Axiom says it has released the first complete computer-checked proof of the Classification of Finite Simple Groups, a result whose original proof runs to roughly 15,000 pages in more than 500 articles by about 100 mathematicians, in a post dated 26 October and announced on X on 27 October.
Over 24 days, engineers and mathematicians at Axiom Math directed swarms of dozens of AI agents in a project that produced approximately 43.1 million lines of code in the Lean proof assistant, the company wrote.
Short explainer · Axiom
A vertical news Short narrated from the published text of this piece and related FourWeekMBA pieces.
Business Pill · Multi-agent systems
An evergreen Business Engineer explainer on multi-agent systems. It explains the general model and does not discuss Axiom.
The key insight: As we read it, Axiom’s post describes a division of labour: dozens of AI agents wrote and cross-checked about 43.1 million lines of Lean in 24 days, Lean checked every deduction, and human mathematicians, including two who worked on the original classification, judged the hard cases.
The classification identifies the basic building blocks of finite groups, the mathematics of symmetry. Axiom calls it widely considered to be the longest mathematical proof ever written by humans, earning it the nickname “the enormous theorem.”
To formalize it, Axiom says it rewrote the proof in Lean, a programming language in which definitions, statements and proofs can be written precisely enough for a computer to check every step. The company says it formalized the first generation proof, because the second generation proof remains incomplete.
Axiom says it opened a Slack channel to coordinate the project on 1 October 2026 and the first swarm went live on 2 October. Pre-existing Lean work and an internally developed proof map divided the project into 108 milestones and about 380 smaller tasks.
The Classification of Finite Simple Groups: the human proof ran from Feit and Thompson’s 1963 odd-order theorem to Aschbacher and Smith’s 2004 quasithin volumes, per Axiom’s history; Axiom says its Lean formalization took 24 days in October 2026 and produced about 43.1 million lines of Lean. The 41-year span is our arithmetic.Axiom’s formalization of the Classification of Finite Simple Groups, by its own count: an original proof of roughly 15,000 pages; 108 milestones and about 380 smaller tasks; 24 days; about 43.1 million lines of Lean. Source: Axiom, 26 Oct 2026.
What Was Posted on X
The largest math proof in human history is now computer-verified.
Today, Axiom announces a complete Lean auto-formalization of the Classification of Finite Simple Groups, the “periodic table of symmetry.”
About 15,000 pages, built by over 100 mathematicians across generations.
Axiom posted that the largest math proof in human history is now computer-verified: a complete Lean auto-formalization of the Classification of Finite Simple Groups, about 15,000 pages built by over 100 mathematicians across generations.
AXIOM’S AI AGENTS VERIFIED A 15,000-PAGE MATHEMATICAL PROOF IN 24 DAYS
The project covers the Classification of Finite Simple Groups, built across hundreds of research papers over decades.
A swarm of agents divided up the work, wrote proofs, reviewed one anotherโs contributions and merged changes through GitHub.
They produced about 40 million lines of formal proof, checked by Lean.
Wall St Engine summarised the project: a swarm of agents divided up the work, wrote proofs, reviewed one another’s contributions and merged changes through GitHub, producing about 40 million lines of formal proof checked by Lean, and said Axiom’s broader goal is to bring this workflow to software and chip verification.
How the Agents Worked
AxiomProver’s agents divided the work, according to the post. Lead agents decomposed difficult arguments and reviewed the statements being formalized, while other agents searched for existing results or wrote Lean proofs. Agents claimed milestones through GitHub issues and reported completed work in Slack.
Human mathematicians investigated mathematical obstructions while engineers kept the system running, Axiom wrote. The company says it consulted group theorists who did the original work, including Richard Lyons and Ronald Solomon.
Early progress exposed weaknesses in the plan, the post says: some milestones concealed enormous prerequisites that agents had assumed rather than proved, and Axiom deleted tens of thousands of lines built on those assumptions.
What the Agents Found
Formalizing the proof exposed problems in the published literature, Axiom says. Most were minor, such as typographical errors, errors in tables or dropped hypotheses, but a few required real mathematics.
In Aschbacher and Smith’s quasithin volumes, the agents found a counterexample to a lemma involving the group SL₂(5), according to the post. In John Walter’s 1986 treatment of the Unbalanced Group Theorem, they found explicit counterexamples to a closure assertion, which Axiom says left a key inference unjustified without refuting the theorem.
Axiom says a new preservation result proved by AxiomProver closed that gap, and that it is published in an accompanying paper along with a shorter route its mathematicians found afterwards.
Connecting the Coverage
It is the latest in a run of AI mathematics claims this week. On 26 October we reported that OpenAI says its AI refuted the Exponential Time Hypothesis, and that an advisory group of mathematicians has since called on OpenAI to release the outputs behind its Yang–Mills claims.
As we read it, the post is candid about where the agents went wrong: some milestones hid enormous prerequisites the agents had assumed rather than proved, and Axiom says it deleted tens of thousands of lines built on those assumptions.
As we read it, the post also shows formalization doing more than transcription: the agents found a counterexample to a lemma in Aschbacher and Smith’s quasithin volumes and counterexamples to a closure assertion in Walter’s 1986 treatment, which Axiom says left a key inference unjustified.
As we read it, Axiom draws the line on what the machine settles: Lean checks the deductions, while humans inspect the final formal statement to confirm it expresses the classification.
Michael Aschbacher, 2005, as quoted by Axiom
“it is impossible to write out a very long and complicated argument without error”
Three Implications
FOR MATHEMATICIANS Axiom says the formalization gives the classification’s dependence on decades of papers an explicit, checkable form, with a dependency graph and the full proof released.
FOR AI LABS Axiom says agents claimed milestones through GitHub issues and reported in Slack, with lead agents decomposing arguments and reviewing statements while others wrote Lean proofs.
FOR VERIFICATION Axiom says it will keep curating and resolving the errata with an advisory board of group theorists, including Richard Lyons and Ronald Solomon.
The framework puts it this way: “Consolidation wins where work is digital, modular, and verifiable — software first, which is precisely the leading edge in both reports.”
As we read it, a Lean formalization is that kind of work: Axiom split it into 108 milestones and about 380 tasks and had every step checked by the proof assistant. The framework was written before this announcement and does not discuss Axiom.
What Is Not Established
The figures and the account of how the work was done come from Axiom’s own post. Axiom says humans inspect the final formal statement to confirm it expresses the classification while Lean checks the deductions; we did not see an independent review of the formal statement, and the post does not state the cost or compute and describes the models only as a “changing mix of models.”
This Week In The AI Supercycle
Execution Fault Lines, Concentrated Demand, and Geopolitical Firewalls
This week’s issue: the AI build is not short of demand for its parts. It is short of cheap money, and its end revenue rests on a narrow base of large, often venture-funded buyers.
Axiom says dozens of AI agents, directed by its engineers and mathematicians, turned a roughly 15,000-page proof into about 43.1 million lines of Lean in 24 days, and found and repaired gaps in the published arguments along the way.
A note on sourcing. All facts about the formalization are from Axiom’s own post, ‘Formalizing the Enormous Theorem’, dated 26 October 2026, which we read in full, and its 27 October post on X. We did not see an independent review of the formal statement. This is not investment advice.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
