Tim O'Reilly: Open AI beats the big labs

Tim O'Reilly built the empire that fed his name into the industry — the conferences, the books, the press. Now AI is eating that business. He's not mad about it. He's mad about who's feeding it.

The big AI labs — OpenAI, Google, Anthropic — trained their models on the same copyrighted material O'Reilly spent decades curating. Now those same models are cannibalizing the market. His fix is simple: the models need to stay open source, not locked behind a walled API. When the weights are public, the community can audit them, fork them, and build on top without paying rent to a single company.

It's a familiar O'Reilly move — push the platform out into the commons so the little guys have a fighting chance. For the small shops, the indie devs, the side hustlers who actually need tools to compete: open weights mean they're not at the mercy of whoever controls the tap.

Why this matters for us: when the big labs lock AI behind paywalls, the tools we need to run our businesses get priced out — open source is the only thing keeping the doors open for the rest of us.