5 AI-Generated Math Papers Accepted! Post-00s Founder Hong Letong Raises $2 Billion
Axiom Math, founded by Chinese post-00s entrepreneur Hong Letong, has had 5 out of 8 AI-generated math papers accepted in peer-reviewed journals. The company raised $2 billion in March, achieving a $16 billion valuation.
Article intelligence
Key points
- Five of eight math papers generated by Axiom Math's AI system, AxiomProver, have been accepted by academic journals.
- Founder Hong Letong dropped out of Stanford to start the company, which secured $2 billion in funding and is valued at $16 billion.
- AxiomProver uses a pipeline of generation, formal proof, and verification to ensure mathematical correctness.
- The company aims to build a self-improving superintelligent reasoner and has already validated a theorem in game theory.
Why it matters
This matters because five of eight math papers generated by Axiom Math's AI system, AxiomProver, have been accepted by academic journals.
Technical impact
May affect model selection, inference cost, product capability, and evaluation benchmarks.
Axiom Math, a startup founded by the 23-year-old Chinese prodigy Hong Letong, has achieved a major milestone: five of eight mathematics papers generated or formally verified by its AI system, AxiomProver, have been accepted by peer-reviewed journals. The company announced the results on May 28, 2026, exciting the AI and mathematics communities.
The accepted papers cover a range of fields including number theory, combinatorics, commutative algebra, algebraic geometry, representation theory, and Dyck path models. One paper, "Reciprocals of Partition Polynomials," which was accepted by the Annals of the Romanian Academy of Sciences, solved six out of ten conjectures posed by Ballantine, Beck, Feigon, and Maurischat—and even found a counterexample to one original proposition.
What sets AxiomProver apart is its approach: instead of generating natural language proofs that may contain hidden logical gaps, it produces formal proofs in the Lean theorem prover. A separate verification system checks every step. Human mathematicians then translate the formal proof into readable academic prose and handle communication with journal editors and reviewers. This human-AI collaboration proved effective: in some cases, the AI generated a fully verified proof within 24 hours of being given an open research problem.
Hong Letong, born in Guangzhou in 2001, has a dazzling academic record. She earned dual bachelor's degrees in mathematics and physics from MIT in just three years, won the Rhodes Scholarship and the Morgan Prize—the highest undergraduate honors in North American math—and then pursued a master's in neuroscience at Oxford before entering Stanford's combined PhD/JD program. She dropped out of Stanford in fall 2024 to focus on Axiom Math, co-founding the company with Shubho Sengupta, formerly of Meta. Ken Ono, a renowned mathematician, left his tenured position at the University of Virginia to join as the founding mathematician.
Axiom Math has quickly attracted huge investment. After a $64 million seed round, the company closed a $2 billion Series A in March 2026, reaching a valuation of $16 billion. Investor Matt Kraning highlighted the importance of verifiable AI: "AI will write all the code, but mathematics will prove whether it works."
The company's vision extends beyond mathematics. It aims to build a superintelligent reasoning engine that can be applied to other scientific domains and high-stakes decision-making. Indeed, a paper submitted on May 27 ventures into game theory and economics, formalizing a classic theorem by Robert Aumann in collaboration with Harvard Business School professor Scott Duke Kominers.
Hong Letong once said that entrepreneurs should choose the hardest problems, ones that might take five to ten years to solve. With Axiom Math, she is tackling one of the most fundamental challenges in AI: ensuring that machine-generated outputs are not just plausible but provably correct.