Axiom Raises $200 Million Series A to Build Verified AI via Mathematics

Axiom, a San Francisco-based AI research lab, has announced a $200 million Series A funding round led by Menlo Ventures. This investment places the company’s valuation at over $1.6 billion just six months after its public launch. The new capital will support the company’s mission to establish mathematics as the core foundation for autonomous reasoning systems.

The startup is built on the conviction that mathematics provides the necessary framework for building systems that truly reason. Unlike standard language models that often hallucinate, Axiom’s technology auto-formalizes natural language into Lean, a machine-verifiable proof language. This allows the system to verify its own logic with absolute certainty.

Axiom utilizes a self-improving loop where a conjecturer model proposes new mathematical problems and a prover model attempts to solve them. By verifying results against a Lean compiler, the system operates a continuous cycle of discovery that does not rely on human data. This architecture aims to create a mathematical superintelligence that surpasses current AI limitations.

The company recently reached a significant milestone when its system, AxiomProver, achieved a perfect score on the 2025 Putnam Mathematics Competition. The AI autonomously solved all twelve problems, outperforming the highest human scores recorded during the exam. This feat demonstrated the practical power of combining transformers with formal proof assistants.

To support the wider research community, the lab has released AXLE, its proprietary engine for mathematical reasoning at scale. Developers can now access these tools through the AXLE Playground or via a Python API. This infrastructure was critical in training the models that achieved the recent Putnam success. Led by CEO Carina Hong, the Axiom team includes a deep bench of talent from both the mathematical and engineering sectors. The lab is currently deploying its technology to attack hard open problems and generate novel mathematical objects. With this new funding, Axiom intends to expand into commercial formal verification for high-stakes software and hardware.