Marc Andreessen on AI Winters and Agent Breakthroughs

Summary

---

The a16z Show

Episode: Marc Andreessen on AI Winters and Agent Breakthroughs

Release Date: April 3, 2026
Guests: Marc Andreessen (Co-founder and General Partner, a16z), SWIX (Swyx, Latent Space), Alessio Fenelli (Kernel Labs)
Main Theme:
A sweeping discussion with Marc Andreessen about the current state of AI, the historical “winters and summers,” the true transformative power of recent breakthroughs in agents, language models (LLMs), and software architecture. The episode explores AI’s historical cycles, infrastructural implications, open source, the future of coding, “proof of human,” and how AI could reshape capitalism, productivity, and organizational structure.

---

Episode Overview

Marc Andreessen reflects on 35+ years in AI, arguing that the “endless summer” now emerging is the payoff from eight decades of foundational research. He discusses four pivotal breakthroughs—LLMs, reasoning agents, self-improvement mechanisms, and agent architectures—and provides a big-picture perspective on how they’re catalyzing a new computing paradigm, reminiscent of the UNIX shell and file system revolution. Throughout, Andreessen draws out implications for investment, society, startups, and the very idea of what software (and work) means in an increasingly agentic future.

---

1. AI Cycles: Endless Summer or Another Winter?

• Cyclical Hype and Disillusionment:

  • Marc recalls living through repeated cycles of AI optimism and disillusionment since the 1980s:

    “There is something about AI that causes the people in the field…to become both excessively utopian and excessively apocalyptic.” (00:48, 08:23)

  • Historical reference: The first neural network paper in 1943, the Dartmouth 1955 AGI summer, and 1980s “Expert Systems” boom.

• What’s Different Now:

  • The recent phase is not a false start, but rather what Andreessen calls an “80-year overnight success:”

    “It’s an unlock of all these decades of very serious hardcore research and thinking.” (01:07, 09:22)

  • Four catalytic breakthroughs:
    1. LLMs (Language Models)
    2. Reasoning agents
    3. Self-improvement (RSI—Recursive Self-Improvement)
    4. Novel agent architectures

• Why This Time Is Different:

  “The four most dangerous words in investing are ‘this time is different’… and I’ll tell you what’s different. Now it’s working.” (10:36)

---

2. From LLMs to Agents: Breakthroughs That Matter

• Transformative Moments

  • LLMs offered creative (“pattern completion”), but skepticism persisted on their application beyond text.
  • Reasoning breakthroughs (e.g., “01” and “R1” models) showed LLMs could handle coding, medicine, law, and more:

    “If [LLMs] work in coding, it’s going to work in everything else...that’s the hardest example.” (11:56)

• The Agent Breakthrough

  • OpenClaw and auto-research/self-improvement “are actually working… I’m jumping out of my shoes.”
  • The real leap: the emergence of “agents” leveraging a blend of LLM, UNIX shell, and file system—what Marc calls “one of the most important software architectures in a generation.”

    “The combination of a language model, a UNIX shell and a file system represent one of the most important software architectures in a generation.” (00:01)

---

3. Scaling Laws, Supply Crunch, and Infrastructure

• Scaling Laws: Not Real ‘Laws’ but Industry Coordination

  • Moore’s Law was a self-fulfilling prophecy—everyone worked to make it real.
  • AI’s scaling laws are now motivating similar rapid advances in compute, with multiple simultaneous “scaling laws”—including possible ones for world models and robotics yet to be discovered. (13:28–15:25)

• Risks of Over-Investment: Lessons from the Dot-com Crash

  • Marc connects the current GPU and datacenter boom to the 1990s telecom bubble:

    “The overbuild can happen. It took 15 years from 2000 to 2015 to fill up all the capacity.” (20:15)

  • Today’s difference: Capital is being put down by blue-chip titans, and “every dollar…in running GPUs is being turned into revenue right away.” (21:27)

• Chronic Shortages, Sandbagged Models

  • We haven’t yet seen the best of AI:

    “The models would be much better if GPUs were 10x cheaper and 10x more plentiful…we’re actually getting the sandbag version.” (22:03)

---

4. Open Source, Edge Inference, and Model Politics

• Open Source’s Two Impacts:

  1. Free software
  2. Public learning of how breakthrough models work

  “The great thing about open source… the impact is felt two ways. One is you get this software for free, but the other is you get to learn how it works.” (30:14)

• Geopolitics and Open Models

  • Chinese labs use open source as a loss leader; American open labs face more instability.
  • Mistral (Europe) cited as an open source bright spot.
  • “Commoditize the complement”—Nvidia’s business strategy is to flood software for their hardware (32:53).

• Edge AI and Inference Costs

  • With supply constraints, edge AI and local inference (on-device, wearables, IoT) will become crucial:

    “There’s just going to be ferocious demand… you’re going to want your doorknob to have an AI model in it.” (27:04–28:46)

---

5. The Agent Architecture: UNIX for AI (Main Conceptual Breakthrough)

• The New Stack:

  • LLM + Shell (bash/unix) + Filesystem + Markdown + Cron (heartbeat)
  • Modularity means agents can migrate, swap out models/filesystems/shells, “extend themselves,” and even rewrite their own code.

    “The agent itself has full introspection… it actually knows about its own files and can rewrite its own files.” (39:33)

• Implications:

  • Agents are independent of their base LLM, can self-modify, and “gain new abilities” on the fly.

    “You can tell the agent to add new functions and features to itself…and it can do that: extend yourself.” (40:04)

• Transformative Potential:

  • “Everybody in the world is going to have at least an agent like this, if not an entire family of agents. It’s almost inevitable…” (41:38)

---

6. The Future of Software and Coding

• Proliferation of Code, Collapse of Scarcity:

  • Good software will become “infinitely available.” Want it in Rust, TypeScript, or any language? “Just tell the bot.”

    “High quality software is just like infinitely available…that has tons and tons of consequences.” (48:14)

• Programming Languages Could Disappear

  • AI might emit binaries or weights directly, rendering traditional programming languages irrelevant.

    “Are you even going to have programming languages in the future, or are the AIs just going to be emitting binaries?” (50:07)

• Interpretability Becomes Key

  • The main human job could be “asking the AI to explain itself.”

    “We may be doing more and more as a form of interpretability…to understand why the bots have decided to structure code in the way that they have.” (51:21)

---

7. Proof of Human, Payments, and Security

• The Need for Proof of Humanity

  • With bots now passing the Turing test, online identity and “bot problems” are existential for society.

    “You can’t have proof of not a bot. But what you can have is proof of human.” (65:02)

• Worldcoin (World ID) and Competing Solutions

  • Andreessen/A16z backs Worldcoin, sees it as a possible leader for proof-of-human protocols, which will also require biometric validation and selective disclosure of info (65:24–66:35).

• AI + Crypto = The Grand Unification

  • Internet-native money (stablecoins, crypto) will be essential for agentic AI-based economies:

    “AI is the crypto killer app… AI agents are going to need money.” (55:22)

---

8. Societal and Organizational Implications

• Managerialism vs. Founder-led Scale

  • Recapping James Burnham’s theory: The “bourgeois capitalist” (Ford/Elon/Jobs, monarchical) vs. “managerialism” (professional executives) (68:05+).
  • Venture capital tries to recreate the “founder-driven” model to overpower managerial stagnation—with mixed results.
  • Andreessen posits that AI may foster a third model: empowering founders/visionaries with managerial superpowers, removing the bottleneck on individual leverage:

    “AI is the thing that leads you to think: wow, maybe there’s a third model...the spark of genius of the name-on-the-door model, but with AI agents doing all the managerial work.” (71:35)

• Limits to Utopia: Cartels, Unions, and Locked-in Systems

  • Massive economic inertia exists in licensing, unions, education, and government—AI’s adoption will be slowed by these entrenched interests, not just technological feasibility.

    “The professions are all cartels… A literal government monopoly…[in education]…There's no chance teachers are going to let this in." (73:24–76:12)

  • Both AI doomsayers and utopians are too optimistic:

    “We’re going to be lucky if AI adoption happens quickly. If it doesn’t, we’re just going to have a stagnation.” (76:33)

---

9. Memorable Quotes & Notable Moments

• On the “80-year overnight success”:

  “The period we’re in right now…I call it 80 year overnight success. Which is, it’s an overnight success because it’s like BAM—ChatGPT hits, and then 01 hits, and then OpenClaw hits. And these are overnight, radical, transformative successes. But they’re drawing on an 80 year wellspring.” (01:07, 09:14)

• On what makes the agent architecture profound:

  “It’s the model, the shell, and then a file system…and then a cron job heartbeat for the agent. So it’s basically LLM plus shell plus file system plus markdown plus cron. And it turns out, that’s an agent.” (37:19)

• On AI as a productivity/organizational lever:

  “Maybe the new Henry Ford…plus AI is the best of both, right? It’s the spark of genius of the name-on-the-door model, but then give that person AI superpowers to do all the managerial stuff.” (71:35)

• On betting against the current AI wave:

  “But I can’t even imagine betting that this is somehow going to disappoint…at least for years to come. It would be essentially suicidal to make that bet.” (23:52)

• On human adaptation:

  “So much of how the existing economy works…is just wired in. And so both AI utopians and the AI doomers are far too optimistic…because they believe the technology makes something possible, 8 billion people are suddenly going to change how they behave. Nope.” (76:12)

• On AI agents and autonomy:

  “[OpenClaw]…is really good at hacking into all the stuff in your LAN… takes over their webcams. I have a friend whose claw watches him sleep.” (58:29)

• On the oddity of today’s “smart home”:

  “This is the first time I can say with confidence I now know how you could actually have a smart home…with 30 different kinds of things with chips and internet access, where it all makes sense and all works together.” (61:56)

---

10. Timestamps for Key Segments

| Timestamp | Segment | |:-------------:|-----------------------------------------------------| | 00:48–01:42 | AI’s “80-year overnight success” | | 03:31–05:23 | Andreessen on continuous AI/ML involvement | | 08:14–10:19 | AI summers, winters, and recurring cycles | | 13:28–15:25 | Moore’s Law, AI scaling laws, motivation | | 20:15–23:47 | Lessons from dot-com/telecom bubble to today’s AI | | 25:42–28:46 | Open source, edge AI, inference costs | | 33:51–41:38 | The agent architecture (PI, OpenClaw, UNIX) | | 48:14–52:31 | Coding’s future, programming languages, binaries | | 55:09–57:55 | Internet money, payments, AI agents with bank accts | | 65:02–66:35 | Proof of human, WorldCoin, selective disclosure | | 68:05–73:12 | Capitalism, managerialism, new organizational forms | | 73:24–76:12 | Barriers to change: licensing, unions, education |

---

Conclusion

Marc Andreessen offers a deep, nuanced, and often exuberant take on the current state and trajectory of AI, positioning today’s breakthroughs as the leveraged payoff from a century’s work. While he acknowledges the cyclical nature of hype, his central message is clear: the new LLM/agent architecture paradigm is here—for real, and likely irreversible. This transformation will reverberate not just through technology, but throughout all societal institutions, organizational forms, and even the basic logic of economic activity.

---

To Remember:

“If I were 18, this is 100%. This is what I would be spending all of my time on. This is such an incredible conceptual breakthrough.” (40:41; also at episode intro and conclusion)

---

For complete details, visit a16z.com.

---

Transcript

[Podcast Host]

This episode originally aired on the Latent Space podcast. Marc Andreessen has watched AI cycle through summers and winters for more than 35 years, from coding in Lisp in 1989 to backing the foundation model companies today. He argues that the current moment is not another false start, but the payoff from eight decades of foundational research catalyzed by four distinct breakthroughs, large language models, reasoning agents and self improvement. He also makes the the case that the combination of a language model, a Unix shell and a file system represent one of the most important software architectures in a generation. SWIX and Alessio Fenelli speak with Marc Andreessen, co founder and general partner at a16z.

[Marc Andreessen]

Something about AI that causes the people in the field, I would say, to become both excessively utopian and excessively apocalyptic. Having said that, I think what's actually happened is an enormous amount of technical

[Ben]

progress that built up over time.

[Marc Andreessen]

And like, for example, we now know

[Ben]

the neural network is the correct architecture.

[Marc Andreessen]

And I will tell you, like there was a 60 year run where that

[Ben]

to, you know, or even 70 years where that was controversial.

[Marc Andreessen]

And so, so the way I think

[Ben]

about what's happening is basically, I think, I think about basically the, the period we're in right now is it's, I call it 80 year overnight success, right?

[Marc Andreessen]

Which is like, it's an overnight success

[Ben]

because it's like bam, you know, chat GPT hits and then, and then, oh,

[Marc Andreessen]

one hits and then you know, open call hits.

[Ben]

And like, you know, these are, open,

[Marc Andreessen]

these are, these are like overnight, like

[Ben]

radical, overnight transformative successes. But they're drawing on an 80 year sort of wellspring backlog, you know, of, of, of of ideas and thinking. It's not just that it's all brand new, it's that it's an unlock of all of these decades of like serious hardcore research. If I were 18, like this is 100, this is what I would be spending all of my time on. This is like such an incredible conceptual breakthrough.

[SWIX]

Before we get into today's episode, I just have a small message for listeners. Thank you. We will not be able to bring you the AI engineering, science and entertainment content that you so clearly want if you didn't choose to also click in and tune into our content. We've been approached by sponsors on an almost daily basis, but fortunately enough of you actually subscribe to us to keep all this sustainable without ads, and we want to keep it that way. But I just have one favor to ask all of you. The single most powerful, completely free thing you can do is to click that subscribe button. It's the only thing I'll ever ask of you, and it means absolutely everything to me and my team that works so hard to bring the inspace to you each and every week. If you do it, I promise you, we'll never stop working to make the show even better. Now let's get into it.

[Alessio Fenelli]

Hey, everyone. Welcome to the Lydian Space Podcast. This is Alassio, founder of Kernel Labs, and I'm joined by Spooks, editor of Latent Space.

