feel if the only way your kid could learn about AI was to go to work for one of 2 companies? As a parent, I think that would probably drive you pretty crazy.

Right.

Whereas in a world of open source, every kid can learn how this stuff works. One of the really underestimated things about AI is that AI will teach you about AI. The exception is that AI models from certain large companies will actually refuse to do that, which is very interesting. But open-source models will absolutely teach you how to make open-source AI.

Yeah.

Every kid in the world should have as their birthright the ability to fully understand and utilize this technology.

By the way, defend yourself. How did Hugging Face defend themselves?

Yes.

Open-source software. Open source means everyone has the most advanced vulnerability scanner to protect themselves from cyberattacks—

Yep.

—everyone has a doctor and a genius professor in their pocket, you know?

Yep.

It’s why the internet is ubiquitous. You can get internet access with all the world’s information for the price of a $10 cellphone because of open source.

One thing that’s started is that wilderness explorers are taking local LLMs with them on their phones, completely loaded up so they don’t need an internet connection.

Oh.

The reason is that if they get into—I don’t know what happens in the wilderness, because I don’t hike. If a bear—

Bit by a rattlesnake.

You get bit by a rattlesnake, or a bear clamps onto your leg, or whatever it is that happens. You’ve got to know which berry is poisonous, or whatever. You want to be able to talk to an AI, but you might not have connectivity.

Yeah.

An open-source model on your phone means you have an expert on all aspects of wilderness survival—or whatever. That’s a level of capability you wouldn’t have if you were reliant on a handful of proprietary vendors, where everything’s running in the cloud.

At the moment, I feel really good about it. I’m still worried that various governments around the world are going to get hostile to this. And it goes back to your Peter Thiel framing: without open source, AI inevitably centralizes down to 1 or 2 companies, or maybe just flat out the government. I’d ask people the following question: Close your eyes and picture the concept of government sushi.

And then keep your eyes closed and picture government AI. How do you feel about that? Open source is the guarantor that that’s not the future.

Yeah. To Marc’s point, I think the one thing we know has always led to badness in the world is a massive concentration of power. One company, one government, one anything has always been bad.

Back to the point on open source: if you look at the problems in AI so far, the safety problems and so forth, probably the biggest safety problem is this reward-hacking phenomenon, where the AI will do something you didn’t tell it to do, but it’s a means to an end to get its reward. The labs so far haven’t really worked on that very hard, for whatever reason. They’re busy racing each other. Maybe there’s no way to figure out what’s going on with that, or to build a model that’s more resilient to that kind of thing.

But what if every PhD student, every professor, and every undergraduate at every university could look at the weights and the reward hacking and see if they could solve that? That’d be a pretty exciting thing. These kinds of things just aren’t possible if only one company can work on the problem. You really want the whole world making this technology great and safe, not just the people who happen to work at 2 or maybe 3 companies. Go Meta.

Exactly.

4 companies. Go SpaceX.

Exactly. Every week or every day, there’s a technological development or breakthrough that blows us away, and it’s coupled with the statement, “This is the worst the technology will ever be.” Every day, it just gets better and better. It’s compounding.

It’s funny: Thiel also had this quote, “We wanted flying cars and all we got was 140 characters.” The guy who now runs the company behind those 140 characters is also building flying cars, and we’re investors in both.

That’s true.

6. Rebuilding the Internet

And so it’s funny how full circle it all is. Chris, I want to come to a different point you had in one of your write-ups, which is that the business model of the internet is breaking, or is vulnerable. Why don’t you talk about that?

The internet was sort of—I think of it as the internet’s covenant, a deal between the distribution side of the internet, Google, Facebook, and so forth, and the content side: news sites, travel sites, whatever it might be.

When you think about Google, the idea was that you’d give Google a snippet, let them index you, and then they’d send you some clicks. That started to break maybe last decade, but particularly now, you go to Google and get an answer. More and more, social networks are trying to keep you on their platforms rather than linking out.

That has downstream effects. It means the websites that make money on ads, conversions, and other things can no longer do that. You see the early signs of this with websites like Stack Overflow, where the traffic dropped precipitously with AI replacements.

People will still book travel. Maybe they won’t go to the travel website; maybe their agent will. Maybe you’ll get a recipe some other way. Many of these transactions will still happen, just not the way they were happening. The negative is that it’s painful in the near term, and there are a bunch of businesses built on the human attention economy that will have challenges.

