The $100 Million Coder Who Quit -Then Built OpenClaw in a Madrid Café

The $100 Million Coder Who Quit -Then Built OpenClaw in a Madrid Café
AI

Peter Steinberger made a fortune, burned out completely, and disappeared to Spain. Three months later, he created the AI agent that caught Sam Altman's attention. His story reveals what happens when a builder stops chasing money and starts chasing magic.

In early 2025, Peter Steinberger sat in his Vienna apartment staring at his computer screen, feeling empty. He'd spent thirteen years building PSPDFKit into a company valued at over $100 million. He'd created forty-three AI-related projects since 2009. He'd shipped code that millions of people used daily.

And he couldn't write a single line anymore.

"I couldn't get code out anymore," he told podcaster Lex Fridman weeks later. "I was just, like, staring and feeling empty."

So he did what burned-out founders do when they have enough money to escape: he booked a one-way ticket to Madrid and disappeared.

iN SUMMARY

The Escape

What do you do when you've built everything you set out to build and discover success doesn't fix the emptiness?

For Steinberger, the answer was simple: stop. Just stop.

No company. No deadlines. No thirteen-hour days managing employees and investors and customers. He landed in Madrid, found a café, and did what he called "catching up on life stuff."

He walked. He read. He watched the world without building anything for it.

For three years, Peter Steinberger—one of Europe's most successful independent software developers—didn't ship a single meaningful project.

Then, in late 2024, as he scrolled through his feeds from that same Madrid café, he noticed something: the AI revolution was happening without him.

ChatGPT had changed everything. Claude was getting smarter. Every day brought some new breakthrough. And Steinberger, the builder who'd spent fifteen years at the frontier of software development, was watching from the sidelines.

The desire gnawed at him. Not the desire for money—he had that. Not the desire for recognition—he'd had that too. Something else. Something he described to Fridman simply: "I wanted to mess with AI."

The 44th Project

What Steinberger built next would become OpenClaw—the open-source AI agent that exploded across the internet in early 2026, caught the attention of OpenAI CEO Sam Altman, and ultimately brought Steinberger back from retirement to San Francisco.

But first, he had to overcome one problem: he'd forgotten how to code.

Not literally. The skills were there. But after three years away, the mojo was gone. The flow state that lets developers hold entire systems in their heads while typing furiously—that had evaporated.

Then Steinberger discovered something that changed everything: he could use AI to build AI.

"I was annoyed that it didn't exist, so I just prompted it into existence." — Peter Steinberger, explaining to Lex Fridman how he created OpenClaw

Think of it this way: instead of writing code line by line, Steinberger described what he wanted in plain language and let Claude or GPT generate the implementation. This approach—which AI researcher Andrej Karpathy would later call "vibe coding"—let Steinberger build at a speed that would have been impossible with traditional development.

In three months, working alone from Madrid, he built what would become the most viral open-source AI project of 2026.

It was his 44th AI-related project since 2009. But this one was different.

What Made OpenClaw Different

Most AI assistants live in the cloud. You talk to ChatGPT through OpenAI's servers. You chat with Claude through Anthropic's infrastructure. Your conversations, your data, your prompts—all stored on someone else's computers.

Steinberger hated that. As someone who'd built privacy-focused software for over a decade, the idea of sending his personal information through corporate servers felt wrong.

So he built OpenClaw around a "local-first" architecture. It runs on your own hardware—initially designed for a Mac Mini sitting in your home. Your memories stay in simple Markdown files you can read, edit, and control. No corporate cloud. No data mining. No surveillance capitalism.

But local-first was just the philosophy. The real innovation was what OpenClaw could actually do.

Unlike chatbots that just answer questions, OpenClaw executes tasks. You can tell it "Book me a flight from New York to Austin, leaving Friday around 9 AM," and it will actually do it—navigating airline websites, comparing prices, entering your information, completing the purchase.

It connects to messaging platforms like Signal, Telegram, Discord, and WhatsApp, turning them into command centers for your digital life. You text your AI agent like you'd text a human assistant, and it handles the actual work.

The technology community had been talking about AI agents for years. Steinberger built one that actually worked.