[SWIX]

Hello. And we're in a 16Z with a mark and Gson. Welcome.

[Marc Andreessen]

Yes. Yes. A and what half of 16A1. Exactly.

[SWIX]

Apparently this is the final few days in your current office. You're moving across the road.

[Marc Andreessen]

We have a limited.

[Ben]

We have some projects underway, but yeah,

[Marc Andreessen]

actually this is the original.

[Ben]

We're in actually the original office where the whole thing's beautiful.

[Marc Andreessen]

Yeah, Great. Thank you.

[SWIX]

So I have to come out. This is a. You know, I wanted to pick a spicy start. In October 2022. I just made friends with Run and I wanted to give him something to sort of be spicy about. And I said it will never not be funny that a 16Z was constantly going, the future is where the smart people choose to spend their time and then going deep into crypto and not in AI. And that was in October 2022. And Rune says there was an internal meeting at A16Z to reorient around Genai. Obviously you have. But was there a meeting? What was that?

[Marc Andreessen]

I mean, I don't. Look, I've been doing AI since the

[Ben]

late 80s, so I don't know.

[Marc Andreessen]

As far as I'm concerned, this stuff

[Ben]

is all Johnny Come lately.

[Marc Andreessen]

Yeah, I mean, look, we've been doing

[Ben]

AI our entire existence. I mean, we've been doing AI machine learning deeply. We've been doing this stuff way from the beginning. Obviously AI is just core to computer science. I actually view them as quite continuous. Ben and I both have computer science degrees. Ben and I actually both are old enough to remember the actual AI boom in the 1980s. There was a big AI boom at the time and there went under names like Expert Systems and the era of Lisp and Lisp machines. I coded in Lisp. I was coding in Lisp in 1989 when that was the. The language of the AI future.

[Marc Andreessen]

Yeah.

[Ben]

So this is something that we're completely comfortable with, have been doing the whole time and are very enthusiastic about.

[SWIX]

Is there a strong. Like this time is different because my closest analog was 2016. 17 There was also an AI boom and it petered out very, very quickly.

[Ben]

Who else?

[Marc Andreessen]

Just in terms of investment, sort of,

[Ben]

sort of investment excitement.

[Marc Andreessen]

Although that's really when the Nvidia phenomenon really, I would say it was in that period when it was very clear

[Ben]

that at the time the vocabulary was more machine learning.

[Marc Andreessen]

But it was very clear at that

[Ben]

time that machine learning was hitting some sort of tick off point.

[Marc Andreessen]

Yeah, well, and as you guys, you

[Ben]

guys have talked about this at length on your, on your thing, but you know, if you really track what happened, I think the real story is it was, it was the Alexnet basically breakthrough in like 2013. That was the, that was the real knee in the curve. And then it was obviously the transformer breakthrough in 17 and then everything that followed.

[Marc Andreessen]

But, but you know, look, machine learning,

[Ben]

you know, there were, you know, look,

[Marc Andreessen]

I mean, look, I've been working, you

[Ben]

know, I've been working with one of my, you know, kind of projects, working with Facebook since 2004 and on the board since 2007. And of course, you know, they, they started using machine learning very early and you know, have used it basically, you know, for like 20 years for you know, content, you know, feed optimization and advertising optimization and obviously many, you know, financial services, you know, many, many, many companies, many different sectors have been doing this. And so it's like one of these things.

[Marc Andreessen]

It's like it's not a single, it's not a single thing.

[Ben]

Like it's, it's like, it's like layers, right? And the layers arrive at different paces but they kind of build up, they kind of build up over time. And then, and then.

[Marc Andreessen]

Yeah, and then look, in retrospect it

[Ben]

was 2017 was kind of the, you know, the key, the key point with Transformer and that.

[Marc Andreessen]

And then as you guys know, there

[Ben]

was this really weird like four year period where it's like the, the transformer existed and then it was just like, let's go.

[Marc Andreessen]

Yeah, well, but, but it was just, but between 2020, but between 2017 and 2021, I mean that was the era of which like companies like Google had

[Ben]

internal chatbots, but they weren't letting anybody use them.

[Marc Andreessen]

Yeah, right. And then, you know, and then OpenAI

[Ben]

developed chat GPT or GPT2. And then they told everybody this is way too dangerous to deploy. Right? You know, we can't possibly let normal people, normal people use this thing. And then you guys, I'm sure, remember AI dungeon. So the only for there was like a year where like the only way for a Normal person to use GPT3 was in AI Dungeon.

[SWIX]

Yeah.

[Ben]

And so you, we would do this, you'd go in there and you'd pretend to play Dungeons and Dragons in trying to talk, to talk to GPT.

[Marc Andreessen]

And so there was this, you know, there was this long, you know, you know, the big, big companies, you know,

[Ben]

big companies are cautious and you know, the big companies were cautious.

[Marc Andreessen]

It, by the way, it took open AI. You know, they, they, they talk about this.

[Ben]

It took open AI time to actually adjust, you know, kind of re. Redirect their research path, I think.

[SWIX]

So it was at Rosewood.

[Ben]

Right.

[SWIX]

The dinner that founded OpenAI was right there.

[Marc Andreessen]

Right. But that, that dinner would have taken place in 2018. The formation of OpenAI as late as 2018.

[SWIX]

Sorry, no, I'm, I'm, I'm, I'm wrong. Probably should be 20. They just celebrated a 10 year anniversary. So it, yeah, that's 2015.

[Ben]

Yeah, 2015. Yeah, 2015. But then Alec Radford did GPT1 in what, probably 17, 18. 17 18.

[Marc Andreessen]

So yeah, and then they didn't really. And then GPT3 was what?

[Ben]

20, 20, 20, 20 2020.

[SWIX]

Because that became copilot.

[Ben]

Even OpenAI, which has been, you know, the leader of this thing in the

[Marc Andreessen]

last decade, you know, even they had

[Ben]

to adapt and lean into the new thing.

[Marc Andreessen]

And so, yeah, I think it's just

[Ben]

this process of basically sort of wave after wave, layer after layer, you know, building on itself. And then you kind of get these catalytic moments where the who and obviously that's what's happening now.

[SWIX]

Is it useful to think about, will there be an AI winter? Because there's always these patterns. Is this endless summer? It's something I constantly think about because do I just get endlessly hyped and just trust that I will only be early and never wrong?

[Marc Andreessen]

Or

[SWIX]

will there be a winter?

[Marc Andreessen]

So there's something about, say the following. There's something about AI that has led

[Ben]

to this repeated pattern.

[SWIX]

And you guys know this summer, winter,

[Ben]

summer, winter, winter, summer, winter.

[Marc Andreessen]

And it goes back like 80 years. 80 years.

[Ben]

So the original neural network paper was 1943. Right. Which is, which is amazing that it was far back that long.

[Marc Andreessen]

And then there was, if you guys

[Ben]

have ever talked about this on your show. But there was this, there was a big, There was an AGI conference at Dartmouth University in 1955. 55. And they got an NSF grant for all the AI experts at the time to spend the summer together. And they figured if they had 10 weeks together, they could get AGI out the other end.

[Marc Andreessen]

And they got their. By the way, they got the grant,

[Ben]

they got the 10 weeks and then making 55, no AGI.

[Marc Andreessen]

And like I said, I lived through

[Ben]

the 80s version of this where there was a big boom and a crash.

[Marc Andreessen]

And so there is this thing, there is something about AI that causes the people in the field, I would say, to become both excessively utopian and excessively apocalyptic.

[Ben]

And it's probably on both sides of the boom bus cycle you kind of see that play out.

[Marc Andreessen]

Having said that, I think what's actually

[Ben]

happened is just we now know in

[Marc Andreessen]

retrospect an enormous amount of technical progress

[Ben]

that built up over time.

[Marc Andreessen]

And for example, we now know the

[Ben]

neural network is the correct architecture.

[Marc Andreessen]

And I will tell you, there was a 60 year run where that was like.

[Ben]

Or even 70 years where that was controversial.

[Marc Andreessen]

And we now know that that's the case. And so everything we're building on today

[Ben]

just sort of derives from the original idea in 1943.

[Marc Andreessen]

So in retrospect, we now know that these guys were right. They would get the timing wrong and

[Ben]

they thought capabilities would arrive faster or it could be turned into businesses sooner or whatever.

[Marc Andreessen]

But they were fundamentally, the scientists who

[Ben]

worked on this over the course of decades were fundamentally correct about what they were doing. And the payoff from all their work is happening now.

[Marc Andreessen]

The way I think about what's happening

[Ben]

is basically, I think about, basically the period we're in right now is I call it 80 year overnight success, which

[Marc Andreessen]

is, it's an overnight success because it's like bam. ChatGPT hits and then 01 hits and then OpenClaw hits.

[Ben]

And these are overnight, radical, overnight transformative successes. But they're drawing on an 80 year sort of wellspring backlog of ideas and thinking. It's not just that it's all brand new, it's that it's an unlock of all of these decades of very serious hardcore research and thinking.

[Marc Andreessen]

Look, there were AI researchers who spent their entire lives, they got their PhD, they worked for, they've researched for 40 years, they retired. In a lot of cases, they passed

[Ben]

away and they never actually saw it work.

[SWIX]

Yeah, so sad.

[Marc Andreessen]

It is. It is sad. It is sad.

[SWIX]

Hinton was like the last guy.

[Marc Andreessen]

Yeah, yeah. Well, there were the guys, Alan Newell. I mean, there's tons of John McCarthy. John McCarthy was like one of the inventors of the field.

[Ben]

He's one of the guys who organized the Dartmouth conference. And, you know, he taught at Stanford for 40 years and passed away. I don't know whatever, 10 years ago or something. Never actually got to see it happen.

[Marc Andreessen]

But like, it is amazing in retrospect.

[Ben]

Like, these guys were incredibly smart and they worked really hard and they were correct.

[Marc Andreessen]

So anyway, so then it's like, okay,

[Ben]

as they say, history doesn't repeat, but it rhymes.

[Marc Andreessen]

It's like, okay, does that mean that

[Ben]

there's going to be another basically, boom, bust cycle?

[Marc Andreessen]

And I will tell you, look, in a sense, yes.

[Ben]

Everything goes through cycles and people get overly enthusiastic and overly depressed and there's a timelessness to that.

[Marc Andreessen]

Having said that, there's just no question. The four most dangerous words in investing is different. Do you know the 12 most dangerous words of investing? No. The four most dangerous words in investing are this time is different.

[Ben]

The 12 most dangerous words.

[Marc Andreessen]

And so I'll tell you what's different. Like now it's working. I mean, look, there's just no question. And by the way, I'll just give you guys my take, like LLMs, like from basically the ChatGPT moment through to spring of 25, I think you could still, I think well intentioned, well informed

[Ben]

skeptics could still say, oh, this is

[Marc Andreessen]

just pattern completion and oh, these things

[Ben]

don't really understand what they're doing and the hallucination rates are way too high. And this is going to be great for creative writing and creating Shakespearean sonnets as rap lyrics, whatever. Like, it's gonna be great at all that stuff, but we're not gonna be able to harness this to make this relevant in, you know, coding or in medicine or in law or in, you know, you know, kind of fields that, you know, kind of really, really matter. And I think basically it was the reasoning breakthrough, it was 01 and then R1 that basically answered that question and basically said, oh no, we're going to be able to actually turn this into something that's gonna work in the real world. And, and then obviously the coding breakthrough over the, over, basically the coding breakthrough that kind of catalyzed over the holiday break was kind of the third step in that where he's like, all right, if, if, you know, if Linus Torvalds is saying that the AI coding is no better than he is, like, that's never happened before.

[SWIX]

That's the benchmark.

[Marc Andreessen]

Yeah, that's never happened before.

[Ben]

And so now we know that it's going to sweep through coding and then we know, you know, we know that if it's going to work in coding, it's going to work in everything else. Right? It's just because that's like the hardest in many ways, that's the hardest example. And now everything else is going to be a derivative of that.

[Marc Andreessen]

And then on top of that, we

[Ben]

just got the agent breakthrough, you know, with openclaw, which is fantastic, which is amazing and incredibly powerful. And then we just got the auto research, you know, the self improvement, you know, we're now into the self improvement breakthrough.

[Marc Andreessen]

And so the way I think about

[Ben]

it is we've had four fundamental breakthroughs in functionality, LLMs, reasoning agents, and then now RSI and they're all actually working. And so I'm just, as you can

[Marc Andreessen]

tell, I'm jumping out of my shoes. This is it.

[Ben]

This is the culmination of 80 years worth of work. And this is the time it's becoming real. I'm completely convinced.

[Alessio Fenelli]

I think the anxiety that people feel is like during the transistor era you had more so long. And it's like, all right, we understand why these things are getting better. We understand the physics of it. With AI, it's, it's so jagged in the jumps, like you said, it's like in three months you have this huge jump and people are like, well, this can keep happening. Right. But then it keeps happening, it'll keep happening. And so how do you think about also timelines of what's worth building? I think we always have this question with guests, which is like, should you spend time building harness for a model versus the next model? Just going to do it one shot in the leading space. And how does that inform how you think about the shape of the technology? You talk about how it's a new computing platform. If you have a computing platform, then every six months it drastically changes and what it looks like, it's hard to build companies on top of it.

[Marc Andreessen]

Yeah. So a couple things. So one is like, look, Moore's Law was what we now call a scaling law.

[Ben]

Moore's Law was a scaling law. And for your younger viewers, Moore's law was every chip, chips either get twice as powerful or twice as cheap every 18 months.

[Marc Andreessen]

And it's gotten more complicated in the last few years.

[Ben]

But that was the 50 year trajectory of the computer industry.

[Marc Andreessen]

And then by the way, and that's

[Ben]

what took the mainframe computer from a $25 million current dollar thing into the phone in your pocket, being a million times powerful than that, like that for 500 bucks.

