Research-grade AI becomes the new competitive front as labs race to own scientific discovery
With OpenAI shipping ten formalised math results and Anthropic's Fable 5 cracking an 87-year-old conjecture, the frontier labs' competition is moving from chat and coding into research itself. Owning scientific discovery — not just answering questions about it — is emerging as the next strategic front, and the next product.
The pivot reframes what a frontier lab sells. Chat was a consumer product, coding an enterprise one; research assistance targets the highest-value knowledge work there is — and a lab that can credibly accelerate discovery in mathematics, and next in biology and materials, is selling something no incumbent software company offers. The math results are the demonstration reel for that pitch.
The competitive dynamic is a capability arms race with academic scoreboards. When OpenAI's ten results and Anthropic's conjecture disproof land in the same quarter, each lab is proving the same underlying thing — long-horizon verifiable reasoning — on the most credible possible stage. Research mathematics has become a benchmark the labs compete on precisely because it cannot be gamed.
The honest caveat is that 'AI does science' is a young and partly-unrefereed claim. Several headline results still await specialist review, and accelerating discovery reliably across a field is far harder than a handful of dramatic proofs. But as a statement of where the labs intend to compete, the pivot to research is unambiguous — and it is where the next round of valuation stories will be written.
Quanta Magazine — The AI revolution in math has arrived → · 36Kr — 2026 AI crisis for mathematicians: key impacts and challenges →