OpenAI solved the Navier-Stokes equation with brute force for $40 million

The Astra-next model reached for the Millennium Prize in 88 hours

OpenAI has published a solution to one of the deepest mathematical problems. According to a report from Latent Space, the Astra-next model needed about 10,000 running agents and 130 billion tokens, costing over $40 million. During 88 hours of search, the agents discovered a singularity in the Navier-Stokes equations, meeting the conditions for winning the second Millennium Prize in history.

Math faces the same fate as chess

This is the definitive end of the era when deep mathematical proofs required decades of human intuition and elegant leaps of thought. For research institutions and universities, this means the playing field has tilted radically. Those without a budget in the tens of millions of dollars for a single computational run are dropping out of the major league of theoretical discoveries.

Brute force does not imply holistic understanding

Finding a singularity by brute-force searching the state space does not mean that AI understands the fundamental physics of gases and liquids better than we do. It found an anomaly we asked it to find. The log from the 88-hour run will likely be completely unreadable, so human verification of the proof will take years.

Peer review and the battle for credit will follow

The key will be to watch how the Clay Mathematics Institute approaches the proof. The question is not just whether the proof is formally correct, but to whom the institute will eventually award the million-dollar prize (the company, specific researchers, or if they will block it, stating it can only be awarded to a human).

Lilith's verdict

Math has become a competition over who can afford the most expensive server farm in the world.