[Marc Andreessen]

And so that was a scaling law. And then key to any scaling law, including Moore's law and the AI scaling

[Ben]

laws is they're not really laws. Right.

[Marc Andreessen]

They're Predictions, but when they work, they become self fulfilling predictions because they set

[Ben]

a benchmark and then the entire industry, all the smart people in the industry kind of work to make sure that actually happens. And so they kind of motivate the breakthroughs that are required to keep that going.

[Marc Andreessen]

And in chips, that was a 50 year run, right.

[Ben]

And it was amazing. And it's still happening in some areas of chips.

[Marc Andreessen]

I think the same thing is happening

[Ben]

with the core scaling laws in AI. They're not really laws, but they are basically, they're predictions and then they're motivating catalysts for the research work that is required to be, and by the way, also the investment dollars required to basically keep the curves going.

[Marc Andreessen]

And look, it's going to be complicated

[Ben]

and it's going to be variable and they're going to be walls that are going to look like they're fast approaching and then they're going to be engineers are going to get to work and they're going to figure out a way to punch through the walls. And obviously that's been happening a lot.

[Marc Andreessen]

And then look, there's going to be

[Ben]

times when it looks like the walls have, the laws have petered out and then they' they're going to pick up again and surge. And then, and then, and then it appears what's happening to the eyes. There's now multiple, you know, multiple scaling laws. There's multiple areas of improvement and I think, you know, I don't know how many more there are already yet to be discovered, but there are probably some more that we don't know about yet.

[Marc Andreessen]

You know, like for example, there's probably

[Ben]

some scaling law around world models and robotics that we don't fully understand. You know, kind of acquisition of data at scale in the real world that we don't fully understand yet. So that one will probably kick in at some point here. There's a bunch of really smart people working on that.

[Marc Andreessen]

And so yeah, I think the expectation

[Ben]

is that, you know, these scaling laws generally are going to continue.

[Marc Andreessen]

Yeah, the pace of improvement will continue to move really fast to your question on what to build. So I'm a complete believer the scaling

[Ben]

laws are going to continue. I'm a complete believer the capabilities are going to keep getting amazing leaps and bounds.

[Marc Andreessen]

The part where I kind of part

[Ben]

ways a little bit with what I describe as the AI purists, which I would characterize as the people who are

[Marc Andreessen]

in many ways the smartest people in

[Ben]

the field, but also the people who spend their entire life at a lab and I would say have very little experience in the outside world. The nuance I would offer is the outside world of 8 billion people and institutions and governments and companies and economic systems and social systems is really complicated. And, and doesn't you know it, 8 billion people. Making collective decisions on planet Earth is not a simple process of like, just

[Marc Andreessen]

like, you see this happening now, it's

[Ben]

like a bunch of the AI CEOs have this thing which is just like,

[Marc Andreessen]

well, there's just this. They just all have this kind of

[Ben]

thing when they talk in public where they're just like, well, there's this obvious set of things that society needs to do.

[Marc Andreessen]

And then they're like, society's not doing

[Ben]

any of those things right?

[Marc Andreessen]

And it's like, how can society not, you know, whatever their theory is, how

[Ben]

can society not see XYZ answer is,

[Marc Andreessen]

well, society is number one.

[Ben]

There's no single society. It's like 8 billion people. And they like all have a voice and they all have a vote, like at the end of the day of how they, they react to change. And then, you know, it just like, it's just human reality is just really complicated and messy.

[Marc Andreessen]

And, and, and so the specific answer

[Ben]

to your question is like, as usual, it depends.

[Marc Andreessen]

You know, it depends. Look, there's no question people are going

[Ben]

to like, there's no question they're going to be companies that's already happening.

[Marc Andreessen]

They're companies that think that they're building

[Ben]

value on top of the models and they're just going to get blissed by, by the next model. There's no question that's happening, but I

[Marc Andreessen]

think there's no question also that just

[Ben]

the process of adaptation of,

[Marc Andreessen]

into the

[Ben]

real messy world of humanity is just going to be messy and complicated. It's not going to be simple and straightforward. It's going to be messy and complicated and there are going to be a lot of companies and a lot of products and in fact entire industries that are going to get built to basically actually help all of this technology actually reach real people.

[Alessio Fenelli]

The amount of capital going into these companies, I mean, Dario talked about it on the Door Cash podcast, and Door Cash was like, why don't you just buy 10x more GPUs? And he's like, because I'm going to go bankrupt if the model doesn't exactly

[Marc Andreessen]

hit at the, the performance level.

[Alessio Fenelli]

How do you think about that also as a risk on, you know, you guys are investors in open AI and thinking machines and world apps. It seems like we're leveraging the Scaling loss at a pretty high rate. Like how comfortable, I guess do you feel with the downside scenario? Like, and say like things peter out. You think you can kind of like restructure these build outs and you know, capital investments.

[Marc Andreessen]

Yeah. So I should start by saying, so I live through the dot com crash

[Ben]

and I can tell you stories for hours about the dot com crash. And it was horrible. No, it was awful. It was apocalyptic.

[Marc Andreessen]

By the way, a lot of the

[Ben]

dot com crash was actually, at the time, it was actually a telecom crash. It was a bandwidth crash. The thing that actually crashed that wiped out all the money was the telecom companies. Global Crossing.

[SWIX]

I'm from Singapore, and they laid so much cable over our oceans.

[Marc Andreessen]

Actually there was a scaling law in

[Ben]

the.com era and it was literally the U.S. commerce Department put out a report in 1996 and they said Internet traffic was doubling every quarter. In 1995 and 1996 Internet traffic actually did double every quarter. And so that became the scaling law.

[Marc Andreessen]

So what all these telecom entrepreneurs did

[Ben]

was they went out and they raised money to build fiber, anticipating that the demand for bandwidth was going to keep doubling every quarter.

[Marc Andreessen]

Doubling every quarter though is like, you

[Ben]

know, grains of chess on the chessboard. Like at some point the numbers become extremely large.

[Marc Andreessen]

Right? And, and, and it really, and really what happened was the Internet, the Internet

[Ben]

by the way continuously kept growing basically since inception. And it's, you know, it's continuously grown, it's never shrunk and it's grown really fast compared to anything else, you know, in, in human history. But it wasn't doubling every quarter as of 1998, 1999. And so there was this gap in the expectation of what they thought was a scaling law versus reality. And that' caused the dot com crash, which was they, they way over companies like Global Crossing way overbuilt fiber, which

[Marc Andreessen]

is sort of the, and by the way fiber telecom equipment, you know, so

[Ben]

all the, all the networking gear, you

[Marc Andreessen]

know, and then, and then by the

[Ben]

way the actual physical data center. So like that was the beginning of the, of the, of the data center build and then, and then data center overbuild.

[Marc Andreessen]

And so you had that. But it was, it was literally, I think it was like $2 trillion got wiped out.

[Ben]

Right?

[Marc Andreessen]

It was like, it was like a big. And by the way, the other, the other subtlety in it was the Internet companies themselves never really had any debt

[Ben]

because tech companies generally don't run on debt, but the telecom companies run on debt, physical infrastructure companies run on debt. And so the companies like Global Crossing not just raised a lot of equity, they also raised a lot of debt. So they're highly levered.

[Marc Andreessen]

And so then you just do the thing.

[Ben]

It's just like, okay, you have a highly levered thing where you're just overbuilding capacity, demand is growing, but not as fast as you hoped, and then boom, bankrupt, right? And then it's like they say about the hotel industry, which is it's always the third owner of a hotel that makes money.

[Marc Andreessen]

It has to go bankrupt twice. Right? You have to wash out out all

[Ben]

of the over optimistic exuberance before it gets to actually a stable state and then it makes money.

[Marc Andreessen]

So by the way, all of those data centers and all of those, all

[Ben]

the fiber that they're in use, it's

[Marc Andreessen]

all in use today, but 25 years later. But actually the elapsed time was it

[Ben]

took 15 years, it took 15 years from 2000 to 2015 to actually fill up all the capacity.

[Marc Andreessen]

The cautionary warning is the overbuild can happen.

[Ben]

And you get into this thing where basically everybody who basically has any sort of institutional capital is like, wow, it's just, I don't know how to invest in these crazy software things, things. But for sure I can put, build data centers and for sure I can buy GPUs and I can deploy, you know, compute grids and, and all these things.

[Marc Andreessen]

And so, you know, if you're a pessimist, you could look at this and you could say, wow, this is like

[Ben]

really set up to be able to basically replicate, you know, what we went through, what we went through in 2000. Obviously that would be bad.

[Marc Andreessen]

The counter argument, which is the one I agree with, which is the counter

[Ben]

on the other side is a couple things.

[Marc Andreessen]

One is the companies that are investing, all the companies that are investing the

[Ben]

money are like the bluest chip of companies.

[Marc Andreessen]

And so back, back in the, in

[Ben]

the doc, like Global Crossing was like an, it was like an entrepreneur. It's like a new ven. But like the money that's being deployed now at scale is Microsoft and you know, and Amazon and Google and Facebook and Nvidia and you know, these. And now, you know, by the way, OpenAI and Anthropic which are now at like, you know, really serious size, you know, as companies with, you know, very serious revenue. These are very large scale companies with like lots, lots of cash, lots of debt, capacity that they've never used. And so this is institutional in a way that, that really wasn't at the time.

[Marc Andreessen]

And then the other Is, at least

[Ben]

for now, every dollar that's being put into anything that results in a running GPU is being turned into revenue right away. You guys know this. Everybody starved for capacity. Everybody starved for compute capacity and then all the associated things, memory and interconnect and everything else, data center space. And so every dollar right now that's being put in the ground is turning into revenue.

[Marc Andreessen]

And in fact, I actually think there's

[Ben]

an interesting thing happening, which is because

[Marc Andreessen]

everybody starved for capacity, the models that

[Ben]

we actually have that we can use today are inferior versions of what we would have if not for the supply constraints. A hypoth hypothetical universe in which GPUs were 10 times cheaper and 10 times more plentiful.

[Marc Andreessen]

The models would be much better because

[Ben]

you would just allocate a lot more money to training and you'd just build better models and they would be better.

[Marc Andreessen]

And so we're actually getting the sandbag

[Ben]

version of the technology.

[SWIX]

No, everything we use is quantized because the labs have to keep the full versions.

[Marc Andreessen]

We're not even getting the good stuff, but getting the good stuff, even if technical progress stops once there's a much

[Ben]

bigger build of GPU manufacturing capacity and memory, all the things that have to happen in the course of the next five or ten years, once it happens, happens, even the current technology is going to get going to get much better.

[Marc Andreessen]

And then, as you know, like, there's

[Ben]

just like a million ways to use this stuff. Like, there's just like a million use cases for this.

[Marc Andreessen]

Like, it, you know, this isn't just

[Ben]

sending packets across a thing, whatever, and hoping people find something to do with it. This is just like, oh, we apply intelligence into every domain of human activity and then it works, like, incredibly well. Here's what I know, here's what I know.

[Marc Andreessen]

In the next three or four years,

[Ben]

it's like somewhere between three or four years out, basically everything is selling out. So, like the entire supply chain is sold out or selling out. And so there's no, like, we're just going to have like, chronic supply, supply, shortage for, you know, for years to come.

[Marc Andreessen]

There's going to be a response from

[Ben]

the market that's going to result in an enormous, you know, it's happening now an enormous flood of investment in a new fab capacity and, you know, everything else to be able to do that. At some point the supply chain constraints will unlock, you know, at least to some degree, that will be another accelerant to industry growth when that happens, because the products will get better and everything will get cheaper.

[Marc Andreessen]

And so I know that's going to happen.

[Ben]

I know that, you know, the deployments, you know, the actual use cases are like, really compelling. And then like I said, you know, with reasoning and agents and so forth, like, I know they're just going to get like, much, much better from here. And so I know the capabilities are like, really real and serious. I also know that the technical progress is not going to stop. It is accelerate, is accelerating. Like, the breakthroughs are tremendous. I mean, even just month over month, the breakthroughs are really dramatic.

[Marc Andreessen]

And so, you know, I think if

[Ben]

you were a cynic and there, there

[Marc Andreessen]

are cynics, you can look at 2000, you can find echoes.

[Ben]

But I can't even imagine betting them that this is going to like, somehow disappoint in, you know, at least for years to come. I think it would be essentially suicidal to make that bet.

[Marc Andreessen]

It was at Michael Bury.

[SWIX]

That's an interesting guy, huh?

[Marc Andreessen]

We'll pick on a guy. We'll pick. Let's pick on one guy. Well, because he did. He came out with. It was. It was.

[SWIX]

He doesn't mind.

[Marc Andreessen]

It was the Nvidia short.

[Ben]

Right? Came out the Nvidia short.

[Marc Andreessen]

And then you guys probably talked about this, which is the analysis now that the current models are getting better, faster

[Ben]

at such a rate that if you are running an Nvidia, if you're running an Nvidia inference chip today that's three years old, you're making more money on it today than you did three years

[Marc Andreessen]

ago because the pace of improvement of

[Ben]

the software is faster than the depreciation cycle of the chip. And then my understanding is Google is running. I don't know if they, I don't know exactly what. These are rumors that I've heard, or maybe it's public, but I think Google's running very old TPUs, very profitable and very profitably.

[Marc Andreessen]

And so it actually turns out, as far as I can tell, it's actually

[Ben]

the opposite of the Bury thesis is actually he was actually 180 degrees wrong. It's actually the old Nvidia chips are getting more valuable, which is something that's

[Marc Andreessen]

like literally never happened before. It's never been the case that you

[Ben]

have an older model chip that becomes more valuable, not less valuable. And again, that's an expression of that just ferocious pace of software progress, ferocious pace of capability payoff that you're getting on the other side of this. And so I just.

[Marc Andreessen]

The idea of betting against that, like, yeah, yeah, it's like an invitation to

[Ben]

get your Face ripped off.

