Carina Hong, CEO of Axiom Math, discusses how formal verification can scale brilliance and compound intelligence, contrasting it with informal AI approaches. The company achieved a perfect score on the Putnam exam and recently raised a $200 million Series A to advance verified reasoning across math, software, and hardware.
Summarized by Podsumo
Axiom Math raised $200 million at a $1.6 billion valuation, nearly matching the entire US annual math research budget.
The company's system achieved a perfect score (120/120) on the 2025 Putnam exam, outperforming top humans and frontier LLMs.
Carina argues that formal verification is not about eliminating errors but about 'scaling brilliance' and enabling human-AI collaboration.
Axiom open-sourced mathematical discovery toolkits to help mathematicians form conjectures before proof.
The future vision includes offering a verification API for frontier labs, claiming formal verification is critical for superintelligence.
"Verification to me is not about lausiness. Verification to me is about scaling brilliance, compounding brilliance."
"We do not believe that an informal math system is going to be the mass AGI solution."
"Anything that can be specified can be proven."