OpenClaw Creator Joins OpenAI: Job, Foundation and Independence

A job, not an acquisition

Steinberger announced on his own blog on February 15, 2026, that he was joining OpenAI to work on bringing agents to mainstream users. Sam Altman announced the hire the same day, putting it as Steinberger joining to drive the next generation of personal agents. [1][2]

Steinberger described the decision as a choice of speed over ownership: "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." [2]

Atlas interpretation: What OpenAI acquired was Steinberger, not OpenClaw. The company hired the person who had just shown, faster than any of its own product teams, what a personal agent people actually wanted to run looked like, and left the thing he built to keep running without it. [1]

Why a foundation instead of a fork

OpenClaw itself did not go to OpenAI. Steinberger said it would move to a foundation and stay open and independent, and OpenAI committed to sponsoring the project rather than taking it over. [1]

By the time of the announcement, OpenClaw had already been renamed twice, first from Clawdbot after Anthropic objected to the resemblance to Claude, then to OpenClaw once Steinberger settled on the name he preferred. [2]

Atlas interpretation: A model-agnostic personal agent is worth less to OpenAI if it becomes an OpenAI product, since its value to users came from running whichever model or subscription they already had. Sponsoring the foundation buys goodwill and a seat near the project without narrowing what it does, which is a different trade than the acquisitions OpenAI made of tools it wanted to own outright. [1]

The growth that made this worth announcing

OpenClaw had shipped as Clawdbot less than three months earlier, in November 2025. In the weeks around this announcement it was on its way to overtaking React's GitHub star count, a position React had held for years, reaching the top spot in under four months from a standing start. [3]

Atlas interpretation: A star count is a weak proxy for usage, but the rate of it here is the point: no comparably fast climb to the top of that list had happened before. That is the leverage Steinberger brought to the negotiation, and it is why OpenAI's interest was in the person who built the thing rather than in owning the repository itself. [3]