OpenAI just claimed a proof of Navier–Stokes — a problem open since the 1930s — using about 10,000 AI agents running for 88 hours. So I asked the obvious question: if agents can do that, what happens when you point a million of them at SHA-256 and try to break Bitcoin? The answer is nothing, and the reason is worth understanding. Navier–Stokes was a search for an idea, and the winning proof is only a few dozen pages long — throw enough agents at it and one of them finds the insight. Brute-forcing a hash is a search through a list of 2²⁵⁶ possibilities with no structure and no shortcuts, and every time you double your fleet you gain exactly one bit out of 256. A million agents buys you 20 bits. A hundred million buys 27. Run a hundred million agents flat out from the Big Bang until today and you'd reach 2¹¹⁸ — still a thousand times short of the easier target. And past a certain point it stops being an engineering problem and becomes a physics one: just counting to 2²⁵⁶, ignoring the hashing entirely, would take more energy than ten trillion Suns produce in their lifetimes. Bitcoin mining is already the largest brute-force attack in history, running 24/7 for seventeen years, and it's about 10⁴⁸ short.
Here’s a link I created on Claude that breaks it all down

Claude Artifact
Try out Artifacts created by Claude users