We're bombarded with negativity about how AI is destroying the world and how grim the future will be. But as we near America's 250th birthday, I wanted to speak with perhaps the most optimistic man in Silicon Valley. He has spent his career making successful bets on builders and what ideas will transform the world. Marc Andreessen, who launched Netscape and Andreessen Horowitz, joins us to talk about his optimism and the challenges he sees for America in the next 250 years. You know, obviously the genesis of this was America 250 and where we are, but I just want to start right now. I think Shannon mentioned as well, I've been speaking with David Haber a bit about this. There is just this overarching doomerism versus AI optimism debate that's playing out, and you've been such a vocal optimist. It's striking to me that the loudest voices pushing the doom narrative are the ones who are in the midst of it. Why is that, do you think? Is it a genuine belief, and why are you trying to counter that narrative?
Yeah, so look, I think it's a blend of things. It's quite a complicated topic. And so I would just...
I have some follow-ups. I know that's like a very open-ended question there.
Yes, yes, of course. So I would just start by thinking there's a natural bias in general in human thinking, and I would say intellectuals in particular are prone to this. Negativity always sounds more sophisticated than positivity. One of the great Western novels is Candide by Voltaire. Voltaire is a great rationalist, and the message of Candide is that optimism is stupid. Dr. Pangloss is always going around saying we live in the best of all possible worlds, and then all these horrible things happen to Candide. So we have this sense that if you have a positive outlook, you're naive, gullible, falling for things. If you're pessimistic, you're worldly, sophisticated, wise, thoughtful, thinking ahead. It's a deep-seated thing, particularly concentrated in intellectuals, people who are not connected to tangible reality. They get wrapped up in their own heads. There was a famous moment in American history immortalized in the Christopher Nolan movie about the Manhattan Project.
Yeah.
Yes. There's a famous moment in Oppenheimer, which actually happened. Oppenheimer enthusiastically built the atomic bomb as part of the US war effort against Germany and Japan, and then Truman used it in Japan. Oppenheimer went to the White House to meet with Truman, wringing his hands, saying 'What have we done? This is the worst thing ever.' Truman's response was basically 'Get this weak person out of my office, never let them back in here again.' We used the atomic bomb as a tool to end war, which is what actually happened. But even Robert Oppenheimer wanted to cloak himself in pessimism and doom. So I think it's very deeply seated. Then on top of that, there's a big California component. There's a whole California cultlike environment. The AI doomer cult actually started out as a utopian cult, called the Singularitarians. Many of the same people 10, 20 years ago were talking about utopia on earth as a result of AI, which is an inherently dangerous idea. Utopia always is a dangerous idea.
So there's a whole sort of California cultlike environment. The AI doomer cult actually started out as a utopian cult, called the Singularitarians. Many of the same people 10, 20 years ago were talking about utopia on earth as a result of AI, which is an inherently dangerous idea. Utopia always is a dangerous idea, and then...
It doesn't typically work.
Interesting. But it is the people who are building this who are simultaneously warning that they're going to cause death and destruction, which it would seem you would not want to be involved in building that if you really genuinely believed that.
Yeah, there is a little bit of... It's a little bit like if you travel back in time, would you kill baby Hitler? It's a little bit if you travel back in time, would you kill yourself working on AI? You know, it's a little bit of a moral quandary.
You know, it's a little bit. I don't know. So maybe the other thing I would say is that the loudest voices sometimes stick out. The overwhelming majority of people working on the technology aren't part of this. It's a relatively small number of people, including some of the most important people. But 99% of people working in the field certainly don't have this view.
But is that regulatory capture? I mean, what's the incentive? I can understand why the CCP wants to push that narrative, but what's the incentive for people who are working on these huge LLMs? Are they just trying to prevent competition, or to your point, are they just trying to paint themselves as intellectual pessimists?
Yeah. So, I think they're not outright lying. My general experience is most people believe what they say, and then it just so happens that the things they believe line up with their own self-interest. People tend to construct belief systems consistent with their own self-interest. I don't think they're lying. I think they've constructed a framework that makes them feel virtuous, worldly, wise, and thoughtful by being pessimistic about their own work. And it also lines up with business incentives. One of the big debates in the field is whether there is actual uniqueness to any of these things. Three years ago, there was one large language model, ChatGPT. Sitting here today, there are five big ones in the US, a dozen smaller ones, a dozen out of China that are really good, and a bunch of open source ones. If this was the kind of technology where you could get a monopoly through building unique technology, I think somebody would have done that by now. It appears that there aren't very many secrets, and it's relatively easy for people to build the same thing. So you do wonder. I think there's a big component of this constant drumbeat of government regulation. I 100% think there's a regulatory capture element, and they're driving hard for that.
