← Back to archive

OpenAI, Navier-Stokes, and the cost of huge AI agent swarms

9 September 2026 · 46 views · 0 likes · 0 reposts · 127 blog views

AI neuroscience product research

OpenAI has solved the Navier–Stokes Millennium Prize Problem. It was one of the famous $1M problems that had remained unsolved for almost 90 years!

But... with OpenAI, there is always some drama.

For a customer, running a similar swarm of 10,000 AI agents would reportedly cost around $15M (15 times more expensive than the price).

It wasn’t strictly an AI-only achievement. Two mathematicians were already behind closely related work: Tristan Buckmaster and Levent Alpöge, who works at Anthropic.

Buckmaster says OpenAI learned that they were close and then threw massive compute at the problem.

Both mathematicians had also been using OpenAI tools in their research. So Buckmaster asked the obvious question: did the model learn something from our private research?

That’s concerning, right? But what’s even more concerning is when a big lab employee says he’s resigning - and both companies don’t act responsibly - while self-improving superintelligence is gambling with our lives.

OpenAI, Navier-Stokes, and the cost of huge AI agent swarms

Originally posted on X

PS: I'm pretty active on 𝕏 if you'd like to follow along

Related posts

Archive

2026

2025

2024

2018