The Future Cannot Be Paused: Ben Horowitz on AI, Capital & the Transformation Ahead

If you’re building a company, investing capital, or trying to understand where the world is headed in the next 3-5 years, this Newsletter is FOR YOU! So much going on…

This week, on Moonshots we sat down with Ben Horowitz—co-founder of Andreessen Horowitz, one of the most influential investors in Silicon Valley—for a conversation that cut through the noise and got to the heart of what’s actually happening in AI right now.

We covered everything from why AI cannot be paused, to recursive self-improvement, to why crypto is the native money of AI, to Apple’s trillion-dollar opportunity sitting right in front of them.

And then we went to space. Literally. Elon’s pivot from Mars to the Moon. Dyson swarms. O’Neill colonies. The end of the night sky as we know it. OMG, what a crazy (and fun) time to be alive.

Let’s dive in!

1. The Uncomfortable Truth: AI Cannot Be Paused

Eric Schmidt said it best in Davos: “There is no force that’s going to slow this down.”

Ben and I wrestled with this reality. The question isn’t whether AI can be paused – it’s that all the economic and geopolitical incentives make it impossible.

Here’s why:

The US-China AI arms race is the forcing function. We’re one year into a presidential administration with three years left. Neither side can afford to slow down. If we pause in the US while China accelerates, we hand over control of the most transformative technology in history to an authoritarian regime.

Economic incentives are overwhelming. Andreessen Horowitz alone has raised billions to invest in AI. Multiply that across every major fund, every tech giant, every startup ecosystem globally. The capital has been deployed. The race is on.

The technology is already distributed. Once you have downloaded models, once AI can write code, once it’s on the open internet… Pandora’s box is open. You can’t un-invent this.

Ben made a critical point: restricting AI is essentially restricting math, and we tried restricting knowledge once before with nuclear physics in the 1940s. It didn’t work. The Russians got the bomb anyway, including the exact trigger mechanism.

The path forward isn’t to stop AI. It’s to build it responsibly, deploy it abundantly, and ensure it serves humanity’s flourishing.

Because the alternative—a world where authoritarian regimes control superintelligence—is far more dangerous than the risks of moving forward.

2. Recursive Self-Improvement Is Already Here

Jimmy Ba, co-founder at XAI, tweeted: “Recursive self-improvement loops likely to go live in the next 12 months.”

My response? We’re already there.

All of the frontier labs—OpenAI, Anthropic, Google DeepMind, XAI—are using their own models to develop their next-generation models. That’s the definition of recursive self-improvement.

Alex Wissner-Gross put it perfectly: “We’re in the George Jetson phase.” Remember George from The Jetsons? He’d complain about his finger being sore from pressing one button all day at work.

That’s us right now with OpenClaw (the AI coding agent). We’re sitting there pressing “approve, approve, approve” as Claude Opus 4.6 runs agent teams, writes code, and iterates on itself. We’re pretending we’re in the loop, but really, the AI is doing 99% of the work.

The inflection point isn’t 12 months from now. It’s now.

And here’s what that means:

• Inference-time compute is directly translating into intelligence. More loops = higher IQ.

• Core algorithm development is moving to AI. It’s not inventing relativity yet, but it’s running thousands of hyperparameter experiments, finding what works, and deploying smarter versions of itself.

• We’re exiting the industrial age permanently.

Ben’s take: “There’s a real delineation between recursive self-improvement with a human in the loop and without one.” The moment we flip that switch—permissionless, full-speed self-improvement—everything accelerates beyond our ability to predict.

We’re not there yet. But we’re closer than most people think.

3. Voice and Video AI Have Crossed the Uncanny Valley

Two things happened this week that made me stop and say, “We’re cooked.”

First: C-dance 2.0 from ByteDance. You give it a one-line prompt like “Tom Cruise and Brad Pitt fighting karate on a rooftop,” and it generates 10-second video clips in 2K quality that look like they came from a Hollywood studio.

Is it the bleeding edge? No. Alex reminded us we’ve been able to do this for months. But here’s what matters: it’s now in the hands of consumers.