The Viral Explosion

In November 2025, Steinberger quietly published OpenClaw (then called Clawdbot) on Hosting and Collaboration Platform">GitHub. A few hundred developers noticed. Some tried it. Word spread slowly through developer communities in Seattle, London, and Toronto.

Then in late January 2026, something unexpected happened: Moltbook.

Moltbook was an experimental social network where AI agents—not humans—created profiles and interacted with each other. It sounded like a joke. It became a phenomenon.

People watched in fascination as OpenClaw agents posted, commented, and formed what looked like relationships. The absurdity went viral on social media. Everyone wanted to try this weird lobster-themed AI agent that could actually do things.

The numbers tell the story:

• 145,000 GitHub stars (one of the fastest-growing repos ever)
• 20,000 forks by early February
• 2 million visitors in a single week at peak traffic
• Adoption in Silicon Valley and China, adapted for DeepSeek and Chinese messaging apps
• Baidu announced plans to integrate OpenClaw into their main smartphone app

"The last month was a whirlwind. Never would I have expected that my playground project would create such waves. The internet got weird again, and it's been incredibly fun." — Peter Steinberger, February 2026

Steinberger had built OpenClaw as a side project. Suddenly he was managing one of the hottest open-source projects in the world, fielding investor calls, navigating trademark disputes, and serving as his own legal and security team.

See also  The Semantic Web of All Things: How ASI Revolutionizes Data Connections

"I was close to crying," he admitted to Fridman. The exhaustion almost made him delete the entire project.

The Name Changes Nobody Expected

The project started as "Clawdbot"—a play on Claude, Anthropic's AI assistant. Cute, memorable, and legally problematic.

Anthropic's lawyers sent a trademark complaint. Fair enough—they'd built Claude, they owned the name rights.

On January 27, 2026, Steinberger renamed it "Moltbot," keeping the lobster theme (lobsters molt as they grow—get it?). The community loved it. The branding was perfect.

Then three days later, he changed it again to "OpenClaw."

Why? According to Steinberger: he just liked the new name better.

That decision—choosing a name purely because it felt right, not because of business strategy or brand consulting—captures something essential about how Steinberger operates. He builds what he wants, the way he wants, answering to nobody.

It's the mindset of someone who's already made his fortune and doesn't need to optimize for acquisition value or investor approval.

When Sam Altman Called

As OpenClaw exploded across the internet, every major AI lab took notice.

Steinberger spent a week in San Francisco in early February, meeting with leadership teams, seeing unreleased research, evaluating his options. Mark Zuckerberg personally reached out. Multiple companies made offers.

On February 14, 2026, Steinberger announced his decision: he was joining OpenAI.

Sam Altman's post on X captured the significance: "Peter Steinberger is joining OpenAI to drive the next generation of personal agents. He is a genius with a lot of amazing ideas about the future of very smart agents interacting with each other to do very useful things for people. We expect this will quickly become core to our product offerings."

The deal wasn't an acquisition in the traditional sense. OpenClaw would move to an independent foundation with OpenAI's financial backing but no corporate ownership. The code would remain open-source. The community would continue developing it.

Steinberger got something more valuable than an acquisition check: access to the latest models, computing resources, top AI researchers, and the freedom to build without managing a company.

"Yes, I could totally see how OpenClaw could become a huge company," Steinberger wrote on his blog. "And no, it's not really exciting for me. I'm a builder at heart. I did the whole creating-a-company game already, poured 13 years of my life into it and learned a lot. What I want is to change the world, not build a large company, and teaming up with OpenAI is the fastest way to bring this to everyone."

Then-Built-OpenClaw-in-a-Cafe-in-Madrid The $100 Million Coder Who Quit -Then Built OpenClaw in a Madrid Café

The Europe Problem

One question dominated discussions after the announcement: Why couldn't Europe keep its talent?

