Anthropic mathematician Levent Alpöge claimed on Friday that he used the company's Claude AI model to prove the famous 170-year-old Riemann hypothesis. The announcement came just one day after the artificial intelligence firm completed its historic public listing on the US stock market at a $1.3 trillion valuation.

Key Facts
- Levent Alpöge used a non-public version of Claude over six weeks to create a Lean-language proof of the Riemann hypothesis.
- The formal mathematical proof spans more than 250,000 pages of computer code in the Lean verification language.
- The announcement occurred one day after Anthropic went public with a $1.3 trillion market valuation.
- The Clay Mathematics Institute requires proposed solutions to undergo at least two years of community scrutiny.
Alpöge spent approximately six weeks working alongside a non-public version of Claude to construct the detailed proof. The resulting work was written in Lean, a specialized computer language designed for confirming mathematical proofs, and spans over 250,000 pages. Writing on his website, Alpöge stated that the underlying methodology can be extended to resolve both the Generalized Riemann Hypothesis and the Extended Riemann Hypothesis. He added that he and the artificial intelligence system formalized all arguments without unverified assumptions.
The timing of the release drew immediate attention because it arrived right after Anthropic's public stock market debut. Alpöge disclosed that he messaged Anthropic Chief Executive Officer Dario Amodei late on Sunday night regarding how to announce the discovery, choosing to wait until after the company completed its listing. Acknowledging that critics might view the timing cynically, Alpöge took personal responsibility for the decision, noting that he felt unable to delay publishing any longer. Anthropic has not formally commented on the release, having previously stated in August that it did not expect Claude's techniques to yield a proof.
Also read: Google Gemini 4 Carbon Tops AI Benchmarks in Cybersecurity
Verification and Industry Reaction
Prominent mathematician Terence Tao observed on social media that evaluating such a massive submission would require considerable time, BBC reported. Unlike earlier automated efforts across the tech sector, Tao noted that having a human author to consult will assist experts during the verification process. The proposed solution has not yet been accepted by the Clay Mathematics Institute, which awards the prestigious $1 million Millennium Prize for solving the problem, according to The Verge. Under institute guidelines, any candidate proof must be officially published and withstand a minimum of two years of rigorous peer examination across the global mathematical community before qualifying for the award.
The announcement follows a series of high-profile mathematical claims by major technology firms, including OpenAI's September assertion that it solved part of the fluid dynamics Navier-Stokes equations using 10,000 autonomous AI agents. That effort faced scrutiny and criticism from academic organizations, which cautioned that commercial racing to solve landmark mathematical problems risks misaligning AI capabilities with scientific integrity. A coalition of scholars signed an open letter asserting that commercial posturing around mathematical breakthroughs could prove detrimental to higher research. Despite those concerns, OpenAI later claimed progress or solutions on 372 additional complex mathematical problems.
Also read: Mass Effect Modders Warn AI Code Tools Are Ruining Game Mods
Alpöge has been an active figure in this emerging intersection of advanced AI and higher mathematics. In July, the mathematician claimed to have resolved the Jacobian conjecture using Claude, underscoring his ongoing work with automated reasoning tools. He also previously collaborated with New York University professor Tristan Buckmaster on fluid dynamics problems before OpenAI released its own controversial solution. The rapid succession of AI-assisted claims has left academic institutions working to digest a rapidly shifting landscape where machine learning models generate thousands of pages of formal proof code.
Background
Formulated in 1859 by German mathematician Bernhard Riemann, the hypothesis addresses the precise distribution of prime numbers along the number line. Because prime numbers serve as the fundamental building blocks of arithmetic and modern cryptography, cracking the puzzle carries profound implications for theoretical mathematics and computer science. It remains one of the original seven Millennium Prize problems established to highlight the most challenging open questions in mathematics. Across the quarter-century since the bounty was established, only the Poincaré conjecture has been officially solved and accepted by the Clay Mathematics Institute.
Also read: OpenAI Hits $1.25 Trillion Valuation in $11 Billion Funding
The mathematical claim coincided with Anthropic completing the largest initial public offering in stock market history, as reported by The Guardian. The startup raised $81.3 billion late Wednesday by selling 980 million shares at $83 each, surpassing SpaceX's previous record. On Thursday, trading opened with shares surging 34 percent before closing at $97 per share, placing the company's valuation at nearly $1.3 trillion. Financial filings reveal Anthropic generated $16.2 billion in revenue during the first half of 2026, though its net losses expanded to $104 billion over the first nine months due to massive computational spending on semiconductor chips.
Anthropic's debut made it the first pure-play artificial intelligence lab to enter the public markets without an accompanying advertising or space venture. Chief Executive Officer Dario Amodei rang the Nasdaq opening bell, hailing the listing as a crucial milestone that places the enterprise on stronger financial footing. The listing took place against a backdrop of wider stock market volatility and ongoing investor anxiety following management changes at major hardware suppliers like Nvidia. Despite broader economic uncertainty, enterprise adoption of Anthropic's Claude platform has continued to expand rapidly among software developers and technical institutions.
Why It Matters
If confirmed through multi-year peer review, Alpöge's proof could demonstrate that AI systems are capable of resolving humanity's most complex theoretical challenges alongside human researchers. For the digital business and technology sectors, the achievement highlights how frontier reasoning models are expanding beyond routine code generation into foundational scientific discovery.
Illustration: AI-generated image, not a photograph of the event.