Sam Altman admitted defeat to Google

OpenAI CEO Sam Altman sent a memo to his researchers that sounds like an admission of defeat. “Google has done excellent work across all areas lately,” he wrote. And this is not just a compliment to a competitor.

The OpenAI CEO warned the team about “temporary economic difficulties” after the release of Gemini 3 Pro. “The atmosphere will be tense,” he stated directly. And here’s what’s particularly telling: at OpenAI they believed that pre-training of models had reached its limit. So they focused on “reasoning” models. And Google simply continued to improve the basic technology. And overtook them.

Now Altman promises to catch up. He announced a new language model under the code name Shallotpeat, which will be used to correct pre-training errors. And calls on the team to focus on “very ambitious tasks,” even if OpenAI “temporarily falls behind competitors”. Most of the research group should work on achieving “superintelligence”. “Under short-term pressure from a competitor, we need not to forget the main thing,” he emphasizes.

It turns out the OpenAI CEO acknowledges the lag and promises to “catch up quickly”. But at the same time wants to focus on distant goals, risking falling even further behind here and now. OpenAI’s aura of invincibility was destroyed by a powerful competitor and falling demand from corporate clients. Now they will have to make “bold strategic decisions”. There’s only 1 question: isn’t it too late?