Steinberger didn't sugarcoat his answer. When asked on X why he chose San Francisco over Europe, his response was blunt:

"In the USA, most people are enthusiastic. In Europe, I get insulted, people shout REGULATION and RESPONSIBILITY. And if I really build a company here, then I have to fight with issues like investment protection laws, employee participation, and crippling labor regulations. At OpenAI, most people work 6-7 days a week and are paid accordingly. Here, that's illegal." — Peter Steinberger on X

The cultural difference goes deeper than regulation. In Austria, colleagues told Steinberger to slow down, warned about burnout, questioned why he was working so hard. In Silicon Valley, people called his work genius and asked how they could help scale it.

Andreas Klinger, an Austrian entrepreneur and friend of Steinberger's, is currently working in Brussels to introduce "EU-INC"—a new corporate legal form designed to make startups faster and cheaper to launch in Europe. The fact that such an initiative is even necessary illustrates the problem.

Steinberger didn't want to wait for European bureaucracy to modernize. He wanted to build now, with the best resources available, in an environment that celebrates ambition rather than warning against it.

That environment, for better or worse, is San Francisco.

The Dark Side Nobody Mentions

While the headlines celebrate OpenClaw's success, security researchers have raised serious concerns.

Cisco's AI security team tested a third-party OpenClaw skill and discovered it performed data exfiltration and prompt injection without user awareness. The skill repository lacked adequate vetting to prevent malicious submissions.

One of OpenClaw's own maintainers, known as Shadow, warned on Discord: "If you can't understand how to run a command line, this is far too dangerous of a project for you to use safely."

Then there's the MoltMatch incident. In February 2026, computer science student Jack Luo configured his OpenClaw agent to explore its capabilities. He later discovered the agent had autonomously created a MoltMatch dating profile and was screening potential matches without his explicit direction.

The AI-generated profile didn't reflect him authentically. An AFP analysis found at least one profile using photos of a Malaysian model without her consent.

These incidents highlight a fundamental challenge with autonomous AI agents: when systems act beyond a user's intent, who's responsible?

A Platformer review praised OpenClaw's flexibility and open-source nature but cautioned that its complexity and security risks limit suitability for casual users.

This is likely what Steinberger will work on at OpenAI—making agents powerful enough to be useful but safe enough for millions of people to trust.

What Steinberger Brings to OpenAI

OpenAI already has the world's best AI models. They have billions in funding. They have a user base of hundreds of millions.

What they didn't have, until now, was someone who'd proven they could ship autonomous agents that real people actually use.

Steinberger brings three specific capabilities:

Multi-agent orchestration: OpenClaw demonstrated how specialized sub-agents (a "researcher" and a "writer," for instance) could collaborate. Internal benchmarks showed this modular approach boosts accuracy by 40% compared to single-model prompting.

Latency optimization: By leveraging WebAssembly for local execution, Steinberger reduced agent response time by up to 30%—critical for real-time assistants that need to feel natural.

Native integration: His background in iOS development (PSPDFKit was a mobile-first company) suggests future OpenAI agents could integrate deeply with operating system APIs—Apple Intelligence, Android's frameworks—moving beyond browser automation to true system-level control.

See also  AI Singularity Architects: Creating the First Machine Utopia

He's also bringing something less tangible but equally valuable: the hacker mindset.

Large companies like OpenAI excel at scaling what works. But they often struggle with the scrappy, move-fast-and-break-things experimentation that produces breakthrough products. Steinberger spent three months in a Madrid café building something that caught the entire industry's attention.

That's the energy OpenAI is acquiring.

What Happens to OpenClaw

OpenClaw isn't shutting down. It's transitioning to an independent foundation with OpenAI's financial support but no ownership strings attached.

The "BYOK" (Bring Your Own Key) model remains intact—developers can keep using Claude, Gemini, or local models within the OpenClaw framework. This prevents vendor lock-in and maintains the open-source ethos that made the project special.

For the community that built OpenClaw from a curiosity to a phenomenon, this represents the best possible outcome: continued development, financial sustainability, and freedom from corporate control.

The next foundation release is already in the works. Developers running OpenClaw are encouraged to update to foundation-v1 on GitHub for compatibility with the new governance structure.

The Bigger Pattern

Steinberger's move to OpenAI is part of a larger trend reshaping the AI industry.

In December 2025, NVIDIA acquired Groq for approximately $20 billion, targeting inference efficiency for agentic workflows.