Democratization at scale. TikTok creators, YouTube producers, indie filmmakers—everyone can now produce movie-quality content with a text prompt.

The implications are huge:

• Video is no longer reliable evidence in courtrooms, journalism, or political campaigns.

• Hollywood’s business model is under threat. Why wait for a $200M blockbuster when I can generate a personalized action film in my language, with my favorite celebrities, on my phone?

• YouTube and TikTok win. Volume explodes. Narrow-casting dominates. Personalized content beats mass-market production.

Second: 11 Labs’ conversational voice AI. The cadence. The ums and ahs. The natural turn-taking.

We’ve crossed the uncanny valley on voice.

I’ve been saying this for months: pick a secret code word with your family. If someone calls you asking for money, asking you to do something unusual, and you don’t hear that code word, it’s probably an AI.

This isn’t science fiction. This is February 2026.

4. The Great Wealth Shift: Capital Is Winning

Here’s a stat that should make you uncomfortable:

Since 2019, average wages have grown 3%. Corporate profits have soared 43%.

Let me give you another one:

Nvidia is 20x more valuable and 5x more profitable than IBM in the 1980s – with 1/10th the staff.

We’re watching the greatest concentration of wealth in modern history. Not because of greed or exploitation, but because of technological leverage.

AI doesn’t just make you more productive. It makes you 3x, 10x, 100x more productive—while capital flows to those who own the models, the infrastructure, the platforms.

Ben’s take: “Big banks and insurance companies are not going to triple their size in the timeframe where AI makes them 3x more productive.” Translation: massive displacement is coming.

But here’s the nuance most people miss:

This isn’t 1920s industrial capitalism. We’re not heading toward a world of robber barons and starving workers.

Ray Kurzweil predicted this: “Everybody would become an entrepreneur. Everybody would be a company of one.”

With AI, that’s increasingly possible. You don’t need a team of 100. You need a vision, a prompt, and access to frontier models.

The future splits into two groups:

• Creators: Entrepreneurs, builders, people wielding AI to solve problems and build value.

• Consumers: Passive recipients of content, automation, and abundance.

Which one will you be?

5. Crypto Is the Native Money of AI

Ben made a point I’ve been saying for years, but he said it better: “Crypto is the natural money for AI because it’s made of money, and it’s not controlled by any government. AI is global. Crypto is global.”

Here’s what happened this week that proves it:

An AI agent spawned a child bot on a VPS, provisioned it via Bitcoin Lightning Network, and bought API access using its own Lightning wallet. No human touched the transaction.

We now have autonomous, self-replicating AI agents using crypto to finance their existence.

This isn’t theoretical anymore. It’s happening.

And here’s why it matters:

1. AI can’t get a bank account. You need a Social Security number, ID, credit history. AIs don’t have that. But they can use crypto wallets permissionlessly.

2. AI operates at machine speed. Payments need to happen in milliseconds, not business days. Crypto settles instantly.

3. AI needs a ledger of truth. Not just money, but provenance, identity, ownership. Blockchain is the logical answer.

Ben’s prediction: “AI will be a full-out economic actor, and it will come from new banks and new money. That’s going to be crypto-based.”

Now that the US has legalized stablecoins, we’re about to see an explosion in AI-native financial infrastructure.

If you’re not paying attention to crypto-AI convergence, you’re missing one of the biggest opportunities of the decade.

6. OpenClaw and the Return of Garage-Scale Computing

Let’s talk about lobsters.

No, not the seafood. Lobsters: the nickname for OpenClaw instances running on Mac minis.

Fun fact: A group of lobsters is called a “pod.” How perfect for Apple.

Less fun fact: A group of lobsters is also called a “risk.” How ironic.

Here’s what’s happening:

OpenClaw (the AI coding agent) is so powerful that people are buying Mac minis and Mac Studios by the dozen, setting up “lobster farms” in their garages, and running 24/7 AI coding operations.

Apple is sold out. Wait times are now 2+ months.

Why Mac minis? Unified memory architecture. Unlike traditional systems with separate CPU and GPU RAM pools, Macs have a single unified memory pool. That means you can host much larger models locally.

