# Andreessen sees AI as a new kind of computer

Markdown copy of https://realalpha.app/s/andreessen-sees-ai-as-a-new-kind-of-computer/ on Real Alpha, the best minutes from long interviews with the people building AI. Cite and link the page, not this copy.

- Speaker: [Marc Andreessen](https://realalpha.app/people/marc-andreessen/) · Marc Andreessen is cofounder and general partner at Andreessen Horowitz. · Co-founded Netscape and co-created Mosaic, then co-founded a16z, a venture firm backing AI, crypto and frontier tech.
- Host: [Jack Altman](https://realalpha.app/people/jack-altman/) · General partner, Benchmark · Co-founded Lattice and ran it until 2023, then started his fund, Alt Capital.
- Show: Uncapped
- Episode: [Marc Andreessen on The Future of Venture Capital | Ep. 12](https://realalpha.app/v/53FImKtf2i0/), Jun 11, 2025
- Moment: 40:05 to 44:52 of the episode (5 min), playing on https://realalpha.app/s/andreessen-sees-ai-as-a-new-kind-of-computer/; the episode from there: https://realalpha.app/v/53FImKtf2i0/?t=2405

## Quote

> “I think the analogy is to the invention of the microprocessor. I think this is a **new kind of computer**.”
> — [Marc Andreessen](https://realalpha.app/people/marc-andreessen/) at [42:38](https://realalpha.app/v/53FImKtf2i0/?t=2558) · Uncapped

## Takeaways

1. O1 and DeepSeek R1 marked a turning point for him.
2. He expects AI to remake many incumbent industries.
3. He says venture investors should target startup-led disruption.

## About this segment

Marc Andreessen says reasoning advances made him confident AI would work, and compares it to the microprocessor. He expects both existing industries and new categories to be rebuilt around it.

From “Marc Andreessen on The Future of Venture Capital | Ep. 12”, Jun 11, 2025. [Watch the full episode](https://realalpha.app/v/53FImKtf2i0/)

## Transcript

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

[40:09](https://realalpha.app/v/53FImKtf2i0/?t=2409) sure we get everybody who can do it and get them to do it that's a good segue into ai do you feel that we're now at the beginning of what is like the new next important you know paradigm like is this cloud but on steroids yeah much i think much i think much larger and i'll explain why so um yeah so so i described you know i described described before right you know the the triangle people technology market the the technology is ultimately the driver is the the technological for venture the technological step function changes drive drive the industry and they always have right and so if you talk to the lps you can see this is like when when there's a giant new technology platform it's an opportunity to reinvent a huge number of companies and products that you know now have become obsolete and create a whole new generation of companies often you know generally

[40:51](https://realalpha.app/v/53FImKtf2i0/?t=2451) end up being bigger than the ones that they replaced and so so so and the venture returns map this and so they come in waves and the lps will tell you it's just like yeah there was the the PC wave, the internet wave, the mobile wave, the cloud wave. Like, that was the thing. And then, by the way, in venture, when you get stuck between waves, it's actually very hard, right? Because you've seen this for the last, like, five years.

[41:07](https://realalpha.app/v/53FImKtf2i0/?t=2467) Like, for the last five years, it's like, how many more SaaS companies are there to found? Like, just, we're just out of ideas. We're just out of categories. They've all been done. Yeah, yeah, yeah. Right? And so, it's when you have a fundamental technology paradigm shift that gives you an opportunity to kind of rethink the entire industry. It would have been very sad, by the way, if the AI breakthrough didn't happen.

[41:23](https://realalpha.app/v/53FImKtf2i0/?t=2483) Like, the state of venture would be sad, I think. Three years ago, this was, I mean, so when we were talking to RLPs three years ago, we're just like, basically like, you know, we're in, you know, so Chris Dixon has this framing he uses, he calls it your adventure. You're either in search mode or hill climbing mode. And in search mode, you're looking for the hill. And it was search mode. Right. And three years ago, we were all in search mode.

