Axiom Math Formalizes Mathematics to Stop AI from Producing Poor Results

Axiom Math has formally verified one of the largest collections of mathematics ever created by humans, converting the work into computer-checked code.

The startup, valued at $1.6 billion, announced it had verified the Stacks Project, a 20-year-old open source reference that mathematicians worldwide depend on. Columbia mathematician Peter Woit called it rare and compared it to the famous Bourbaki books, which rebuilt mathematics from its foundations in the 20th century.

The Stacks Project began in 2005 as an open source effort to document mathematical results from basic algebra to advanced ideas in algebra, geometry, and number theory. Over two decades, mathematicians have contributed to this 7,654-page library.

Axiom's team translated the Stacks Project into approximately 16 million lines of code written in a programming language called Lean. A computer program called a proof assistant then checked each line to confirm the logic was sound.

This process, called formal verification, is the same approach engineers use to test chip designs.

The challenge lay in maintaining every connection between the mathematics. The Stacks Project is built like a family tree, where later theorems depend on earlier ones.

If one proof changed, many others could break. Axiom's team had to keep all 16 million lines of code connected without contradiction.

To manage this complexity, Axiom gave its AI agents a loose structure and allowed them to work in parallel, finding their own tasks instead of following strict orders. The CEO, Carina Hong, explained that the team wanted agents with "taste"—good judgment about mathematics.

The company trained its own mathematical models rather than relying on general-purpose AI.

This work fills a major gap in mathematics. Algebraic geometry, an advanced field, had almost no formally verified work available before now.

Researchers can now build new proofs on this verified foundation instead of starting from scratch.

Hong co-founded Axiom after leaving a Stanford mathematics PhD program. The company raised $200 million in Series A funding in March 2026.

She believes verified mathematics will become the foundation for trustworthy AI-written software. "If you slop drop in a way that's bad engineering choices in the beginning, then it's very hard for them to grow that boundary," she said.