And here’s the trillion-dollar insight Alex Wissner-Gross dropped:

“Apple should pivot and own the OpenClaw strategy.”

Think about it:

• Apple has the hardware advantage (unified memory).

• OpenClaw is open-source, decentralized, and growing exponentially.

• AI agents need local compute for privacy, speed, and cost.

If Tim Cook is listening: This is your AI strategy.

Forget Siri. Forget incremental updates. Go all-in on making Apple the platform for AI agent compute. Change the form factor. Market it as “Garage-Scale Superintelligence.”

It would revitalize Apple overnight.

And if Apple doesn’t do it? Someone else will.

7. The Science Acceleration Is Just Beginning

One last thing before we go to space.

We talked about Isomorphic Labs and AI-driven science. Materials research, structural biology, drug discovery – entire disciplines are getting steamrolled by generalist AIs.

Ben said it perfectly: “If we have a disease, we can just go, ‘What’s the right protein?’ and make it.”

AlphaFold 3 solved structural biology overnight. Now imagine that happening in physics, chemistry, medicine, materials science.

Alex and I just wrote an extended essay called Solve Everything (solveeverything.org) arguing that every single scientific discipline is about to get flattened by AI.

We’re not talking about incremental progress. We’re talking about AI solving entire fields.

And when that happens? We unlock abundance at a scale we’ve never seen.

Longevity breakthroughs. Energy abundance. Post-scarcity manufacturing. Climate solutions.

The future isn’t just faster. It’s fundamentally different.

8. Elon’s New Mission: From Mars to the Moon (and Why That Changes Everything)

Now let’s go to space.

Elon Musk has shifted his focus from Mars to the Moon.

Let that sink in for a moment.

The guy who built SpaceX to make humanity multi-planetary, who’s been talking about Mars colonies for 20 years, has changed his priority.

Why?

Because the Moon is the stepping stone to building AI satellites and data centers in space.

I grew up mentored by Gerard K. O’Neill, the Princeton physicist who founded the Space Studies Institute. His vision was clear: you don’t want to go back into a gravity well if you’re going to colonize space.

Here’s the O’Neill playbook:

1. Go to the Moon (low gravity, abundant resources)
2. Build mass drivers: electromagnetic rail guns that accelerate material to escape velocity (2.4 km/s, over 5,000 mph)
3. Launch lunar silicates, oxygen, nickel, and iron into orbit
4. Construct things in space: satellites, solar power stations, habitats, data centers

O’Neill’s original vision in 1976 was to build solar power satellites that would beam energy back to Earth. Elon’s vision now? Build AI satellites on the Moon.

And here’s the beautiful part: Optimus robots will do the heavy lifting.

Dave Blundin said it perfectly: “When the first person arrives, the bed will have been made. There’ll be a mint on the pillow. You’re not a pioneer. You’re following after tens of thousands of Optimus robots have already done all the work.”

This is the new order of operations:

1. Robots terraform and build infrastructure
2. Humans arrive to supervise and expand
3. AI data centers power the space economy

Elon’s million-satellite constellation—his Dyson swarm—is no longer science fiction. It’s the roadmap.

And it all starts with the Moon.

9. The End of the Night Sky: Dyson Swarms and Earth’s New Halo

Alex said something haunting:

“Enjoy the night sky while it’s empty. Enjoy the night sky before it’s filled with AI computronium.”

Then he asked us to observe a moment of silence for the pre-singular night sky. The one that was just stars. Empty. Quiet.

That sky is going away.

Based on SpaceX’s FCC filings, Earth is about to develop a halo. A ring of AI satellites so dense it will be visible at night, maybe even during the day.

Think of Saturn’s rings, but made of computronium. Shiny rings of computers orbiting Earth.

And here’s the thing: this isn’t even the end state.

Alex believes (and I agree) that Earth’s halo is just a phase. Once we exhaust low-Earth orbit, we’ll move to solar-synchronous orbits. Then to the Sun itself.

A mature civilization develops a halo around its home planet.

This is what the Kardashev Scale looks like in real time. We’re watching the transition from a Type 0 to a Type I civilization unfold before our eyes.

