On February 17, Sam Altman announced that Peter Steinberger, creator of OpenClaw, was joining OpenAI "to drive the next generation of personal agents." OpenClaw would remain open source. The next day, sources revealed that Meta had offered Steinberger more money. He chose OpenAI anyway. The same week, Raspberry Pi stock surged as much as 42% in a record two-day rally, driven by demand for single-board computers to run OpenClaw locally.
One person. Two bidders. One chose less money. A $4 single-board computer company's stock moved more than either bidder's. Something about the standard acquisition playbook doesn't apply here.
The Rebranding Arc
Steinberger's trajectory matters. He started the project as Clawdbot, rebranded it in late January to avoid confusion with a similarly named project, and relaunched as OpenClaw on January 31. Within ten days, Tencent Cloud, DigitalOcean, and Alibaba Cloud added support. Within two weeks, Google's VirusTotal was scanning its skill marketplace. Within three weeks, the founder was fielding acquisition calls from Sam Altman and Mark Zuckerberg.
The pace isn't normal. Open-source projects usually take months to attract cloud provider integrations and years to attract acquirers. Steinberger described on a podcast receiving the call from Altman personally. The speed of the courtship tells you that OpenAI and Meta aren't buying a product. They're buying a community.
Why Meta Lost
Meta offered more money. This is what Meta does. It outbid competitors for Instagram ($1 billion when the company had no revenue), for WhatsApp ($19 billion), for Oculus ($2 billion). The Meta acquisition playbook is: identify a network effect, pay above market, absorb it.
But OpenClaw isn't a network. It's an open-source agent framework. The value isn't in the code — it's in the community of developers building skills and extensions on top of it. That community chose OpenClaw partly because it wasn't controlled by a platform company. The moment Meta acquires it, the community's relationship with the project changes. Open-source developers have a finely tuned instinct for when "open" becomes "Meta open."
Steinberger seems to have understood this. By joining OpenAI instead — and keeping OpenClaw open source — he preserved the community's trust while gaining access to the models that power the framework. The community keeps building. The framework stays independent. And OpenAI gets the person who built the thing without absorbing the thing itself.
The Raspberry Pi Signal
The strangest artifact of the OpenClaw story isn't the acqui-hire. It's Raspberry Pi. The $4 single-board computer company's stock rose 42% in two days because OpenClaw runs well on cheap local hardware. Developers were buying Raspberry Pis to run AI agents at home, offline, without cloud dependencies.
This is a signal about where AI agent infrastructure is heading. The dominant assumption in the industry is that AI agents will run in the cloud, on the big providers' hardware, generating API revenue for OpenAI and Anthropic and Google. The Raspberry Pi rally suggests a parallel track: local agents running on cheap hardware, using open-source frameworks, independent of any cloud platform.
Both Meta and OpenAI want to be the platform that AI agents run on. The market is telling them that some fraction of agents will run on a $4 computer instead.
What OpenAI Actually Bought
The acqui-hire isn't about the code. OpenAI could build its own agent framework. It already has one. What OpenAI bought is legitimacy in the open-source agent ecosystem — the community's most respected builder, working inside OpenAI but maintaining the open-source project externally.
This is a different kind of platform strategy. Instead of acquiring the community and absorbing it (the Meta playbook), OpenAI is placing one of its people at the center of the community while letting the community remain independent (the Google-Android playbook). The framework stays open. Developers keep building. And the most popular agent framework in the world has its creator inside OpenAI, shaping the APIs and models that agents call.
Meta offered more money because Meta's playbook is to own things. OpenAI offered less because OpenAI's play is to orbit things. Platform gravity — the ability to shape an ecosystem without controlling it — turns out to be worth more than a bigger check.