[41:42](https://realalpha.app/v/53FImKtf2i0/?t=2502) And that's how we described it to everybody, which is like, we're in search mode and there's all these candidates for what the things could be. And AI was one of the candidates, right? It was like a known thing, but it hadn't broken out yet in the way that it has now. And so we were in search mode. Now we're in hill climbing mode. Thank goodness, yeah. Big time. Yeah. And then, you know, look, like I said, on the technology breakthrough itself, I think a year ago, you could have made the argument that like, I don't know if this is really going to work because LLMs, you know, hallucinations, you know, it's great that they can write Shakespearean poetry and hip hop lyrics. Can they actually do math? You know, can they do, can they write code?

[42:10](https://realalpha.app/v/53FImKtf2i0/?t=2530) And now it obviously is. And now they obviously can. And this, I think for me, the turning point moment, the moment for certainty for me was the release of O1. So O1 from OpenAI, the Reasoner, and then DeepSeek R1. The minute I, and those happened kind of back to back. And the minute those popped out, you saw what's happening with that. And the scaling law that was around that, you're just like, all right, this is going to work because reasoning is going to work.

[42:29](https://realalpha.app/v/53FImKtf2i0/?t=2549) And in fact, that is what's happening. Like it's, you know, and I would say just every day I'm seeing product capabilities. You know, I'm seeing new technologies I never thought I would live to see. Like really profound. I actually think the analogy isn't to the cloud or to the internet. I think the analogy is to the invention of the microprocessor. I think this is a new kind of computer. Being a new kind of computer means that essentially everything that computers do can get rebuilt, I think.

[42:52](https://realalpha.app/v/53FImKtf2i0/?t=2572) So we're investing against the thesis that basically all incumbents are going to get nuked. Yeah. And everything is going to get rebuilt. Just across the board. Just across the board. Now, we'll be wrong in a bunch of those cases because some incumbents will adapt. But power law, the things that are right will be super right. Will be super right. Exactly. And then, look, the AI makes things possible that were not possible before.

[43:09](https://realalpha.app/v/53FImKtf2i0/?t=2589) And so there's going to be entirely new categories. By the way, is your mindset there that you should just bet on, like, obviously incumbents are going to win some percentage and startups are going to win some, but it's basically the dominant strategy as a venture capitalist to just plan to bet that startups are going to win it all and go for the power law? Yeah, that's right. That's right. And again, the reason is, because remember, two customer sets.

[43:28](https://realalpha.app/v/53FImKtf2i0/?t=2608) The way the LPs think of us, the way the LPs think of us is as complementary to all their other investments. Yeah. And so RLPs all have like major public market stock exposure. Like they don't need us to bet on incumbent healthcare, you know, whatever company, right? They need us to fit a role in their portfolio, which is, you know, to try to maximize alpha based on, you know, based on disruption.

[43:48](https://realalpha.app/v/53FImKtf2i0/?t=2628) And then again, and then just again, the basic math adventure, which is you can only lose one X, you can make a thousand X and you just like slam that forward as hard as you can. So when you have a moment in time worldview like this, do you? As a firm leader, do you give a directive that's basically like, hey, everybody, we need to deploy in this kind of way right now? Or do you just build a system that's always picking birds out of the flock from like the bottoms up and you're just like, well, they're smart.

[44:14](https://realalpha.app/v/53FImKtf2i0/?t=2654) They're going to see that every opportunity is good. Like, how much is it like a top down guidance versus, you know, the market's just obviously good all around? Yeah. So we don't do like I said, we don't do top down investment decision making. And so Ben and I aren't sitting saying, you know, we need to invest in category X or we need to invest in this company versus that company. And we don't run, we run, we have a legal investment committee, but we don't run a process where they come to us to get approval.

[44:35](https://realalpha.app/v/53FImKtf2i0/?t=2675) Because you're letting the leader of each group sort of make that decision. Yeah. And often in those groups, it's actually delegated further, it's delegated to the individual, individual GP or check writer. And the reason for that is we just think that the knowledge of knowing what's going on and which one's likely to win is going to be focused in the mind of the person who's closest to the specific thing.