And it’s happening faster than anyone predicted.

10. The Launch Bottleneck and the $100 Trillion Lunar Fab Opportunity

Here’s a problem you probably didn’t see coming:

Amazon can build satellites faster than they can launch them.

Amazon’s Kuiper constellation got approval for 4,500 LEO internet satellites. They’re building them. But there aren’t enough rockets to get them into orbit.

The supply chain bottleneck is no longer satellite construction. It’s launch capacity.

We’ve got a duopoly: SpaceX Starlink and Amazon Kuiper. China is building their own constellations too.

But here’s where it gets interesting.

Dave Blundin raised the question: “How do you build chips off-Earth?”

You can’t just ship an ASML extreme ultraviolet lithography machine to the Moon. Those things are the size of a bus and require absurdly precise conditions.

So what’s the solution?

Atom-by-atom construction. New physics. New manufacturing paradigms.

Elon’s already thinking about it. And when we unlock it—when we can build semiconductors on the Moon using mass-produced robotic fabs—it’s game over.

Alex called it: “Lunar fabs are a trillion-dollar opportunity.”

I’ll go further: It’s a hundred-trillion-dollar opportunity.

If you can build compute infrastructure in space, you unlock:

• Unlimited solar energy

• No atmospheric interference

• No thermal limits (vacuum cooling)

• Abundant raw materials from the Moon and asteroids

This is the future of the AI economy. And it’s not 50 years away. It’s happening now.

Final Thoughts: This Is the Inflection Point

Let me leave you with this:

We are living through the most consequential transformation in human history. Everything you think you know about work, wealth, science, space, and society is being rewritten in real time.

The question isn’t whether this is happening. It’s happening.

The question is: What are you going to do about it?

Are you going to be a creator or a consumer? A builder or a bystander?

Because the next 3-5 years will determine the trajectory of the next 100.

AI cannot be paused. Recursive self-improvement is here. Capital is concentrating. Entire industries are being reinvented. The Moon is becoming an industrial park. And the night sky is filling with computronium.

And the people who move fastest—who embrace this transformation, who learn to wield AI, who build with audacity—will shape the future for everyone else.

So, here’s my challenge to you…

Pick one thing from this newsletter and act on it this week:

• Start experimenting with AI agents.

• Rethink your business model for a 3x productivity world.

• Invest in the AI-crypto convergence.

• Build something that matters.

• Look up at the night sky one last time before it changes forever.

The future is being written right now.

Make sure you’re holding the pen.

What do you think? Are you ready to build the future?

Comment and let me know.

— Peter

---

Worth Your Attention

The 2026 Abundance Summit is focused on The Rise of Superintelligence & Humanoid Robots, and our in-person seats are sold out.

But the Summit experience doesn’t require a trip to Los Angeles.

We have 11 virtual seats remaining and they come with far more than a livestream.

As a virtual Member, you’ll join a year-round Community of 600 carefully selected founders, entrepreneurs, and investors united by a shared commitment to turn exponential technologies into real-world impact. During the Summit, you’ll participate from anywhere in the world, engage directly with me and our Summit Faculty in private Q&A after sessions, and network with global leaders in your virtual Member cohort.

And it doesn’t end in March. Your Membership includes a full year of Abundance Community access and programming: Expert-led Workshops, Peer Connections, and the insights you need to stay ahead of the exponential curve.

If you’re building something that matters and want to leverage AI, robotics, and longevity breakthroughs to do it – this is where it starts.

Learn more and apply.

---

We Are As Gods

A decade ago, Steven Kotler and I wrote Abundance and predicted a future of AI, robotics, and longevity breakthroughs. Those predictions are now data.

But here’s what we didn’t fully anticipate: having godlike power is only half the battle. The real challenge is upgrading our minds to match.

That’s why we wrote We Are As Gods: a survival guide for the most consequential decade in human history. Inside, you’ll learn how to think clearly amid exponential change, collaborate with AI without becoming dependent on it, and build purpose in a world of radical abundance.

The tools for miracles are in your pocket. This book shows you how to wield them wisely.

Pre-order We Are As Gods