OpenClaw Plans Hang in the Balance as Peter Steinberger Weighs Meta and OpenAI Buyouts

Austrian developer Peter Steinberger never expected his side project to ignite a movement.

What began as a rapidly built, self-modifying AI assistant has morphed into OpenClaw—one of the fastest-growing open-source AI agents in history. With more than 180,000 GitHub stars amassed in record time, the project has caught the attention of Silicon Valley heavyweights, including Mark Zuckerberg, Sam Altman, and even Satya Nadella.

Now, Steinberger faces a pivotal decision: accept billion-dollar acquisition interest or preserve the open-source principles that fueled OpenClaw’s explosive rise.

Billion-Dollar Interest, Open-Source Conditions

Speaking on the Lex Fridman Podcast, Steinberger revealed that both Meta and OpenAI have made serious overtures.

Zuckerberg reportedly contacted him directly via WhatsApp, where the two debated model performance and agent capabilities. Altman’s proposal came with a different hook—access to significant compute resources, potentially supercharged by infrastructure partnerships that could dramatically accelerate OpenClaw’s performance.

But Steinberger’s stance is firm: any deal must preserve the project’s open-source DNA.

He floated the possibility of a dual structure similar to Google’s Chrome and Chromium model, where a corporate layer coexists with a fully open foundation. “This is too important to just give to a company and make it theirs,” he said.

Building at a Loss

Despite its popularity, OpenClaw isn’t profitable. Steinberger estimates he’s losing between $10,000 and $20,000 per month to keep the project running. Sponsorship revenue goes directly toward infrastructure and dependencies rather than into his own pocket.

Having previously sold his company PSPDFKit, financial security isn’t his primary driver. Instead, he appears more focused on shaping what he believes is the next paradigm of software: autonomous agents capable of replacing traditional apps.

The 24-Hour Nightmare

OpenClaw’s rise hasn’t been smooth.

Originally launched under a different name, the project was forced into a rebrand after a trademark complaint tied to Anthropic’s Claude model. What followed was chaos.

Within seconds of updating accounts, crypto scammers hijacked his digital footprint—serving malware from GitHub repositories, compromising NPM packages, and flooding his social feeds with spam. The coordinated attack nearly pushed him to delete the project entirely.

The second rebrand required extreme operational secrecy—simultaneous account changes, decoy names, and tightly timed rollouts—to prevent another exploit. Steinberger later described it as the worst online harassment he had ever experienced.

“Agentic Engineering” Over Apps

Steinberger prefers the term “agentic engineering” over “vibe coding,” a label he sees as dismissive. His workflow reflects that philosophy: running multiple AI agents simultaneously and generating thousands of commits in a single month by interacting conversationally with models instead of manually coding.

He believes this shift will fundamentally reshape the software economy.

In his view, standalone apps are becoming obsolete. If an AI agent already understands a user’s calendar, preferences, health data, and habits, why open a separate fitness, food delivery, or scheduling app? The agent can simply call APIs directly.

“Every app is just a very slow API now,” he argued.

The Fork in the Road

Steinberger is weighing several paths: accept an acquisition under strict open-source conditions, launch a venture-backed company around OpenClaw, or continue independently—even if it means operating at a loss.

Each option carries trade-offs. Corporate backing would provide compute scale and financial stability. Independence would preserve full autonomy but extend the financial burn.

For now, he appears confident in one thing: whichever route he chooses, OpenClaw has already changed the trajectory of AI development.

“I can’t go wrong,” he said.