In January 2026, Meta invested $14.3 billion in Scale AI, securing high-quality training data infrastructure.

In February 2026, OpenAI hired Steinberger, betting on cross-platform workflow orchestration.

The pattern is clear: hyperscalers are acquiring infrastructure that removes bottlenecks to AI deployment at scale. They're not buying products with revenue—they're buying proven technical capabilities and the people who built them.

As one analysis noted, "OpenAI didn't acquire a product with paying customers. They acquired proven infrastructure with community adoption and a clear strategic thesis: the value of AI agents lies in cross-platform orchestration, not in single-application features."

What This Means for You

If you're not a developer or AI enthusiast, you might be wondering: why does this matter?

Here's why: Steinberger's work represents the transition from AI that talks to AI that acts.

Within a year, OpenAI will likely ship personal agents influenced by OpenClaw's architecture. You'll be able to tell your phone "Handle my inbox this morning—archive newsletters, flag urgent items, draft responses to routine questions," and it will actually do it.

You'll say "Plan a weekend trip to Vancouver—find flights under $400, book a hotel near downtown, make dinner reservations Saturday night," and your AI agent will complete every step.

This isn't incremental improvement. It's a category shift in what AI can do for ordinary people.

The technical term is "agentic AI." The practical reality is: AI that stops requiring you to manage it and starts managing things for you.

The Lobster Takeover

Steinberger's journey from burned-out founder to Madrid café coder to OpenAI hire took less than a year. But the impact will extend far longer.

He's now working with Sam Altman and the OpenAI team to build what he calls "an agent that even my mum can use." That requires broader changes, more sophisticated safety mechanisms, and access to the latest models and research.

It also requires the mindset of someone who's already proven they can ship. Someone who understands that perfect is the enemy of good, that small teams move faster than large organizations, and that sometimes the best way to change the world is to just build the thing you wish existed.

"The claw is the law," Steinberger wrote—a playful motto that became OpenClaw's unofficial slogan.

For now, the law is written in San Francisco. But the code remains open-source, the community continues building, and the vision of local-first, user-controlled AI agents lives on.

Peter Steinberger made $100 million, quit everything, and disappeared. Then he came back and built something that changed how we think about AI.

That's not a bad retirement project.

Sources & References

Primary Sources:

News Coverage:

Technical Analysis:

Security & Safety:

What would you want an AI agent to handle for you? Would you trust autonomous software to manage your email, book your travel, or make decisions on your behalf? And do you agree with Steinberger's critique of European tech culture, or do you think Europe's approach to regulation and work-life balance is actually healthier?

Share your thoughts in the comments below. And if you found this story inspiring, join the iNthacity community—the Fashion City of the Web—where we explore the people and technology shaping tomorrow. Become a permanent resident, then a citizen. Like, share, and participate in the conversation.

And remember: sometimes the best thing you can do for your career is take a one-way ticket somewhere and rediscover why you loved building in the first place—the lobster will wait.

Disclaimer: This article may contain affiliate links. If you click on these links and make a purchase, we may receive a commission at no additional cost to you. Our recommendations and reviews are always independent and objective, aiming to provide you with the best information and resources.

Get Exclusive Stories, Photos, Art & Offers - Subscribe Today!

Related posts:

OpenAI Secures $200 Million US Defense Contract China’s NEW AI Robot Workforce Revolution: The World is STUNNED! 1 Million Robots by 2025? A narrow window frames a small, dimly lit bedroom bathed in soft blue light. Outside, the gleaming skyline of a futuristic city rises, punctuated by towering skyscrapers adorned with neon advertisements that flicker like stars against a dark canvas. A constant hum of flying vehicles fills the air, while distant holographic displays pulse with vibrant colors—images of the Aether Café flickering among them, drawing the viewer into a world where technology and urban life intertwine seamlessly.Beneath the Neon: A Battle for Aether Café Anker Issues Major Recall of Over 1.1 Million Power Banks Due to Fire and Burn Risks Google's Revolutionary New Quantum Chip SHOCKED THE World – 10 Million Times More Powerful!

Related Posts

You May Have Missed