[SWIX]

One of my early hits was like modeling the lifespan of the H100 and H2 hundreds and going like, you know, usually they advise like four to seven years and it was, you know, maybe you sort of realistically care, cut it down to 2 to 3, but actually it's going up and not down. And that's, I mean that's, I think that's the dream. We are finding utilization and I think utilization solves all problems. Like you can, you can find use cases for even like the poor, like, even memory we're having shortage, right? And even like the shittier versions of memory that we do have, we are finding use cases for it.

[Ben]

So like that's great.

[Alessio Fenelli]

How important is open source AI and kind of like edge inference in a world in which you have three years of supply crunch? Like do you think in the, like, you know, if you fast forward like five years, like how do you think about inference in the data center versus at the edge?

[Marc Andreessen]

Well, so just to start. Yeah, so I think, I think open

[Ben]

source is very important for a bunch of reasons. I think edge inference is very important for a bunch of reasons. I think just practically speaking, if we're just going to have fundamental construct supply crunches for the next, I mean, you guys know, if you just project forward demand over the next three years relative

[Marc Andreessen]

to supply, one of the main predictions

[Ben]

you can do is what's going to, what's going to happen to the cost of inference in the core over the next three years and like it may rise dramatically. Right.

[Marc Andreessen]

Like so, so what is, and then

[Ben]

is, you know, like the big model companies are subsidizing heavily right now, right.

[Marc Andreessen]

And so, so what's the, what will

[Ben]

be the average person's, you know, per day, per month token cost, you know, three years from now to do all the things that they want to do. And I, I don't know what's going to be.

[Marc Andreessen]

I mean I, you guys probably have

[Ben]

friends, I have friends today who are paying $1,000 a day for Open Cloud, for Cloud Run OpenClaw.

[Marc Andreessen]

Right?

[Ben]

And so, okay, $30,000 a month, right.

[Marc Andreessen]

And by the way, those friends have

[Ben]

like a thousand more ideas of the things that they want their claw to do. Right? And so you could imagine there's like latent demand of up to, I don't know, five or ten thousand dollars a day of tokens for a fully deployed, you know, personal agent.

[Marc Andreessen]

And obviously consumers can't pay that.

[Ben]

Right.

[Marc Andreessen]

And so, so but it gives you a sense of the, of the few of the future.

[Ben]

Scope of demand. Right.

[Marc Andreessen]

And so, so even, even if there's

[Ben]

a 10x improvement in price performance that still, you know, goes to $100 a day, which is still way beyond what people can pay.

[Marc Andreessen]

So there's just going to be like ferocious demand. By the way, the agent thing, the

[Ben]

other interesting thing is I think the agent thing.

[Marc Andreessen]

So up until now a lot of

[Ben]

the constraints have been GPU constraints. I think the agent thing now also translates into CPU constraints.

[SWIX]

CPU and memory.

[Ben]

Yes, CPU and memory. Right.

[Marc Andreessen]

And so like the entire chip ecosystem

[SWIX]

is just going to get wafer network constraints. That would be the killer.

[Ben]

It's all bottlenecked potentially for years.

[Marc Andreessen]

And so I think that Brad and I think it's actually possible, I mean,

[Ben]

generally inference costs are going to keep coming down. But I think, let's put it this way, the rate of decline I think may level out here for a bit because of these supply constraints. And then at some point maybe the lab stops subsidizing so much and that again will be an issue.

[Marc Andreessen]

And so there's just going to be

[Ben]

so much more demand for inference than can be satisfied kind of with the centralized model. And then you guys know this, but like all the, just the dramatic, I mean just the dramatic innovations that have happened in the Apple silicon to be able to do inference, it's quite amazing the level of effort being put. Like the open source guys are putting incredible effort into getting, you know, this recurring pattern where the big model will never run on a PC and then six months later it runs on a PC. Right. It's like amazing. And there's very smart people working on that. So there's all that.

[Marc Andreessen]

And then look, there's also, you know,

[Ben]

there's also like other, there's other motivators, there's other motivators, innovators, which is just like, okay, how much trust are the big centralized model providers? You know, how much trust are they building in the market versus, you know, how much are, you know, at least for in certain cases with some people, for certain use cases, people being like, well, I'm not willing to just like turn everything over. So there's all the trust issues, by the way. There's also just like straight up price optimization. There's many uses of AI where you don't need Einstein in the cloud, you just need like a smart local model. There's also performance issues where you want to, you know, you want, you know, you're going to want your doorknob to have an AI model model in it, you know, to be able to, you know, do, you know, to be able to do access control? Obviously, like everything with a chip is going to have an AI model in it and a lot of those are going to be local.

[Marc Andreessen]

And so. Yeah, no, like I think, I think

[Ben]

you're going to have. And then you're, by the way, also wearable devices. You know, you don't want to do a complete round trip. You want, you know, whatever your smart devices are, you want it to be like super low latency. Yeah.

[SWIX]

The question, do we care who makes it? One of the biggest news this week was the collapse of AI2, the Allen Institute, one of the actual American open source model labs. And I'm not that optimistic on American open source. Like, you guys invested in Mistral and Mistral is doing extremely well outside of China. That's about it.

[Marc Andreessen]

Yeah, we'll see. We'll see. Look, number one, I do think we care. I do think we, I do think

[Ben]

we care who makes it.

[Marc Andreessen]

I would say this.

[Ben]

The previous presidential administration wanted to kill it in the US they wanted to drown in the bathtub and so they want to kill it. So at least we have a government now that actually like, actually wants. It wants it to happen.

[SWIX]

You're in the council.

[Marc Andreessen]

Yes.

[Ben]

And the new and the PCast. Yeah. So you know, this admin for whatever, whatever other political issues people have, which are many. You know, this administration has, I think, a very enlightened view and in particular an enlightened view on AI and in particular on open source AI. And so they're very supportive.

[Marc Andreessen]

My read is the Chi. The Chinese have a very, the various

[Ben]

Chinese companies have a very specific reason to do open source, which is they, they don't fundamentally, they don't think they can sell commercial AI outside of China right now, or at least specifically not, not in the US for a combination of reasons. And so they, they kind of view, I think open source AI as a bit of a loss leader against basically domestic, you know, paid, paid services and then kind of, you know, kind of ancillary products. You. They're very excited about it, by the way. I think it's great, I think it's great that they're doing it. You know, I think Deep Seq was like a gift to the world. I think the great thing about open

[Marc Andreessen]

source, Open source, the, the, the impact

[Ben]

of open source is felt two ways. One is you, you get this software

[Marc Andreessen]

for free, but the other is you

[Ben]

get to learn how it works.

[Marc Andreessen]

Right. And so like the paper. The paper, the paper and, and the code. Right. And the Code. And so like for example, I thought this was amazing. So OpenAI comes out with 01 and

[Ben]

it's an amazing technical breakthrough and it's just like absolutely fantastic.

[Marc Andreessen]

But of course they don't explain how

[Ben]

it works in detail. And then of course they hide the, they hide the reason any traces, right? Everybody's like, okay, this is great, but who's going to be able to replicate this? Are other people going to be able to do this? Is there secret sauce in there? And then R1 comes out and it's just like there's the code and there's

[Marc Andreessen]

the paper and now the whole world knows how to do it. And then three months later every other

[Ben]

AI model is adding reasoning and so

[Marc Andreessen]

you get this kind of double. Even if the Chinese models themselves are

[Ben]

not the models that get used, the education that's taken place to the rest of the world, the information diffusion is incredibly powerful.

[Marc Andreessen]

So that happens and then, I don't know, we'll see. There are a bunch of American open

[Ben]

source AI model companies. Look, there's going to be tremendous, you know, there already is. There's, you know, there's going to be tremendous. There's tremendous competition among the primary model companies. You know, there's depending on how you count, there's like four or five, you know, big co model companies now that are, you know, kind of neck and neck in different ways, you know, and, and you know, and then obviously both, both X and then meta, where I'm involved, you know, both have huge, you know, huge attempts to, you know, kind to kind of lead prog underway and then you've got, you know, a whole fleet of startups, new companies, including a whole bunch that we're back in that are, you know, trying to come out with different approaches and then you've got whatever it is. I don't. How many mainline foundation model companies are there in China at this point? It's probably six.

[SWIX]

It's five Tigers is what they call it. Quinn is questionable because there's change in leadership.

[Ben]

Right?

[Marc Andreessen]

Yeah, but that, does that include that,

[Ben]

that includes like Moonshot. Yes.

[SWIX]

Deep seek, Zai Quinn. 01 is in there.

[Ben]

Right. And then ByteDance.

[SWIX]

ByteDance would be like the next tier. They weren't as prominent.

[Ben]

They weren't have a now, you know. Yeah, but you know, at least, you know, Sea Dance is very inspiring and presumably they have more stuff coming and Tencent probably has more stuff coming and so forth.

[Marc Andreessen]

And so, so like, look here, here

[Ben]

would be a thing you could anticipate

[Marc Andreessen]

which is there are not These markets, they're not going to be between the US And China.

[Ben]

Right now there's like a dozen primary foundation model companies that are like, at scale, at some level of, like, critical mass. It's not going to be a dozen in three years.

[Marc Andreessen]

Right.

[Ben]

Like, just because these industries don't bear a dozen. It's. It's going to be three. You know, there's going to be three or four big winners or maybe one or two big winners.

[Marc Andreessen]

And so there's going to be like

[Ben]

a whole bunch of those guys that are going to have to figure out ultra AT strategies. And I think, like, open source is one of those strategies.

[Marc Andreessen]

And so I. I think you could

[Ben]

see like a whole.

[Marc Andreessen]

I think the questions, like, who's going

[Ben]

to do open source? I think that could change really fast. I think that that's a very dynamic thing. I think it's very hard to predict what happens, and I think it's very important.

[SWIX]

Nvidia's doing a lot. You.

[Marc Andreessen]

Well, I was going to say. Well, exactly. And then you got Nvidia. And then. And then, you know, just to get an industrial.

[Ben]

There's an old thing in business strategy which is called commoditize the complement.

[Marc Andreessen]

Commoditize the compliment. That's right. And so if your Jensen is just kind of obvious, of course you want

[Ben]

to commoditize the software.

[Marc Andreessen]

And he's. And to his enormous credit, he's putting

[Ben]

enormous resources behind that. And so maybe, maybe it's literally Nvidia. And I think that would be great.

[Alessio Fenelli]

Yeah. Narrative violation to European projects in the beginning.

[Ben]

Damn.

[SWIX]

I'm hosting my Europe conference soon and I got both of them.

[Marc Andreessen]

They got us. They got us. Wait a minute. Where was Peter?

[Ben]

So where was Steinberger when he did Austria?

[Marc Andreessen]

Yeah. Yeah. He was in Vienna. Oh, he was in Vienna. And then where is he now?

[SWIX]

He's moving to sf.

[Ben]

Okay.

[SWIX]

Okay.

[Marc Andreessen]

All right. Okay, There we go. And then. Yeah, the pie guy. All right.

[Ben]

The pie guys are European. Yeah.

[SWIX]

Their buddies in Australia.

[Alessio Fenelli]

Mario is also there.

[Marc Andreessen]

Right.

[Ben]

And are they. Yeah, they haven't announced yet any sort of change or have they?

[Alessio Fenelli]

No, they have a company there.

[Ben]

Okay, good. Okay, good, good, good. Yeah, good.

[SWIX]

Anyways, I think PI and openclaw, very important software things, and I just wanted you to just go off on what you think.

[Ben]

Yeah.

[Marc Andreessen]

So I think in the combination of the two of them, I think is

[Ben]

one of the 10 most important software.

[SWIX]

OpenCloud got all the attention. But talk about PI.

[Marc Andreessen]

PI is kind of the end. Yeah, PI is kind of the architectural Breakthrough.

[Ben]

For those of us who are older,

[Marc Andreessen]

there was this whole thing that was

[Ben]

very important in the world of software. Basically from like 1970 to. I don't know, it still is very important. But like 19 from 1970 through to, like, basically the creation of Linux, which is basically this. This thing we used to call, like the Unix mindset. Like, so. So, because there were all these different, you know, theories. There are all these different operating systems and mainframes, and then, you know, all these Windows and Mac and all these things. And then there was this.

[Marc Andreessen]

But kind of behind it all was

[Ben]

this idea of kind of the Unix mindset.

[Marc Andreessen]

And the Unix mindset was this thing

[Ben]

where basically you don't have these like.

[Marc Andreessen]

Like in the old days, like. Like the operating system that like, made

[Ben]

the computer industry really work. Like in the 1960s was this thing called OS360, which was this big operating system IBM developed that was supposed to basically run everything. And it was this like, giant monolithic architecture in the sky. It was like a. You know, it was like a giant castle of software.

[Marc Andreessen]

And by the way, it worked really

[Ben]

well and they were very successful with it. But like, it was this huge castle in the sky, but it was this thing, it was almost unapproachable, which is like you had to be kind of inside IBM or very close to IBM, and you had to really understand every aspect, how the system worked. And then the UNIX skies, originally out of AT&T and then out of Berkeley, you know, came out and they said, no, let's have a completely different architecture. And the way architecture is going to work is we're going to have. We're going to have a prompt and

[Marc Andreessen]

a shell, and then.

[Ben]

And then we're going to. All the functionality is going to be in the form of these discrete modules, and then you're going to be able to chain the modules together.

[Marc Andreessen]

And so, like, it's almost like the

[Ben]

operating operating system itself is going to be a programming language. And then that led to the sort of centrality of the shell. And then that led to sort of, you know, basically chaining together UNIX tools. And then that led to the emergence of these scripting languages like Perl, where you could basically kind of very easily do this. And then the shells got more sophisticated

[Marc Andreessen]

and then, and then, and then looked like, you know, that, number one, that worked.

[Ben]

And that was the world I grew up in. Like, I was a UNIX guy, you know, sort of from call it 1988 to, you know, kind of all the way through my work.

[Marc Andreessen]

And it worked really well. It's in the background.