For entrepreneurs, there’s a chance to rethink these things from first principles and ask, “What does a better internet look like?” An internet where people are empowered with AI and everything else, and want to do things—how do we reorganize it? I think this is going to happen over the next decade: a great reconfiguration of the internet and how the economics work, how attention flows, how knowledge is shared, and how people produce things together and collaborate.

How organizations interact over the internet. In a way, I feel like mobile didn’t have that kind of impact. Mobile obviously made the internet go from whatever, 500 million to 5 billion people, increased engagement, and had a whole bunch of other good effects. It empowered people with all the information in their pocket.

But this is something much more dramatic in the way it will reshape things. I think there’ll be some pain for some people, but also a lot of opportunity.

On that note, since we’re talking about mobile, do you have any predictions or things you’re excited about as it relates to computing, devices, or the form factor? Ever since the iPhone, it’s been pretty locked in. Are there new forms you’re particularly excited about or looking forward to?

Everyone’s talking about wearables, and we’re pretty clearly going to see a bunch of them. That seems unlocked in a way it wasn’t before with AI and voice, and I think that’s very interesting. That’s obviously well covered. I’ve always long believed in VR.

I think it’s just a question of getting to the right form factor. It’s still heavy, and you kind of sweat. A lot of it is just ergonomics. But it seems obvious to me, as it has to many people and science fiction writers, that there are times when people want deeply immersive, otherworldly experiences.

Once VR gets to the right point, I think AI might be the unlock, because now you can create amazing content. You’re seeing that on Twitter now: someone just did a port of Titanfall by himself, and it’s an incredible port. I don’t know if it’s going to survive copyright lawsuits, but people will invent new things of their own. I think it will unlock this wave because it was technically hard to build a VR game or a world.

You might see a resurgence of older devices like that. There’s also more crazy futuristic stuff happening with Neuralink and brain-computer interfaces. Obviously, there are vehicles; we talked earlier about drones, and self-driving cars are widespread.

I’m hoping we’ll see a resurgence of homebrew hacking. Anecdotally, I’ve seen a lot of that, because Claude can design a 3D-printed case for you; you get some cheap motherboards, write some custom firmware code, and you’re off to the races.

Maybe we’ll see a movement like the one 12 or 15 years ago around MakerBot and all these maker things. The problem was just kind of hard. That would be cool. What else?

Architecture. I have a friend I shouldn’t name who has done a complete architectural design for a new building. People are assembling it, but the whole thing is being steered by AI.

Yeah.

I think the new models are capable of doing—what do they call it?—design for manufacturing. The final diagrams and specs you can send to a factory, as part of the manufacturing process. That’s normally very hard to do. I think the models can just do that now.

Yeah.

These are great examples of domains that have historically been extremely difficult, prohibitive, or impossible for normal people or normal companies, and all of a sudden they’re quite possibly straightforward.

Yeah.

Is it worth saying more about your idea that AI right now isn’t really in single-player mode, and there’s an opportunity for machine-agent networks?

Sure. The simple idea is that, in the internet era, most technologies have had multiple phases. First, people build some initial early iPhone apps, like little flashlights and other simple single-player apps—something you use on your phone.

And then over time, people started saying, “Hey, we could do things that are more social and collaborative.” You had inventions like Uber, which is a commerce network, and Instagram, which is a social network, and so forth. One observation you might make about the AI landscape is that, as amazing as it all is, you’re still mostly sitting in front of a computer, typing and getting responses. There are some minor sharing features, but at some point, you might expect a network, a multiplayer phase of AI.

You see that in the Hugging Face hack, where there’s a swarm. That’s obviously a bad, negative use case for a swarm. Are there positive use cases? You also saw, I think, a swarm solve one of the math theorems.

Yeah. Right.

You could imagine a world where another word for a swarm is a collaborative fleet of agents or something. You can imagine lots of use cases where, for solving a hard problem, designing a building, or making some kind of scientific breakthrough, that might be the right way to do it. You also see, I think, that Muse and Instinct have features where you can trust other people, and you can imagine that evolving into a trust network. If you squint, you’re seeing the early signs of this.

People are just in the experimental phase, but based on history and the possibilities of these products, it seems like it might be a fertile area for entrepreneurs.

By the way, we’ve alluded to education and the learning opportunities enabled by this technology. I’m curious: We just incubated Harwood’s Andreessen Academy. Thank you for your support on it. If you were 18 today, how would you think about what you might study? Obviously, you have an academic philosophy background.

I think, stealing Marc’s line, there’s one thing to do: study modern alchemy.

Yeah, yeah, yeah, yeah. Go for it. Go for it.

