Research as product — the frontier labs' pivot to discovery

Chat was a consumer product. Coding was an enterprise one. The next front the labs are competing on is scientific discovery itself — and mathematics is the demo reel.

With OpenAI shipping ten formalised math results and Anthropic's Fable 5 cracking an 87-year-old conjecture, research-grade AI has become the new competitive front. It reframes what a frontier lab sells: not answers about knowledge work, but acceleration of the highest-value knowledge work there is.

Why mathematics is the proving ground

Math is the most credible possible stage because it cannot be gamed — a proof either checks or it does not. When two labs post research results in the same quarter, each is proving the same underlying thing, long-horizon verifiable reasoning, on a benchmark no marketing team can fake. Discovery is where the capability contest is honest.

But the disruption lands on people. Results are outpacing peer review, unsettling how mathematics credits work and refereeing it. Selling "AI does science" as a product collides with a profession whose institutions were built for human cadence — and the collision is the friction the pivot has to work through.

The young, unrefereed frontier

The honest caveat is that accelerating discovery reliably is far harder than a handful of dramatic proofs, several still unreviewed. But as a statement of where the labs intend to compete, and where the next valuation stories will be written, the pivot to research is unambiguous.

Quanta Magazine — The AI revolution in math has arrived → · 36Kr — 2026 AI crisis for mathematicians →