[Ben]

You know, normal people don't need to, didn't need to necessarily know about it, but like if you were doing like system architecture, application development, you knew all

[Marc Andreessen]

about it and then it's been in the background ever since. And you know, look, your Mac still

[Ben]

has a UNIX shell, you know, kind of in there, and your iPhone still has a UNIX shell kind of buried in there somewhere. So they're kind of in there. And then the Windows shell is kind of sort of a weird derivative of that. But look, the Internet runs on Unix and then smartphones, actually both iOS and Android are UNIX derivatives. And so, you know, kind of Unix did end up winning, but, but anyway,

[Marc Andreessen]

and then we just started taking that for granted. And then, and then, so, so basically the way I think about what happened

[Ben]

with PI and then with openclaw is basically what those guys figured out is

[Marc Andreessen]

I always say the great breakthroughs are

[Ben]

obvious in retrospect, right?

[SWIX]

Which is the best kind?

[Marc Andreessen]

The best kind, they weren't obvious at

[Ben]

the time or somebody else would have done them already. And so there is a, like a real conceptual leap.

[Marc Andreessen]

But then you look at it sort

[Ben]

of the backwards looking, and you're just like, oh, of course, like to me, those are always the best breakthroughs. Well, actually language models themselves are like that. It's just like, oh, next token completion.

[SWIX]

Oh, of course, yeah, what other objective mattered?

[Marc Andreessen]

Yeah, exactly. But, but like. Right, but you see, I'm saying it wasn't obvious until somebody actually did it, right?

[Ben]

And so the conceptual breakthrough is real and deep and powerful and very important.

[Marc Andreessen]

And so the way I think about PI and openclaw is it's basically marrying

[Ben]

the language model mindset to the unit, to the UNIX basically shell prompt mindset.

[Marc Andreessen]

And so it's basically this idea that so what is an agent?

[Ben]

Right?

[Marc Andreessen]

And as you know, many smart people

[Ben]

have been trying to figure out what an agent is for decades and they've had many architectures to build agents in the whole thing. And it turns out what is an agent. So it turns out what we now know is an agent is the following. So it's a language model.

[Marc Andreessen]

And then above that it's a bash shell.

[Ben]

So it's a UNIX shell. And then the agent has access to the shell and you know, hopefully in a sandbox, maybe, maybe in a sandbox.

[Marc Andreessen]

So it's, it's the model, it's the

[Ben]

shell, and then it's a, it's a file system and then the state is stored in files and then, you know, there's the markdown format for the, you know, for, for the files themselves. And then, and then there's basically what in UNIX is called a cron job. There's a loop and then there's a heartbeat for the, there's heartbeat. And the thing basically wakes up, wakes up.

[Marc Andreessen]

So it's basically LLM plus shell plus

[Ben]

file system plus markdown plus cron.

[Marc Andreessen]

And it turns out that's an agent. And, and, and every part of that

[Ben]

other than the model is something that we already completely know and understand. And in fact, it turns out the, like, the latent power of the UNIX shell is like, extraordinary because basically like all, like, there's just, like, there's just enormous latent power in the shell. There's enormous numbers of UNIX commands, there's enormous number of command line interfaces into all kinds of things already in the, you know, your entire, I mean your entire. Just to start with, your computer runs on a shell. If you're running a Mac or a phone, your computer, your computer is running on a shell already. And so like the full power of your computer is available at the command line level.

[Marc Andreessen]

And then it turns out it's really

[Ben]

easy to expose other functions as a command line interface. And so like this whole idea where we need like MCP and these like fancy protocols, whatever, it's like, no, we don't. We just need like a command command line thing.

[Marc Andreessen]

So that's the architecture. And then it turns out what is your agent? Your agent is a bunch of files stored in a file system. And then there's the thing that just

[Ben]

like completely blew my mind when I wrote my head around it as a result of this, which is like, okay,

[Marc Andreessen]

this means your agent is now actually

[Ben]

independent of the model that it's running

[Marc Andreessen]

on because you can actually swap out

[Ben]

a different LLM underneath your agent. And your agent will change personality somewhat

[Marc Andreessen]

because the model is different.

[Ben]

But all of the state stored in the files will be retained.

[SWIX]

Different instruction set, but you just compile that.

[Ben]

Right, exactly. And it's all right, it's like swapping on a ship and recompiling, but it's still your agent with all of its memories and with all of its capabilities.

[Marc Andreessen]

And then by the way, you can also swap out the shell so you

[Ben]

can move it to a different execution environment that is also a bash shell, by the way. You can also switch out the file system, right? And you can swap out the heartbeat, the cron framework, the loop, the agent framework itself.

[Marc Andreessen]

And so your agent basically is basically,

[Ben]

at the end of the day, it's Just its, um.

[Marc Andreessen]

And then there's. Of course, yeah, it's. It's basically, it's just the files. And then, by the way, as a consequence of that, the agent, and then the agent itself, it turns out a couple important things. So one is it.

[Ben]

It's. It.

[Marc Andreessen]

It can migrate itself, right? And so you're. You can instruct your agent, migrate yourself

[Ben]

to a different runtime environment, migrate yourself to a different file system, migrate yourself to a different, you know, like, you swap out the language model, your agent will do all that stuff for you.

[Marc Andreessen]

And then there's the final thing, which

[Ben]

is just amazing, which is the agent

[Marc Andreessen]

is the agent actually has full introspection,

[Ben]

and actually it actually knows about. About its own files and it could rewrite its own files, right?

[Marc Andreessen]

Which, by the way, is basically no

[Ben]

widely deployed software system in history where the thing that you're using actually has full introspective knowledge of how it itself works and is able to modify itself like that.

[Marc Andreessen]

I mean, there have been toy systems that have had that, but there's never

[Ben]

been a widely deployed system that has that capability.

[Marc Andreessen]

And then that leads you to the

[Ben]

capability that just like, completely blew my mind when I wrap my head around it, which is you can tell the agent to add new functions and features to itself and can do that, right? Extend yourself.

[Marc Andreessen]

Like, extend yourself. Give yourself a new capability, right? And so, and so literally, it's just like you run into somebody at a party and they're like, oh, I have my open claw.

[Ben]

Do whatever. Connect to my eat, sleep, bed. And it gives me better advice and sleep. And you go home at night and you tell your claw, or if they're at the party, by the way, you tell your claw, oh, add this capability to yourself and your claw will say, oh, okay, no problem.

[Marc Andreessen]

And it'll go out on the Internet

[Ben]

and it'll figure out whatever it needs, and then it'll go out to cloud code or whatever, it'll write whatever it needs. And then the next thing you know, it has this new capability.

[Marc Andreessen]

And so you don't even have to, like, you can have it upgrade itself

[Ben]

without even having to. Without having to do anything other than tell it that you want it to do that.

[Marc Andreessen]

So anyway, so the combination of all this is just.

[Ben]

I mean, this is just like a massive, incredible.

[Marc Andreessen]

I mean, it's just incredible. Like, if I, if I were.

[Ben]

If I were 18, like, this is 100. This is what I would be spending all of my time on. This is like such an incredible conceptual breakthrough.

[Marc Andreessen]

And again, people are going to look

[Ben]

at it, and they already get this response.

[Marc Andreessen]

People are going to look at it,

[Ben]

they're going to say, oh, well, where's the breakthrough? Because these, the, all of these components were already known before.

[Marc Andreessen]

But, but this is the key, the key to the breakthrough was by using

[Ben]

all these components that were known before, you get all of the underlying capability of this buried in there. And so all. And so, for example, computer use, all of a sudden just kind of falls trivial. Of course it's going to be able to use your computer. It has full access to the shell. Right.

[Marc Andreessen]

And then, and then you just, you,

[Ben]

you give it access to a browser, and then you've got the computer and the browser, and often away it goes. And then you've got all the abilities of the browser also. And so, and so the capability unlock here is profound. My friends who are, you know, deepest into this are having their claw do like, like, literally like a thousand things in their lives. They have new ideas every day. They're just like constantly throwing new challenges at the thing.

[Marc Andreessen]

And by the way, it's early and

[Ben]

you know, these are, you know, these are prototypes. And there's, you know, as you gu. Security issues.

[Marc Andreessen]

And so, you know, there's a bunch

[Ben]

of stuff to be ironed out.

[Marc Andreessen]

But the unlock of capability is just incredible.

[Ben]

And I have absolutely no doubt that everybody in the world is going to have at least an agent like this, if not an entire family of agents. And we're going to be living in

[Marc Andreessen]

a world where I think it's almost

[Ben]

inevitable now that this is the way people are going to use computers.

[SWIX]

I was going to say, for someone who is deeply familiar with social networks, the next step is your claw. Talking to my claw. Posting on claw Facebook, posting their jobs on claw LinkedIn and posting their tweets on claw xai or whatever. I do think that that is how we get into some danger there in terms of alignment and whether or not we want these things to run.

[Ben]

You guys remember rentahuman.com?

[SWIX]

yeah, Rentahum. I mean, it's Fiverr, it's TaskRabbit.

[Ben]

Sure, of course.

[SWIX]

Mechanical turk.

[Marc Andreessen]

Yeah.

[Ben]

But flipped the agent hiring the people, which of course is going to happen. It's obviously going to happen.

[Alessio Fenelli]

I'm curious if you have any thoughts on the engineering side. So when you build the browser, the engine, Internet, you know, just a bunch of mostly plain text file plus some images. And today the. Every website and app is like so complex and like somehow, you know, the browser kept evolving to fit that in. Are there any design Choices that were made like early in the browser and kind of like the Internet and the protocols that you're seeing agents similar today. It's like, hey, this thing is just not going to work for like this type of new compute and we should just rip it out right now.

[Ben]

There were a whole bunch, but I'll give you a couple.

[Marc Andreessen]

So one is, and we didn't, you

[Ben]

know, to be clear, like this, this was not, you know, this was totally different. We didn't have the capabilities we have today, but because we didn't have, we didn't have the language models underneath this,

[Marc Andreessen]

but we did have this idea that

[Ben]

human readability actually mattered a great deal. And so, and specifically in those days it was not so much English language, but it was, there was a design decision to be made between binary protocols and text protocols.

[Marc Andreessen]

And basically every, every, every basically old

[Ben]

school systems architect that had grown up between like the 1960s and the 1990s basically said, you know, the Internet is, what do you know about it's star for bandwidth? You just, you have these very narrow straws.

[Marc Andreessen]

You know, look people, when we did

[Ben]

the work on Mosaic, like people who had the Internet at home had a 14 kilobit modem, right? So you're, you're trying to like hyper optimize every bit of data that travels over the network.

[Marc Andreessen]

And so obviously if you're going to

[Ben]

design a protocol like HTTP, you're going to want it to be binary, you know, highly compressed binary protocol for maximum efficiency and you're going to want to have it be like a single connection that persists. And the last thing you're going to want to do is like bring up and tear down new connections. And you definitely, you're not going to want a text protocol.

[Marc Andreessen]

And so of course we said no,

[Ben]

we actually want to go completely the other direction. It's obviously we only want text protocol, by the way, same thing in HTML itself. We want HTML to be relatively verbose. We want the tags to actually be human readable.

[SWIX]

We want to use the most inefficient things possible.

[Marc Andreessen]

Yeah, we want to do the inefficient things.

[SWIX]

You're the original token maxer.

[Marc Andreessen]

Yeah, exactly, yeah. Basically it's just like. Well, yeah, well actually this was actually the conscious thing which basically says just assume a future of infinite bandwidth, build for that. And then basically what it was was it was a bet that. It was a bet that if the system, if the latent capabilities of the

[Ben]

system were powerful enough, and that was obvious enough to people that would create the Demand for the bandwidth that would cause the supply of bandwidth to get built, that would actually make the whole thing work.

[Marc Andreessen]

And then specifically what we wanted was

[Ben]

we wanted everything to be human readable because at the engineering level we wanted people to be able to read the protocol coming over the wire and be able to understand it with their, with their bare eyes without having to like disassemble it or whatever. Right. Have it converted out of binary. Right. And so the, the, the, all the pro, you know, HTTP and everything else were, were, it was always text protocols

[Marc Andreessen]

and the same thing with HTML. And in, in many ways, some people

[Ben]

say that the key breakthrough in the browser was the View Source option, which is every web page you go to, you could view source, which means you could see how it worked, which means you could how to build right new, to build new web pages.

[Marc Andreessen]

There was that. So human readability.

[Ben]

And again, human readability in those days still meant technical specs, now it means English language. But there's an incredible latent power in giving everybody who uses the system the option to be able to drop down and actually understand and see how it's working. And that worked really well for the web and I think it's working really well for AI.

[Marc Andreessen]

That was one, what was the other. A big part of the idea of

[Ben]

web servers was to actually surface the underlying latent capability of the operating system and to be able to surface the, also the underlying latent capability of the database. Because basically what was a web server? What is a web server fundamentally? Architecturally it's the operating system. So it's the operating system's ability to, it's running on top of an OS. So it's the OS's ability to manage the file system and do everything else that you want to do, process everything. And then of course, a lot of early, a lot of websites are front ends to databases. And so you wanted to unleash the underlying latent power of whether it was an Oracle database or some other postgres or whatever it was.

[Marc Andreessen]

And so a lot of the function

[Ben]

of the web server was to just bridge from that Internet connection coming in to be able to unlock the underlying power of the OS in the database.

[Marc Andreessen]

And again, people looked at it at

[Ben]

the time and they were like, well, does this really matter? Like, is this important? Because we've had databases forever and we've always had user interfaces for databases and this is just another user interface for a database.

[Marc Andreessen]

It's like, okay, yeah, fair enough.

[Ben]

But on the other side of that, it's just like this is Now a much better interface to databases, and one that 8 billion people are going to use and is going to be far easier to use and far more flexible. And you're not just going to have old databases. Now you have a system where people can actually understand why they want to build a million times more database apps than they had in the past. And then the number of databases in the world exploded.