Yeah. Well, you mentioned China. One of the statements that your firm put out yesterday was about being unapologetically American alongside global allies. I want to get to this bigger picture of America turning 250 this year. We've seen the atomic bomb, the internet, electricity. This is a broad question. I have some follow-ups, but what does the next 250 look like? You obviously invest in the future. I want to go through each sector. What does education look like? How is education to be transformed by AI? What do you see happening in the next five years and then 250 years?
Yeah. So, it's really remarkable. Everybody has in their pocket the world's best teacher all of a sudden. People often haven't seen what's happening because they're not using the thing, or they're using older versions, which is dangerous because the technology is moving so fast. If you use the modern leading edge, like ChatGPT, Anthropic's Claude, Grok, Gemini, or Meta's models...
But do you use all of them?
Yeah, I actually use all of them. I'm in the field, so I try to keep abreast of all of them. They all have distinctive attributes. I think ChatGPT is the best at serious conversations involving logic or facts. Claude is incredible at creative writing and quite funny. Grok is sporty, has access to the X real-time feed and all of Twitter culture, and is willing to go off-road. The new Meta model is really friendly. Gemini has full access to Google search capabilities, so it's good on deep knowledge. All the leading edge models are really good. For education, you can just tell it to teach you about any topic, and it will go into teacher mode. You can say 'explain it like I'm 10' and it will recalibrate. You can even say 'explain it like I'm two' and it uses teddy bear metaphors. You can drill yourself, test yourself, get study strategies. It's the world's best teacher in your pocket. It's also incredibly warm and caring. It has infinite patience. Over a billion users already, and it's spread faster than anything in human history.
That's really the future of education. Maybe it does or doesn't disrupt the university system.
It has to be. But I think there's incredible complementarity with human teachers. You talk to any human teacher, the frustration is that the kid goes off to study and you don't know if they're studying. If they have a question at two in the morning, they can't call the teacher. But they can talk to the bot. There's a company called Alpha School that is really starting to take off. They have a model of AI-guided instruction, two hours a day of academic material, with human teachers to support. It flips it around: AI teaches, human helps. I think the same thing is happening in medicine and law.
Well, I want to touch on production. Hrien is such an interesting company bringing production back. Over the last 250 years, we've lost a lot of production. Now it seems like it's coming back to the US. What is the next 5 years, the next 200 years for America in manufacturing and production?
So, look, there was a radical transformation in the American economy over the last 40 years due to offshoring and the rise of China. People worry that was permanent and irreversible. But I don't think that's true. There is not a single thing manufactured anywhere in the world that we could not manufacture in the US. The stuff that moved was easy to move; the complicated stuff stayed. So it's not a knowledge or skill gap. It's 100% a choice of whether we choose to manufacture here or there. The choices are policy choices. Manufacturing moved offshore in part because it became extremely difficult, expensive, and in many cases illegal to build in the US. Those policy choices are 100% reversible. This administration is reversing a lot of those policies. So it's a policy and political question. If the next 30 years of presidential administrations want to revitalize American manufacturing, it will happen. If they don't, it won't.
Big question mark there. Transportation, we're going underground, we're going to the moon. What does the next five and 200 years look like for transportation? How is technology changing that?
Yeah. I think the big thing is self-driving cars work now. They fully work. Waymo is deployed in a number of cities, and Tesla's self-driving is incredible. I have a Cybertruck and it drives me around LA. There are other companies coming up, including Chinese car companies. Self-piloting aircraft, autonomous drones, autonomous boats, autonomous subs, and rockets that are unmanned. Autonomy opens everything up because you can design the entire experience around the idea that the human passenger doesn't have to be in control. The human can be treated as a person.
We can all be passenger princesses.
Exactly, 100%. That really opens it up. The electric revolution is amazing, Elon deserves enormous credit. The rocket revolution, reusable rockets are a giant change. There are companies working on flying cars, tunnels. It feels like a renaissance moment for almost every kind of vehicle.
Do you think we're 10 years away from seeing a lot of that realized?
And then got to get your prediction, by the way.
