Anthropic (ANT.US) nearly suffered a "broken IPO," yet Claude is knocking on the door of the "holiest mountain in mathematics"! 17 million lines of Lean code produce a candidate proof of the Riemann Hypothesis.

date
10:59 21/11/2026
avatar
GMT Eight
At 11:20 AM Eastern Time on November 20, Anthropic fell as low as US$83.05 intraday on the second day after its listing, just 5 cents away from its US$83 issue price, and finally closed at US$88.91, down 8.50%. On the same day, the company's mathematician Levent Alpoge announced on his personal website that he and Claude had found a candidate proof of the Riemann Hypothesis, which has troubled the mathematics community for 167 years, and publicly released about 17 million lines of Lean formalized code.
At 5:05 PM Beijing time on November 21, MIT number theorist Andrew Sutherland's publicly posted verification log showed that all 175,187 modules had been checked item by item through the Lean kernel, and that two independently written verification kernels had both accepted the fully exported proof. The five core theorems rely only on Lean's three standard axioms, and independent comparison also confirmed that they match the pre-listed target propositions. Among them, the classical Riemann Hypothesis directly adopts the definition already available in the public mathematics library Mathlib. This significantly strengthens the credibility of the formal proof. What mathematicians need to explain next is which new ideas make the proof work, and which methods can be applied to other difficult problems. The Clay Mathematics Institute still lists the Riemann Hypothesis as an unsolved problem, and its million-dollar prize will be considered only after the result has been published in a qualified channel for at least two years and has gained broad recognition from the international mathematics community. According to the verification log published by MIT number theorist Andrew Sutherland, two independently written verification kernels both accepted the complete proof, and the formulation of the classical Riemann Hypothesis is also consistent with the existing definition in the public mathematics library. For investors focused on the AI industry chain, the most noteworthy aspect of this historic advanceAlpge and Claude finding a candidate proof of the Riemann Hypothesis that has troubled the mathematics world for 167 yearsis that Claude can, under expert guidance, propose key derivations and deliver results that can be independently checked by third parties. How much valuation support Anthropic's scientific research capability can provide depends further on whether this performance can be reproduced in other high-value R&D tasks, and whether the computing power and expert hours required for each success can make customers willing to pay while also enabling the company to continue achieving strong profitability. Anthropic researcher says he found a proof of the Riemann Hypothesis together with Claude Following Anthropic's record-breaking initial public offering, one of the company's researchers said he believes that, with the help of the company's artificial intelligence software, he has solved a famous number theory problem that has troubled mathematicians for more than 150 years. Mathematician Levent Alpge posted an explanation on his personal website on Friday, saying he had found a proof of the Riemann Hypothesis. He shared links to code written in the Lean programming language and said the code provides evidence for his conclusion. According to his explanation, Alpge sent a message to Anthropic CEO Dario Amodei on Sunday night asking whether he could announce the research results to the public. Alpge described the timing of the announcement as "terrible," but said: "If I waited any longer, I really couldn't stand it." Anthropic declined to comment, and Alpge did not respond to requests for comment. The Riemann Hypothesis concerns the distribution of prime numbers and has long been regarded as one of the hardest problems in mathematics. For years, OpenAI and Anthropic have sought to demonstrate the capabilities of their most advanced models by tackling difficult problems in mathematics and the life sciences. If Alpge's result is confirmed, the Riemann Hypothesis will become the second Millennium Prize Problem that may have been solved with the help of AI. In September this year, OpenAI said its model had produced a solution to the 90-year-old NavierStokes problem. Andrew Sutherland, a senior research scientist at MIT, said he verified Alpge's work on the Riemann Hypothesis on his own computer and also saw at least three verifications completed by other professional mathematicians on different systems. Sutherland said Alpge and Claude's research built on a great deal of prior work, including generations of mathematicians' research on the Riemann Hypothesis as well as accumulated progress in the field of mathematical formalization. Speaking about the future, Sutherland said: "I am both very nervous and very excited." He explained that mathematicians will have many years of work ahead, requiring in-depth study of the proof and clarification of how it can be applied. "This is a day worth celebrating," Sutherland said. He also acknowledged that some in the mathematics community worry that AI may disrupt mathematicians' careers and other jobs. Noah Giansiracusa, an associate professor of mathematics at Bentley University and an affiliated scholar at Harvard University, added that if the solution holds, it would be a big deal for AI. "It sounds like this proof is indeed very likely to be rigorous, airtight, and fully valid," Giansiracusa said. Alpge described the experience of attempting to prove the conjecture as "a stream-of-consciousness wander." He wrote: "Over the past few months, Claude and I have explored such a vast mathematical landscape together, and we enjoyed it immensely. But this time, near the finish line, Claude left me far behind." From prime distribution to scientific AI pricing, what is truly scarce is verifiable discovery The Riemann Hypothesis asserts that all nontrivial zeros of the Riemann zeta function lie on the straight line where the real part equals . It matters because these zeros determine how far the actual prime count can deviate from its smooth estimate. The Riemann zeta function, namely the Riemann zeta function. In August this year, Anthropic reported that it had raised the lower bound on the proportion of zeros provably lying on this critical line from 41.6% to 67.2%. This time, Alpge described another route. He directed the research toward Stepanov's classical method over function fields, while Claude proposed stringing together about \(\sqrt X\) theta lifts for the von Mangoldt weighted sum with summation length \(X\), so that each lift simulates the role of one "multiplicity" in the classical proof. According to the author's account, Claude's most original contribution was proposing this step. Terence Tao preliminarily judged that the argument may use modern results from the Langlands program and arithmetic geometry, and that it will still take time to clarify the full picture. Competition in scientific AI therefore now has a more concrete yardstick. Models can try many proof routes, Lean can verify the final derivation step by step, and researchers must judge whether the proposition is valuable and which lemmas are worth reusing. For investors, the more explanatory metric is how much inference compute and expert labor are required in total for each independently verified result that can be used by customers. The full discovery cost of this research has not yet been disclosed. Open-source routes are also competing for the same pool of scientific research users. Caltech professor Sergei Gukov and his students trained specialized mathematical models from scratch; some models excel at specific problems, but cross-domain capability remains limited. Open Math OS, promoted by Terence Tao, integrates open-weight models, cloud computing, and verification tools; research files and conversations are private by default and are used for training only with researchers' explicit consent. For laboratories with unpublished results, whether models can solve problems, whether proofs can be independently reviewed, and who controls research data will jointly determine purchasing and renewal decisions.