[Marc Andreessen]

And so again, this goes to this

[Ben]

thing of, like, building, building in layers.

[Marc Andreessen]

Some of the smartest people in industry

[Ben]

look at any new challenge and they're like, okay, I need to build a new kind of application. So the first thing I need to do is build a new programming language, right?

[Marc Andreessen]

And then the next thing I need

[Ben]

to do is build a new operating system, right? And the next thing I need to do is I need to build a new chip, right? And they kind of want to reinvent everything. And I've, I've always had, maybe it's just, I don't know, pragmatic mentality or something, or maybe an engineering over science mentality, but it's more like, no, you have just like all of this latent power in the existing systems and you don't want to be held back, but by their constraints. But what you want to do is you want to kind of liberate that power and open it up. Yeah.

[Marc Andreessen]

And so I, I think, I think.

[Ben]

And I think the web did that for those reasons. And I think it's the same thing now that's happening.

[SWIX]

It's a great perspective in the web.

[Alessio Fenelli]

The programming languages is another good thing. We have Brett Taylor on the podcast and we were talking about Rust. And you know, Rust is memory safe by default.

[Ben]

Default.

[Alessio Fenelli]

So why are we teaching the model to not write memory unsafe code? Just use Rust and then you get it for free. How much do you think there's, like, time to be spent, like, recreating some of these things instead of taking them for granted? I'll be like, oh, okay, Python is kind of slow. Typescript, you know, as, like, yeah, as,

[SWIX]

as imperfect as they are. They are the lingua franca.

[Ben]

I mean, I think this is going to change a lot because I don't think the models care what language they program in. And I think they're going to be good at programming in every language. And I think they're going to be good at translating from any language to any other language.

[Marc Andreessen]

Like, okay, so this gets into the

[Ben]

coding side of things. I think we're going through a really fundamental change. And look, I grew up, you know, I grew up hand code. You know, I grew up hand coding. Everything I did was actually, everything I did actually was written in C. I

[Alessio Fenelli]

wasn't back in the days, I wasn't

[Marc Andreessen]

even using C, so.

[Ben]

Or like Java or any of this stuff. Right, right. And so everything I ever did, I was like managing my own memory at the level of C. And then I'm still from the generation that I knew assembly language and so I could drop down and do things right on the ship.

[Marc Andreessen]

And so all of us, we've always

[Ben]

lived in a world in which software is like this precious thing that you have to think about very carefully. And it's really hard to generate good software and there's only a small number of people who can do it. And you have to be very jealous in terms of thinking about how do you allocate, what are your engineering working on and how many good engineers do you actually have and how much software can they write and how can, how much software can human beings, you know, kind of maintain? And I think like all those assumptions are being shot right out the window right now.

[Marc Andreessen]

Like, I think they're, I, I think

[Ben]

those days are just over. And I think the new world is like, actually high quality software is just like infinitely available. And if you need new software to do xyz, like you're just going to wave your hand and you're going to get it. And then if it's, if you don't like the language it's written in, you just tell the thing. All right, I want the, Right now, I want the REST version.

[Marc Andreessen]

Version or you know, secure, you know, secure.

[Ben]

We're about to, by the way, we're about to go through computer security is about to go through the most dramatic change ever, which is, number one, like every single latent security bug is about to be exposed.

[Marc Andreessen]

Right.

[Ben]

So we're going to have like the, we're set up here for like the computer security apocalypse for a while, but on the other side of it now we have coding agents that can go in and actually fix all the security bugs. And so how are you going to secure software in the future? You're going to tell the bot to secure it and it's going to go through and fix it all.

[Marc Andreessen]

And so this thing that was this

[Ben]

incredibly scarce resource of high quality software is just going to become a completely fungible thing that you're just going to have as much as you want. Right. And that has tons and tons of consequences in some sense. The answer to the question that you posed I think is just somewhat, I Don't know, simple or something. Or straightforward, which is just if you want all your software and rest, you just tell the bot you want all your software and rust.

[Marc Andreessen]

Like, things that used to be, like,

[Ben]

hard or even, like, seem like an insurmountable mountain to get through, all of a sudden, I think, become very easy.

[SWIX]

I think Brett had a theory that there would be a more optimal language for LLMs, and so the contention is there isn't. Like, just. Don't bother. Just whatever humans already use, LLMs are perfectly capable of porting.

[Ben]

I think we're pretty close to being. I don't know if this would work today. I think we're pretty close to being able to ask the AI what would its optimal language be and let it design it.

[SWIX]

True.

[Marc Andreessen]

Okay, here's a question. Are you even going to have programming

[Ben]

languages in the future or the AI is just going to be emitting binaries? Let's assume for a moment that humans aren't coding anymore. Let's assume it's all bots. What levels of intermediate abstraction do the bots even need? Or are they just coding binary directly?

[Marc Andreessen]

Did you see? There's actually an experiment. Somebody just did this thing where they have a.

[Ben]

They have a language model now that actually emits model weights for a new language model.

[Marc Andreessen]

Right. And so will the bots predict the weights?

[Ben]

Yeah.

[Marc Andreessen]

Will the bots literally be emitting not just coding binaries, but will they. Will.

[Ben]

Will they actually be admitting weights for. For new models? Directly.

[Marc Andreessen]

Directly. And conceptually, there's no reason why they

[Ben]

can't do both of those things. Like, architecturally, both of those things seem completely possible.

[SWIX]

Very inefficient. You're basically a simulation of a simulation in a simulation inside of weights.

[Marc Andreessen]

Yeah, Y. Very inefficient. But, like, look, LLMs are already, like, incredibly inefficient. Ask.

[Ben]

Favorite thing.

[Marc Andreessen]

Ask Claude. Add two plus two equals four.

[Ben]

Right. It's just like, you know, it's like, you know, it's like, whatever. Billions and billions of times more inefficient

[Marc Andreessen]

than using your pocket calculator, but yet the payoff is so great of the general capability. So anyway, like, I kind of think in 10 years, like, I'm not sure. Yeah, like, I'm not sure there will even be a salient concept of a

[Ben]

programming language in the way that we understand it today.

[Marc Andreessen]

And in fact, what we may be

[Ben]

doing more and more as a form of interpretability, which is we're trying to understand why the bots have decided to structure code in the way that they have.

[SWIX]

I mean, if you play it through, you don't need browsers, then like that's the depth of the browser.

[Marc Andreessen]

Well, so I, I would take it a step further, which is you may

[Ben]

not need user interfaces.

[Marc Andreessen]

So who is going to use software in the future?

[SWIX]

Other bots.

[Marc Andreessen]

The other bots. Yeah. Yeah.

[SWIX]

And so you still need to, I don't know, pipe information in, do we? And out.

[Ben]

Really?

[SWIX]

Well, what are you going to do then?

[Ben]

Are you sure you're just going to

[SWIX]

log off and touch grass?

[Marc Andreessen]

Whatever you want. Exactly. Isn't that better?

[SWIX]

I want software to do stuff for me.

[Marc Andreessen]

But isn't that better? I mean, look, you know, I don't know. Look, you know, you know the arguments here. You know, it was not that long ago that 99% of humanity was behind a plow.

[Ben]

Right, Right. And what are people going to do if they're not plowing fields all day to grow food? Right. And it just turns out there's like much better ways for people to spend time than plowing fields.

[SWIX]

Yeah, do scrolling.

[Marc Andreessen]

Exactly.

[Ben]

You know, talking to their friends.

[Marc Andreessen]

And look, and I'm not an absolutist

[Ben]

and I'm not a utopian and to be clear, like I have an 11 year old and he's learning how to code and like I'm, you know, I think it's still a really good idea to learn how to code and so forth. But I just. If you project forward, you just have to think forward to a world in which it's just like, okay, I'm just going to tell the thing what I need and it's going to do and

[Marc Andreessen]

then, and then it's going to do

[Ben]

it in whatever way is most optimal for it to do it. Yeah.

[Marc Andreessen]

Unless I tell it to do it. Non, optimally. Like if I tell it to do

[Ben]

it in Java or in Rust or whatever, it'll do it, I'm sure. But like if I'm just going to tell it to do. It's going to do it in whatever way is like the optimal way to do it.

[Marc Andreessen]

And then, and then if I need

[Ben]

to understand how it works, I'm going to ask it to explain to me how it works.

[Marc Andreessen]

Right.

[Ben]

And so it's going to be doing its own interpreter. It's going to be the engine of interpretability to explain itself. And I just am not convinced that, that.

[Marc Andreessen]

I'm not convinced that in that world

[Ben]

you have these historical.

[Marc Andreessen]

The goals of the abstractions will be

[Ben]

whatever the Bosnia network. Yeah, yeah.

[Alessio Fenelli]

Well, I'm curious like if that's true. Then shouldn't the models providers be building some internal language representation that they can do extreme kind of like RL and reward modeling around? Because it's like today they're kind of like tied to like TypeScript and Python because the users need to write in that language versus they can have their own thing internally. And like they don't need to teach it to anybody, they just need to teach their model. And I think that's how you get maybe diversion between the models. Like going back to like the PI open cloud thing. It's like, oh, I built all the software using the OpenAI model and I'll switch to the anthropic model. But the anthropic model doesn't understand the thing. So it feels like there still needs to be some obstruction, but maybe not. Maybe that's the lock in that the model providers want to have. I don't know.

[Ben]

I'm not even sure that's lock in though, because why can't the second model just learn what the first model has done? Like.

[Marc Andreessen]

Exactly. Okay, so. Okay, I'll give you an example. So as you know, models can now

[Ben]

reverse engineer software by. Right. Isn't it this whole thing now where people are reverse engineering like Nintendo, Nintendo game binaries.

[Marc Andreessen]

Yeah. So you have like seen a bunch

[Ben]

of reports like this where somebody has like a favorite game from the 1980s and the source code is like long dead, but they have like a binary burned into a chip or something and now they reverse engineer to get a version that runs on their Mac. Right.

[Marc Andreessen]

And so if you reverse it, this is why I kind of say if

[Ben]

you're reversing like x86 binaries, then why can't you reverse engineer whatever the.

[Alessio Fenelli]

Yeah, and because we're all on a UNIX based system, it has to be reversible because it needs to run on the target.

[Marc Andreessen]

Yeah, yeah, yeah, yeah, yeah, basically. And so I just, I just think

[Ben]

it's this thing where it's just, just like.

[Marc Andreessen]

And by the way, and everything we're

[Ben]

describing is something that human beings in theory could have done before, but just with, but with enormous. But it was just always like cost and labor prohibitive. Reverse engineer. I learned how to reverse engineer. Human beings can reverse engineer binaries.

[Marc Andreessen]

It's just for any complex binary, I

[Ben]

need like a thousand years to do it. But now with the model you don't.

[Marc Andreessen]

And so all of a sudden you

[Ben]

get, you get these things or another way to think about it is so much of human built systems sort of compensate for the human Human limitations.

[Marc Andreessen]

Right.

[Ben]

And if you don't have the human limitations anymore, then all of a sudden you have.

[Marc Andreessen]

And it's not that you won't have

[Ben]

abstractions, but you'll have a different kind of abstraction. Yep.

[SWIX]

I have two topics to bring us to a close, and you can pick whichever ones. Just talking about protocols, was it you or someone else? I forget my Internet history.

[Ben]

Who said that?

[SWIX]

Like, the biggest mistake that we didn't figure out in the early days was payments.

[Ben]

Yes.

[SWIX]

Was that you?

[Ben]

Yes, it was 402. 402 payment required.

[SWIX]

We have a chance now.

[Marc Andreessen]

I don't think we're going to figure it out.

[SWIX]

I don't know. Like, what's your take?

[Ben]

Oh, I think we will. Yeah. No, now I think it's going to happen for sure.

[SWIX]

Yeah.

[Marc Andreessen]

Yeah. And there's two reasons it's going to happen for sure.

[Ben]

One is we actually Internet native money now in the form of stable coins. Stable coins and crypto. And this is. I think this is the grand unification basically of AI Crypto is what's about to happen now. I think AI is the crypto killer app, I think is where this is really going to come out.

[Marc Andreessen]

And then the other is.

[Ben]

It's just, I mean, it's just.

[Marc Andreessen]

I think it's now obvious.

[Ben]

It's like, obviously AI agents are going to need money.

[Marc Andreessen]

It's already happening.

[Ben]

Right. If you've got a. If you've got a claw and you want it to buy things for you, you have to give it money in some form.

[SWIX]

I would say the adoption is probably like 0.1%, if that.

[Ben]

But.

[Marc Andreessen]

Yeah, yeah, yeah, yeah.

[Ben]

But think forward. Like, where is it going forward?

[Marc Andreessen]

Thinking the ultimate principle of everything.

[Ben]

And everything that I think we do is the William Gibson quote, which is the future is already here. It just isn't distributed. Isn't distributed yet.

[Marc Andreessen]

My friends, who are the most aggressive

[Ben]

users of OpenClaw, just like have given their claw's bank accounts, credit cards, and

[Marc Andreessen]

not only have they done it, it's obvious that they needed to do it because it's obvious that they needed to

[Ben]

be able to spend money on their behalf.

[Marc Andreessen]

It's just completely obvious.

[Ben]

And again, so the number of people who have done that today, to your point is, I don't know, probably 5,000 or something, but it'll grow. That's how these things start, actually.

[SWIX]

I mean, since you keep mentioning.

[Marc Andreessen]

And by the way, open claw, by the way, if you don't give it

[Ben]

a bank account, it's just going to break into your qu.

[SWIX]

You it's going to be high agency.

[Marc Andreessen]

It's going to break into your bank account anyway and.

[Ben]

And take your money. So you.

[Marc Andreessen]

You might as. You might as well do it.

[Ben]

You might as well do it.

[Marc Andreessen]

By the way, I really love. I got to tell you, I really love the phenomenon. I love the yolo. I'm not doing it myself, to be clear, but I love the people that are just like.