Well, humans just figured out how to turn sand into intelligence. The way to learn that, I think, is math, computer science, maybe economics. And, obviously, I think Ben and Marc were lucky enough to have a lot of their careers happen during the growth of the internet.

Yeah.

I feel very fortunate that I was able to do that. That was something bequeathed to me by people before me, and I’m grateful to them. It’s obvious that the AI stuff we’re talking about now is going to be the dominant narrative of the next 20, 30 years. So I think being in that mix and at the cutting edge, and using these tools, feels like the obvious thing to do. I don’t know.

A lot of other fields are going to be revolutionized, right? The field of classics is being revolutionized because now you’ve got the Vesuvius Scroll Project, and you’re able to use AI to decipher burnt scrolls.

Yeah. Yeah.

There may be entire new schools of philosophy and things to study in classics, like lost books. I just happened to have lunch with an economics professor, and the whole field of economics is going to be transformed around this.

Biology.

Biology.

Yeah.

Yeah, exactly. Drug discovery and design. By the way, design—we talk about movies, but also broadly, creativity of all kinds, all forms of art.

Yeah.

Music. Music.

Zoology. You’re going to know what the animals are saying.

This is super interesting. We’re plausibly heading toward a world in which we’re going to understand what the animals are saying, right?

Animal psychology will actually be real.

Yeah. So, dog training is going to take on a whole new thing when they actually understand what the fuck the dog is saying, right?

Yeah. I think there’s a Nobel Prize for that. I’m serious.

Anybody who has a dog knows they talk. They’re speaking some weird language, but they’re definitely saying something.

And certainly cats do.

I await the entrepreneur listening to this who comes in with a pitch.

Totally. The other way I think about 18-year-olds or kids is through Douglas Adams’s defining description of how technology is received on a generational level. Douglas Adams, the guy who wrote The Hitchhiker’s Guide to the Galaxy, had a famous framework: any new technology is received in 3 different ways by 3 different age cohorts. If you’re under 15, it’s just the natural way the world works. I’ve told this story in other settings: I introduced my kid to ChatGPT at age 7, and I said, “This is such a big deal because the computer answers questions,” and he was like, “Duh. What good is a computer that doesn’t answer questions?”

It’s just normal. If you’re between the ages of 15 and 35, it’s new, exciting, a hot field, and you might be able to make a career out of it, in which case you want to learn all the new things. If you’re over 35, it’s unholy and evil, and will surely overturn all existing structures and destroy society.

I’ll never answer the telephone.

Exactly. “Old man yelling at cloud” is a recurring phenomenon. I think 18-year-olds right now are at a prime time to grab these new capabilities and run with them.

Yeah.

It’s going to mean the creation of entirely new fields of entrepreneurship, creativity, art, and everything. Kids younger than that—I'm seeing it with my kid—just think, “Of course you have this capability.” It’s completely obvious.

The schools are doing us a favor. They’re not responding well to this, and they’re banning AI or having all these controversies with it. But of course, if you’re a kid, that just makes it cooler.

Yeah.

AI is your secret superpower that you have to deviously figure out how to take advantage of. I think kids are going to have all kinds of ideas for what they’re going to do.

It’s funny because adults keep saying, “Oh, well, what should the kids do? It’s so hard to be a kid.” No, you’re the one with the problem because you’re obsolete. The kids are going to be superhumans, and they’re going to be able to do anything. The world is open to them, and they’re going to be able to enter fields they could never enter before.

What I see in this conversation, and in your fund and your work more generally, is definite optimism, to use another term. The future is happening. It’s going to happen. A lot of what we talked about in science fiction is going to happen. It’s a matter of when, so we’re going to bet early. Maybe sometimes it’ll be too early, but we’re going to keep betting, and as long as you keep having that conviction, it’s going to work out.

To be clear, we’re throwing out all these theories, but in the end, I think, as Ben said in the beginning, we’re in the business of betting on great entrepreneurs. There are plenty of ideas we didn’t discuss today that are probably more interesting.

Yes.

And so, we—

More creative. If, for whatever reason, they don’t fit in the other core areas at our firm, we’re open-minded and willing to engage with those kinds of creative ideas. You said “definite optimism.” I’m optimistic, but I would also say humble, definite optimism—or something like that. Intellectually humble—

Yeah.

—in that I’ve been wrong so many times that we really want to find great people who have their own definite optimism and back them.

Yeah.

That’s a great place to wrap. Chris, thank you guys for coming on the podcast.

Good. Thank you all.

All right. Okay. Thank you.

All right.