Axiom Math

Axiom Math is an American artificial intelligence company founded in 2025 by Carina Hong, a Rhodes Scholar, MIT graduate, Morgan Prize winner, and former Stanford JD/PhD candidate in mathematics. The company is building formal-mathematics AI systems that produce machine-checkable proofs in the Lean programming language, with the goal of solving the verification problem that limits the trustworthiness of language-model mathematical output. In December 2025 Axiom's AxiomProver scored a perfect 12 out of 12 on the William Lowell Putnam Competition, the principal undergraduate-mathematics competition in North America, a result that out of more than 150,000 attempts in the competition's 98-year history has been achieved by only five human contestants. In March 2026 the company closed a Series A of $200 million at a $1.6 billion post-money valuation, led by Menlo Ventures.

At a glance

• Founded: 2025 by Carina Hong.
• Status: Private. Series A in March 2026 at $1.6 billion post-money valuation.
• Funding: Approximately $264 million-plus cumulative private capital. Seed of $64 million in 2024 at $300 million valuation led by B Capital with Greycroft, Madrona, and Menlo Ventures participating. Series A of $200 million in March 2026 at $1.6 billion post-money valuation led by Menlo Ventures.
• CEO: Carina Hong, Founder and Chief Executive Officer. Rhodes Scholar; MIT graduate (dual degrees in mathematics and physics in three years); Morgan Prize winner; Schafer Prize winner; nine peer-reviewed publications. Former Stanford JD/PhD candidate in mathematics (Knight-Hennessy Scholarship).
• Other notable leadership: Ken Ono, Founding Mathematician (joined December 2025). Senior number-theorist; University of Virginia STEM Adviser.
• Open weights: Limited. Axiom Math develops commercial formal-mathematics AI systems rather than open-weights model releases.
• Flagship outputs: AxiomProver, the formal-mathematics AI system; perfect 12/12 score on the December 2025 Putnam Competition; automated proof of a 20-year-old open number-theory conjecture.

Origins

Axiom Math was founded in 2025 by Carina Hong, a 24-year-old Rhodes Scholar and MIT graduate who completed dual degrees in mathematics and physics at MIT in three years and won the Morgan Prize and the Schafer Prize during her undergraduate career. Hong left a Stanford JD/PhD program in mathematics (Knight-Hennessy Scholarship) in her first year to start the company. The founding research thesis was that formal mathematics in the Lean programming language could solve the verification problem that limits the trustworthiness of language-model mathematical output, with each step of an AxiomProver-generated proof being machine-checkable rather than relying on the model's word.

The company emerged from stealth in 2024 with a seed financing of $64 million at a $300 million valuation led by B Capital, with Greycroft, Madrona, and Menlo Ventures participating. The seed round size and pre-product valuation were unusual for a research-stage AI company, reflecting the founder's research credibility and the strategic interest in formal-mathematics-AI as a category.

In December 2025 two events brought Axiom Math into wider visibility. First, the company's AxiomProver system scored a perfect 12 out of 12 on the Putnam Competition, the principal undergraduate-mathematics competition in North America. The Putnam is graded out of 120 points and has produced only five perfect 120-out-of-120 scores from human contestants in 98 years across more than 150,000 attempted exams, but Axiom benchmarked AxiomProver against the Putnam's 12 problems, where it solved all 12 with formally-verified Lean proofs. Second, the same AxiomProver system automatically proved a 20-year-old open number-theory conjecture that Ken Ono, one of the world's senior number-theorists, had been unable to solve himself. Ono joined Axiom Math as Founding Mathematician at the same time.

The March 2026 Series A of $200 million at a $1.6 billion post-money valuation, led by Menlo Ventures, completed the company's first full year out of stealth. The valuation jump from the $300 million seed to the $1.6 billion Series A reflected the AxiomProver Putnam result, the Ono conjecture, and the broader strategic interest in mathematical-superintelligence-class AI systems.

Mission and strategy

Axiom Math's stated mission is to build mathematical superintelligence: AI systems that can autonomously produce, verify, and discover mathematical knowledge with machine-checkable rigor. The strategic premise is that formal mathematics in Lean provides a substrate where each step of reasoning is verifiable, and that the trust gap between language-model mathematical output and rigorous mathematical proof has been the principal blocker for deploying AI in high-assurance technical domains.

The strategy combines two threads. First, the AxiomProver system, with the Putnam result and the Ono conjecture as the principal capability demonstrations and continued formal-mathematics-AI research as the public-research surface. Second, applied deployment in adjacent verification-critical domains, with stated ambitions including code verification (proving software correctness) and adjacent high-assurance technical applications.

The competitive premise positions Axiom Math within the formal-mathematics-AI insurgent cohort alongside Harmonic, with both companies pursuing mathematical-superintelligence-class research while differentiating on technical approach and applied-deployment focus. Axiom's emphasis on the Lean substrate and the AxiomProver verification-first approach is a distinctive technical positioning relative to the language-model-style approaches that dominate the broader AI-for-math segment.