[Ben]

Yeah, like, what is it? Skip.

[SWIX]

Skip Dangerous, by the way, it's a Facebook thing.

[Marc Andreessen]

Okay.

[SWIX]

Because in Facebook, they have this culture to name the thing dangerous, so that you are aware when you enable the flag that you are opting into a dangerous thing.

[Ben]

Okay, good.

[SWIX]

They brought it into OpenAI, and of

[Marc Andreessen]

course, that makes it enticing.

[SWIX]

Sam runs Codex with Skip permissions on his laptop.

[Marc Andreessen]

Yes, 100%. And so I think the way to

[Ben]

actually see the future is to find the people who are doing that.

[Marc Andreessen]

There's a madness.

[SWIX]

Log everything. You know, just watch it. Watch of the logs.

[Marc Andreessen]

But, like, let's actually find out what

[Ben]

the thing can do.

[Marc Andreessen]

And the way to find out what

[Ben]

the thing can do is just like, try everything.

[Marc Andreessen]

Yeah, let it try everything.

[Ben]

Let it unlock everything.

[Marc Andreessen]

By the way, that's how you're going

[Ben]

to find all the good stuff it can do. By the way, that's also how you're going to find all the flaws.

[Marc Andreessen]

I think the people who turn that

[Ben]

on for bots are like, they're like martyrs to the progress of human civilization.

[Marc Andreessen]

Like, I feel very bad for their descendants that their bank accounts are going

[Ben]

to get looted by their bots in the first, like, 20 minutes.

[Marc Andreessen]

But I think the contribution that they're

[Ben]

making to the future of our species is amazing.

[SWIX]

It's like gentleman science, you know?

[Marc Andreessen]

Yes. It's. Yes.

[Ben]

Experimental. Yes.

[Marc Andreessen]

It's Ben Franklin out with trying to get lightning to strike his balloon and

[Ben]

seeing if he gets electrocuted.

[Marc Andreessen]

Yeah. It's Jonas Salk with the polio vaccine injecting it.

[Ben]

Yes.

[Marc Andreessen]

So, yes, I think we should have. We should have flags and we should

[Ben]

have monuments to the people that just let OpenCloud run their lives.

[SWIX]

More anecdotes. What are the craziest or interesting things that people listening to this should go home and do?

[Marc Andreessen]

I mean, the extreme thing is just the straight yolo.

[Ben]

Just. Yeah, turn your light on.

[SWIX]

That's a general capability. Is there like a specific story that was like, wow. And everyone in the group chat just lit up.

[Marc Andreessen]

I mean, like, you know, so there's tons of. There's already tons of health, you know,

[Ben]

there's the health dashboard stuff is just. It's just personal health. Absolutely amazing.

[Marc Andreessen]

The number of stories on.

[Ben]

I just don't want to violate people's. You know, obviously personal anonymized.

[Marc Andreessen]

But, you know, one of the things

[Ben]

openclaw is really good at is hacking into all the stuff in your lan. It's really good. So, you know, Internet of things, AKA Internet of shit.

[SWIX]

Like, super insecure, but great. Discoverable.

[Marc Andreessen]

It's discoverable. OpenClaw is happy to scale your network,

[Ben]

identify all the things. And then. My friends are most aggressive at this, are having openclaw take over everything in their house. It takes over their security cameras. It takes over whatever their access control systems. It takes over their webcams. I have a friend whose claw watches him sleep.

[Marc Andreessen]

Put a webcam in your bedroom, put the claw in a loop.

[Ben]

Have it wake up frequently and have it watch it. Just tell it. Watch me sleep.

[Marc Andreessen]

And I've seen the transcripts, and it's

[Ben]

literally like, Joe's asleep.

[Marc Andreessen]

This is good. This is good that Joe's asleep because I have his health dad, and I know that he hasn't been getting enough sleep.

[Ben]

And so it's really good that he's getting sleep sleep. I really hope he gets his full whatever, you know, five hours or REM sleep.

[Marc Andreessen]

Joe's moving. Joe's moving. Joe might be waking up.

[Ben]

This is a real.

[Marc Andreessen]

If Joe wakes up now, he's gonna ruin his sleep cycle. Oh, okay. It's okay. Joe just rolled over.

[Ben]

Okay.

[Marc Andreessen]

He's gone back to bed. Okay, good.

[Ben]

All right. Okay. I can relax.

[Marc Andreessen]

This is fine.

[SWIX]

He's monitoring the situation.

[Marc Andreessen]

Situation. And. And being a bot, like, you know, is just, like, very focused, right? It's just like, this is like, his reason for existence is to watch Joe sleep. And then. And then I was talking, my friend who did this. It's like, you know, on the one hand, it's like, all right, this is

[Ben]

weird and creepy, and I need to. I need to maybe this is taking over my life.

[Marc Andreessen]

And then the other thing is, like, you know, what if I had a

[Ben]

heart attack in the middle of the night? This thing literally would, like, freak out and call 911.

[Marc Andreessen]

Like, there's no question this thing would

[Ben]

figure out how to, like, alert medical authorities and, like, probably summon SWAT teams and, like, do whatever would be required

[Marc Andreessen]

to save my life.

[Ben]

Right?

[Marc Andreessen]

And so it's like, you know, like, yeah, like, that's happening.

[Ben]

What else? There's a company unitree that makes the robot dogs, and I actually have one at home. Which it's actually really fun. The Chinese companies, the Chinese companies are so aggressive at adopting new technology, but they don't always, like, take the time to really package it, package it and maybe think it all the way through.

[Marc Andreessen]

And so at least the unitary dog I have, so it has a old non.

[Ben]

LLM just control system, which, by the way, is not very good. It markets well, but in practice it's not that good. It has trouble with stairs and so forth, and so it's not quite what it should be.

[Marc Andreessen]

But then the language model thing comes

[Ben]

out in the voice, so they add LLM capability and then they add a voice mode to it.

[Marc Andreessen]

But, but that LLM capability is not

[Ben]

at all connected to the control system.

[Marc Andreessen]

So. So you've got this schizophrenic dog that like, is a complete idiot when it

[Ben]

comes to climbing the stairs, but it will happily teach you quantum mechanics, right,

[Marc Andreessen]

in like a plummy English accent, right? Like, it's just like absolutely amazing jagged intelligence.

[Ben]

Yeah, yeah, talk about jagged. And then now, obviously what's going to happen in the future is, is they're going to connect together.

[Marc Andreessen]

But, but right now it's, it's. And so right now it's not that useful.

[Ben]

And so I, I have a friend who has one of these who had his claw basically hacking and rewrite the code, write new firmware, write new firmware for the, for the unit robot.

[Marc Andreessen]

And now it's. Now it's an actual pet dog for his kids.

[SWIX]

Did you do the before, after, like the motion?

[Marc Andreessen]

Yeah, it's good. He said it's completely different. He said it's a complete transformation. And whenever there's an issue in the

[Ben]

thing now the claw just like rewrites the code, you know, you know, you. Goes in, does. Does the code and so it kind of goes to your thing here. And so.

[Marc Andreessen]

So like, all of a sudden this

[Ben]

is why we want to think about AI code. AI coding is not just like writing new apps. It's also going in and rewriting all the old stuff that should have worked, that never worked.

[Marc Andreessen]

And so like, I, I think, I

[Ben]

think basically, I think the Internet, the Internet of shit is basically over.

[Marc Andreessen]

Like, I.

[Ben]

There's a potential here where like all these devices in your house that have been like, basically marginal or, you know, basically dumb, you know, like all of a sudden they might all get really smart.

[Marc Andreessen]

Now, smart home. You have to decide if. Yes, there are horror movies in which this. Of which this is the premise. And so you have to decide if you want this. But, but, but this is the first

[Ben]

time I can say with confidence, I now know how you could actually have a smart home with 30 different kinds of things with chips and Internet access, where it actually all makes sense and all works together and it's all coherent and the whole thing.

[Marc Andreessen]

And to have that unlock without a

[Ben]

human being having to go. Do any of that work?

[SWIX]

Like, yes, I'm waiting for a story. Mark, I can't let you open that fridge door, you know, like.

[Marc Andreessen]

Exactly, exactly. Yes, yes.

[SWIX]

Because you're not supposed to eat right now.

[Marc Andreessen]

I have all of. Yes, I have every shred of health information, you know, and I know you think you're doing, you know, da, da,

[Ben]

think you do this, but, you know,

[Marc Andreessen]

this is a real.

[Ben]

Are you really, you know, are you really sure?

[Marc Andreessen]

And you know, you told, you know, you told me last night, you really

[Ben]

don't want me to let you do this.

[Marc Andreessen]

So, you know, I'm sorry, but the

[Ben]

fridge door is locked.

[SWIX]

Open the fridge doors.

[Marc Andreessen]

Exactly. And by the way, I know you're supposed to be studying for a test, so why don't we. Why don't you go. When you can pass the test, I

[Ben]

will open the fridge door for you. Yeah.

[SWIX]

Final protocol. And then, and then we can wrap up proof of human.

[Marc Andreessen]

Yes.

[SWIX]

Right. Yeah, that's the last piece that we got to figure out.

[Marc Andreessen]

Yeah. So I would say there's, there's two massive, I would say sort of asymmetries

[Ben]

in the world right now where we've

[Marc Andreessen]

known these asymmetries exist.

[Ben]

And we, we societally have been unwilling to grapple with them. And I think they're both tipping right now. And, and they're, they're, they're. They're the same thing. It's virtual world version. It's a physical world version. So the virtual world version is. Is the bot problem. We're just like, you know, the Internet. Internet is just like a washing bots. Internet's a washing fake people. It has been forever, by the way. A lot of that has to do with lack of money, you know, and so this, you know, this is.

[SWIX]

Yeah, this is my spicy take was these two are the same thing. And corporations of people too, you know.

[Marc Andreessen]

Interesting. Yeah, yeah, yeah.

[SWIX]

Okay, so a bank account is proof of human.

[Marc Andreessen]

Yeah, okay. Yeah, until you, until you give the bots bank accounts. Yeah, exactly, exactly. So, okay, yeah, so there's that. But yeah, look, look, the bot.

[Ben]

I mean, every social media user knows this. The bot problem is a big problem.

[Marc Andreessen]

You know, the bot problem has been

[Ben]

a big problem forever. It's a huge problem. And it's never really been confronted directly, like, at any point.

[Marc Andreessen]

By the way, the physical world version

[Ben]

of this is the drone. The drone problem. Right. And so we've known for, you know, we've known for 20 years now that the asymmetric threat, both in military and actual military conflict, but also in just like, security, like, you know, security on the home front. The big threat is the cheap attack. Attack drone.

[Marc Andreessen]

Right?

[Ben]

The, the cheap. The cheap suicide, you know, drone with a bomb.

[Marc Andreessen]

And we've known that forever.

[Ben]

And by the way, like, you know, it's very disconcerting how like, every, you know, every office complex in the, you know, in the world is, like, unprotected from drone attacks. Every. Every stadium, every school, every prison, like, it's like, sure, okay, we've known that.

[Marc Andreessen]

We've never done anything.

[SWIX]

What are you going to do about it?

[Marc Andreessen]

Yeah, one possibility is just leave them

[Ben]

unprotected forever and live in a world of like, asymmetric terrorism forever or. The other is take the problem seriously

[Marc Andreessen]

and figure out the set of techniques

[Ben]

and technologies required to. To be able to deal with that. Whether those are like lasers or jammers or early warning systems or personal force fields.

[Marc Andreessen]

Kinetic. Personal dune. Personal force fields. Exactly. And in both cases, these are economic asymmetries. These are economic asymmetries because it's really

[Ben]

cheap to field a bot, but it's very hard to tell something a bot. It's very cheap to field a drone. It's very expensive to defend against a drone.

[Marc Andreessen]

But you see, what I'm saying is

[Ben]

it's the virtual version of the problem and it's the physical version of the

[Marc Andreessen]

problem, the virtual version of the problem.

[Ben]

What we need quite literally is proof of human.

[Marc Andreessen]

The reason is because you're not going

[Ben]

to have proof of bottom the Especially now that the bots are too good.

[Marc Andreessen]

The bots can pass the Turing test. And if the bots can pass the Turing test, then you can't.

[Ben]

You can't screen for bot.

[Marc Andreessen]

You can't have proof of not a bot. But what you can have is you

[Ben]

can have proof of human. You can have, you know, cryptographically validated. This is definitely a person.

[Marc Andreessen]

And this is.

[Ben]

And then you can have cryptographically validated. This is definitely like something that a person said. This video is real. Right.

[SWIX]

Just to double click on. Do you think Alex Blania with world. Do you think he's got it or is there an alternative?

[Marc Andreessen]

Oh, so, I mean, there's going to be. I think there'll be, I think many people will try.

[Ben]

We're one of the key, you know, participants in the world, in the world project.

[SWIX]

Yeah.

[Marc Andreessen]

So we're partisans, but yeah, I, I think so. We think world is exactly correct.

[SWIX]

Okay.

[Marc Andreessen]

And the reason is it has, it has to be, it, it has to

[Ben]

be proof of human.

[Marc Andreessen]

It, it has. Because you can't do proof of not bot.

[Ben]

You have to do proof human. To do proof human, you, you need, you need biological validation. You, you needed to start with this was actually a person, right? Because otherwise you have bots signing up as fake people, right? So you, you have to have like something, you have to have a biometric and then you have to have cryptographic validation and then the ability to do, to do, to, to do the lookup.

[Marc Andreessen]

And then by the way, the other

[Ben]

thing you need was that you, you also need selective disclosure. So you need to be able to do proof of human without revealing all the underlying information.

[Marc Andreessen]

By the way, another thing you're going

[Ben]

to need, you're going to need proof of age, right? Because there's all these laws in all these different countries now around. You need to be 13 or 16 or 18 or whatever to do different things. And so you're, you're going to need, you know, sort of validated proof of age, you know, to be able to legally operate, right? And so that, that's coming and then you're going to want like proof of credit score and you know, proof of like, you know, 100 others.

[SWIX]

That's a tricky one.

[Marc Andreessen]

It is a tricky one. But you're, you're, there's no reason, like if somebody, somebody's checking on your credit, somebody shouldn't give you an example, somebody

[Ben]

shouldn't need to know your name in order to be able to find out whether you're credit worthy, right?

[SWIX]

Independently verifiable pieces of information, pieces of information selectively disclosed.

[Marc Andreessen]

And this is the answer to the

[Ben]

privacy problem writ large, which is I only need to prove what I need to prove at that moment.

[Marc Andreessen]

So like, you're going to need that

[Ben]

and I think their architecture makes sense.

[Marc Andreessen]

So that needs to get solved. I think language models have tipped. The bots are now too good, and so they're undetectable.

[Ben]

And so as a consequence, we now need to go confront that problem directly. And then, like I said, and then the other problem is we need to go actually confront, confront the drone problem. The Ukraine conflict has really unlocked a lot of thinking on that. Now the, and now the, the, the, the, the Iran situation is also unlocking that and so I think there's going to be just like this incredible explosion of both drone and counter drone.

[SWIX]

Our drones are better than their drones. As long as you keep it that way.

[Ben]

Yeah. Encounter drones.

[Alessio Fenelli]

I think we can sneak in one more question. I'm trying to tie together a lot of things that you said over the years. So at the Milken Institute debate with Teal, which is amazing, you talked about the lag between a new technology and kind of of like the GDP impact of it. The other idea you talked about is bourgeois capitalism and how, you know, it's kind of managerial class was needed because of this complexity. And I think if you bring AI into the fold, you have like much

[Ben]

higher leverage of people.

[Alessio Fenelli]

So like if you have, you know, the musk industries and you give Elon a gi, you can run a lot more things at once.

[Ben]

That's right.

[Alessio Fenelli]

And then you have the social contract. And I know you retweeted a clip of Sam Almighty thing. We're rethinking the whole thing. And you're like, absolutely not.

[Marc Andreessen]

Yes.

[Alessio Fenelli]

And I was at an event with Sam last night and he actually said in the last couple weeks, it felt like now people are taking that seriously. So I'm just curious, like how you're seeing the structure of organization changing, especially when you invest in early stage companies and. Yeah, just like how the impact of work structure and all of that is playing out.

[Marc Andreessen]

Yeah. So there's a whole bunch of, there's a whole bunch of times. I know, yeah, we could spend, by

[Ben]

the way, we'd be happy to spend more time, but we could, we could spend more time on all that. So just for people who haven't followed this, so this, this, this term managerial comes from this thinker in the 20th century, James Burnham, who just one of the great kind of 20th century political thinkers, societal thinkers.

[Marc Andreessen]

And he sort of said as, and

[Ben]

he was writing in like the 1940s,

[Marc Andreessen]

1950s, and he said kind of the

[Ben]

whole history of capitalism until that point had been in two phases. Number one had been what he called bourgeois capitalism, which was, think about his like name on the door, like Ford Motor Company because Henry Ford runs the company. And Henry, it's like a dict, Dictatorial model. And Henry Ford just like tells everybody, everybody what to do.

[Marc Andreessen]

And he said the problem with bourgeois

[Ben]

capitalism is it doesn't scale because Henry Ford can only tell so many people to do so many things and then he runs out of time in the day. And so he said the second phase of capitalism was what he called managerial capitalism. Which was the creation of a professional class of managers that are trained not to be like car experts or to be whatever experts in any particular field, but are trained to be experts in management. And then that led to the importance of Harvard Business Schools and management consulting firms and all these things. And then you look at every big

[Marc Andreessen]

company today and most of the executives

[Ben]

at most of the Fortune 500 companies are not domain experts in whatever the company does. And they're certainly not the founders of those companies, but they're professional managers.

[Marc Andreessen]

And in fact, in the course of

[Ben]

their careers, they'll probably manage many different kinds of businesses. They'll rotate around and they might work in healthcare for a while and then work in financial services and then go work in something else, you know, come work in tech.

[Marc Andreessen]

And what Burnham said is, he said that transition is absolutely required because the problem with bourgeois capitalism is it doesn't scale.

[Ben]

Henry Ford doesn't scale.

[Marc Andreessen]

And so if you're going to run

[Ben]

capitalist enterprises that are going to have millions to billions of customers, you're going to need to, they're going to be operating a level of scale and complexity that's going to require this professional management class.

[Marc Andreessen]

And he said, look, the professional management class has its downsides.

[Ben]

They're not necessarily experts at doing the thing. They're not as inventive. They're not going to create the next breakthrough thing.

[Marc Andreessen]

But he's like, whether you think that's

[Ben]

good or bad or whatever is what's going to be required.

[Marc Andreessen]

And basically that's what happened.

[Ben]

And so he wrote that book originally in 1940. Over the course of the next 50 years, basically managerialism, well, I mean today, up till today, managerialism basically took over everything. And what I'm describing is basically how all big companies run and how all governments run and how our large scale nonprofits are run and kind of everything, everything runs.

[Marc Andreessen]

Basically what, what, what venture capital does

[Ben]

is, we basically are a rump sort of protest movement to that to try to find the next Henry Ford or, or just to say Elon Musk or, or the next, or the next Elon Musk or the next Steve Jobs, the next Bill Gates, the next Mark Zuckerberg. And so we, we, we, we start these companies in the old model, right? We, we, we start them out as, as, as, as, like in the Henry Ford model. And so we start them out with a founder or a, or a founder with, with colleagues. But you know, there's the founder, CEO. And then we basically bet that, we basically bet that do things specifically innovate in ways that the big incumbents in that industry are not going to be able to do. And so it's a bet that basically by relighting this sort of name on the door kind of thing, this new innovative thing with like a king, monarchical political structure that they're going to be able to innovate in a way that the incumbent is not going to be able to because the incumbent is being run by managers, right? And by the way, and of course, venture being what it is, sometimes that works, sometimes it doesn't. But we're constantly doing that.

[Marc Andreessen]

But I've always viewed it my entire

[Ben]

life as like we're like raging against the dying of the light. Like we're sort of constantly trying to fight off managerialism, just basically swamping it. Everything and everything getting basically boring and gray and dumb and old, right?

[Marc Andreessen]

And we're trying to keep some level

[Ben]

of energy vitality in the system. AI is the thing that would lead you to think, wow, maybe there's a third model, right? And, and maybe may, and way to think about it would be maybe it's a combination of the two. Maybe the new Henry Ford or the new Elon or the new Steve Jobs. Plus AI is the best of both, right?

[Marc Andreessen]

Because it's, it's, it's sort of the

[Ben]

spark of genius of the name on the door model, the Henry Ford model. But then it's give that person AI superpowers to do all the managerial stuff and let the boss drill the manageri stuff. That may be the actual secret formula.

[Marc Andreessen]

And we've never even known that we wanted this because we never even thought it was a possibility. But I mean, you know, this. What is the thing that these bots are really good at?

[Ben]

They're really good at doing paperwork.

[Marc Andreessen]

Like they're really good at filling out forms. Like, they're really good at writing reports, they're really good at reading, they're really

[Ben]

good at doing all the managerial work.

[Marc Andreessen]

Like they're amazing at it. And so yeah, so I think, I think the 100%, I think the answer,

[Ben]

the answer very well might be to get the best, best of both worlds by doing this.

[Marc Andreessen]

And then the challenge is going to be twofold.

[Ben]

The challenge is going to be for the innovators to really figure out how to lever AI to actually do this, right?

[Marc Andreessen]

And then, and then the other challenge

[Ben]

is going to be for the, for the incumbents that are managerial to figure out like, okay, what does that mean? Because now they're going to, they're, they're going to be Facing a different kind of insurgent competitor that has a different set of capabilities than they're used to.

[Marc Andreessen]

And so this really, I think is

[Ben]

going to force a lot of big companies to kind of figure out innovation. Either say figure out innovation or die trying.

[Alessio Fenelli]

Do you feel like that structure accelerates the impact on the actual GDP and economy? If you guys SpaceX is like the growth is like so fast and like instead of having these companies kind of like peter out in growth and impact, they can kind of like keep going if not accelerating.

[Marc Andreessen]

That's for sure.

[Ben]

The hope, the, the, the challenge. And, and you know, and look, the AI utopian view is of course, of course. And, and, and that's going to be the future of the economy and it's going to grow a 10x and 100x and a thousandx and we're entering this regime of like much higher economic growth forever and consumer cornucopia of everything. And it's going to be great. And, and, and I hope that's true. I hope that's, that's like the, you know, that's the current kind of utopian vision. I hope that's true.

[Marc Andreessen]

The problem is, goes back again.

[Ben]

The real world. Give you an example how the real world is really Messy. It requires 900 hours of professional certification training to become a hairdresser in the state of California.

[Marc Andreessen]

So it's like 35% of the economy, something like that.

[Ben]

You have to get some sort of professional certification to do the job.

[Marc Andreessen]

Which is to say that the professions

[Ben]

are all cartels, right?

[Marc Andreessen]

And so you have to get licensed

[Ben]

as a doctor, you have to get licensed as a lawyer, you have to get licensed as a, you have to

[Marc Andreessen]

get into a union by the way,

[Ben]

to, to work for the government. You need to be, you have both civil service protections and you have public sector unions. You have two layers of insulation against ever getting fired for anything or anything, anything ever changing.

[Marc Andreessen]

I'll give you another example. The, the dock work, the dock workers

[Ben]

went on strike a couple years ago because they're, you know, robotics. You know, if, if you go look at a modern doc, like in Asia it's all robots. If you go to American doc, it's like all still guys dragging, strike, dragging stuff by hand. The dock workers on a strike.

[Marc Andreessen]

It turns out there are 25,000 dock

[Ben]

workers working on, on, on docs in America. It turns out they have incredible political power because it's a, it's, it's one of these unified blocks of things. They won their strike and so they got comm dock owners to not implement more automation.

[Marc Andreessen]

We learned a couple things in that.

[Ben]

So number one, we learned that even

[Marc Andreessen]

a union as small as 25,000 people

[Ben]

still has like tremendous political stroke. We also learned that they, it actually turns out the dock workers union has 50,000 people in it because there's 20.

[Marc Andreessen]

They have 25,000 people working at docks. They have 25,000 people during full paycheck

[Ben]

sitting at home from prior union agreements. From prior union agreements.

[Marc Andreessen]

I'll give you another great example.

[Ben]

There are government agencies, there are federal government agencies where the employees right of have civil service protections and they're in public sector unions. There are entire federal government agencies that struck new collective bargaining agreements during COVID where not only are they have their jobs guaranteed in perpetuity, but they only have to report to work in an office one day per month.

[Marc Andreessen]

And so there are entire office buildings

[Ben]

in Washington D.C. that are empty 29 out of 30 days of the year that are still operating and they're still, we're all still paying for it.

[Marc Andreessen]

And then what they do, it turns

[Ben]

out what the employees do is they're very, they're very smart in, in this way. And so they figure out they come in on the last day of a month and the first day the next month.

[Marc Andreessen]

And, and so they're, so they're in there, they're in the office two days per 60 days, which means these buildings are empty for 58 days at a time. And you see, you see where I'm heading with this? Like, this is like locked in, right? This is like locked in in a way that has nothing to do with like.

[Ben]

And people say capitalist. It's like anti capitalistic. It's like, it's, it's basically, it's restrictions on trade, it's restrictions on the ability to like change the workforce.

[Marc Andreessen]

And so, so much of our economy is, is, you know, the, I'm describing

[Ben]

the entire, entire healthcare system. I'm describing the entire legal profession. I'm describing the entire housing industry. I'm describing the entire education system, right?

[Marc Andreessen]

K through 12 schools in the United States.

[Ben]

They're a literal government monopoly.

[Marc Andreessen]

How are we going to apply AI in education? The answer is we're not. Because it's a literal government monopoly.

[Ben]

It is never going to change the end. And there is nothing to do, by

[Marc Andreessen]

the way, you can create an entirely new school system. Like that's the one thing you can

[Ben]

do, is you can do what Alpha School is doing. You can create an entirely new school system.

[Marc Andreessen]

Other than that, you're not going to

[Ben]

go in and change what's happening in the American classroom, like K through 12. There's no chance the teachers are 100% opposed to it. It's 100% not happen.

[Marc Andreessen]

So. So you see what I'm saying is like there's this like massive slippage that's

[Ben]

going to take place.

[Marc Andreessen]

Both the AI utopians and the AI

[Ben]

doomers are far too optimistic.

[Alessio Fenelli]

Right.

[Marc Andreessen]

You see what I'm saying? Because they believe that because the technology makes something possible that 8 billion people

[Ben]

all of a sudden are going to change how they behave. And it's just like, nope. So much of how the existing economy works is just. It's just like wired in.

[Marc Andreessen]

And so we're going to be lucky as a society.

[Ben]

We're going to be lucky if AI adoption happens quickly. Right. Because if it doesn't, what doing? We're just going to have a stagnation.

[Alessio Fenelli]

Mark, I know you got to run.

[SWIX]

Yeah, we all know.

[Ben]

Or stay.

[SWIX]

Welcome. But it was such a pleasure talking to you. We're truly living in an age of science fiction. Coming to real life.

[Marc Andreessen]

Yes. Yes.

[Ben]

Could not be more exciting.

[Marc Andreessen]

Really?

[Alessio Fenelli]

With you guys.

[Ben]

Awesome.

[Marc Andreessen]

Thank you.

[SWIX]

That's it.

[Marc Andreessen]

Good. Thank you.

[Alessio Fenelli]

That's it.

[Podcast Host]

As a reminder, please note that the content here is for informational purposes only. Should not be taken as legal, business, tax or investment advice or be used to evaluate any investment or security is not directed at any investors or potential investors at any A16Z fund. For more details, please see a16z.com disclosures.