Models and products

• AxiomProver. Formal-mathematics AI system that produces machine-checkable proofs in the Lean programming language. Scored 12/12 on the December 2025 Putnam Competition and automatically proved a 20-year-old open number-theory conjecture.
• Lean-substrate research and tooling. Continued formal-mathematics-AI research output and adjacent infrastructure.
• Applied verification deployment. Stated forward direction including code verification and adjacent high-assurance technical applications.

Distribution channels have been comparatively under-disclosed publicly, with the principal public references being the AxiomProver capability demonstrations and the broader research-and-recruitment surface.

Benchmarks and standing

Axiom Math's evaluation framework focuses on formal-mathematics capability metrics rather than horizontal foundation-model leaderboards. The principal disclosed benchmarks include:

• Putnam Competition (December 2025): AxiomProver solved 12 out of 12 problems with formally-verified Lean proofs. Out of more than 150,000 human contestants in 98 years, only five have achieved perfect scores on the Putnam.
• Open number-theory conjecture (December 2025): AxiomProver automatically proved a 20-year-old open number-theory conjecture that Ken Ono had been unable to solve himself.

Industry coverage has consistently grouped Axiom Math with Harmonic as the two principal mathematical-superintelligence-class AI insurgents of the 2024 to 2026 cohort. The Putnam result and the Ono conjecture provided unusually concrete capability evidence relative to the demonstration-and-research-claim-only positioning common in the broader AI-for-science segment.

Leadership

As of May 2026, Axiom Math's senior leadership includes:

• Carina Hong, Founder and Chief Executive Officer. Rhodes Scholar; MIT graduate; Morgan Prize and Schafer Prize winner.
• Ken Ono, Founding Mathematician. Senior number-theorist; University of Virginia STEM Adviser.
• Senior research and engineering leadership across the AxiomProver and Lean-substrate programs.

The Hong-and-Ono leadership combination provides distinctive mathematical-research credibility that few peer AI-for-mathematics labs have matched.

Funding and backers

• Seed (2024): $64 million at $300 million valuation. Led by B Capital with Greycroft, Madrona, and Menlo Ventures.
• Series A (March 2026): $200 million at $1.6 billion post-money valuation. Led by Menlo Ventures.

Cumulative disclosed private capital approximately $264 million-plus.

Industry position

Axiom Math occupies a distinctive position as one of the two principal mathematical-superintelligence-class AI insurgents alongside Harmonic, with the December 2025 Putnam and Ono-conjecture capability demonstrations providing unusually concrete evidence relative to the demonstration-and-research-claim-only positioning common in the broader AI-for-science segment. The Hong-and-Ono leadership combination provides distinctive mathematical-research credibility, and the Series A unicorn round at $1.6 billion post-money valuation establishes the company in the top tier of AI-for-mathematics insurgents.

The structural risks are characteristic of formal-mathematics AI companies. First, the conversion from research demonstrations to commercially valuable applied deployment is the principal commercial question: the Putnam result and the Ono conjecture are unambiguous research wins, but whether the AxiomProver capabilities translate into revenue-generating verification deployments is unproven. Second, the formal-mathematics-AI category is narrower than the horizontal language-model category, and the absolute size of the addressable market for verification-first AI is uncertain. Third, the competitive dynamics with Harmonic, with frontier-lab AI-for-math research programs, and with the broader language-model-style AI-for-math approaches will shape the segment over the next several years.

Competitive landscape

• Harmonic. Direct mathematical-superintelligence-class AI peer with overlapping mission and a different technical approach (Aristotle model with International Mathematical Olympiad result).
• OpenAI, Anthropic, Google DeepMind AI-for-math research. Frontier-lab AI-for-math programs including AlphaProof, Gemini-with-Math, and adjacent research output.
• Lean community and academic formal-mathematics research. Carnegie Mellon, Stanford, MIT CSAIL, and broader academic formal-mathematics research peers.
• Code-verification AI peers. Adjacent applied-verification AI companies.

Outlook

• Continued AxiomProver capability disclosures and benchmark results beyond the December 2025 Putnam and Ono-conjecture results.
• The pace of applied-verification deployments in code verification and adjacent high-assurance technical applications.
• The competitive dynamics with Harmonic as both companies progress through 2026 to 2027.
• Continued senior mathematical-research recruiting against academic formal-mathematics research programs.
• The Series B or adjacent fundraising timeline beyond the recent unicorn round.

Sources

• Axiom Math official site. Company reference.
• Menlo Ventures: AI Will Write All the Code. Mathematics Will Prove It Works. Menlo investment-thesis reference.
• Madrona: Axiom Math Tackles One of AI's Hardest Problems. Madrona investment-thesis reference.
• Thought Economics: Superintelligent AI Mathematician. Founder interview reference.
• B Capital: Toward Mathematical Superintelligence. B Capital investment-thesis reference.
• Tech Funding News: Axiom Math AI mathematician $64M seed. Seed-round and Series A coverage.
• The Standard: Carina Hong's AI startup hits $1.6b valuation. Series A reference.