
Anthropic Soars Then Slides on Nasdaq Debut as ANT Opens at $111.50 and Closes at $97.17
Anthropic opened at $111.50 on its Nasdaq debut, briefly touching a $2 trillion value, then slid to close at $97.17, still 17% above its $83 IPO price.
By Toijam Bimoljit Singh • 21 November 2026 • 7 min read
Photo: A.Savin, Free Art License, via Wikimedia Commons — the Aula of the University of Goettingen, where Bernhard Riemann was professor, 27 June 2022. The photo is from an earlier date, not the announcement itself.
TL;DR — the direct answer: Levent Alpoge, a mathematician at Anthropic, posted a four-page note on his personal website on Friday, 20 November, saying he believes he and Claude discovered a proof of the Riemann Hypothesis, along with the Generalized Riemann Hypothesis. He released a Lean formalization of the proof, more than 17 million lines of code in a 14-part archive, but no traditional paper, no preprint and no journal submission. Within hours, Andrew Sutherland, a mathematician at the Massachusetts Institute of Technology, independently compiled the Lean code and said the chance that anything is fishy is approximately zero. But the claim is not yet verified: a final comparator run to confirm the code really states the standard theorem was still awaited on Friday afternoon. The same day brought a second, entirely human claim: a draft by Stanford postdoc Alessandro Masullo, shared by American Mathematical Society president Ravi Vakil, that was under private review for months. The hypothesis, posed by Bernhard Riemann in 1859, concerns the distribution of prime numbers and carries a 1 million dollar Clay prize, but that prize needs publication and two years of scrutiny.
What happened: Alpoge posted his note on Friday morning, writing: I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis. He said the same method also settles the Generalized Riemann Hypothesis, a broader version that extends the claim to a wider family of functions. With the note he released the proof as code in Lean, a programming language that checks every step of a mathematical argument by computer, split into a 14-part archive running to more than 17 million lines. He released only that version for now. A full written argument would be extremely long, he said, adding that he had not yet mastered all of the mathematics involved and planned to write it up properly. Science reports that the natural-language write-up runs to more than 250,000 pages. Alpoge had consulted Anthropic chief executive Dario Amodei before announcing the result, and he acknowledged that the timing could be viewed cynically, without elaborating. The announcement came the day after Anthropic's record debut on the Nasdaq. According to DataCamp, Alpoge had been working with Claude since 6 October, steering the work toward the Riemann problem, with several key steps dated 15 November. Science also reports that he released advances on the Euler and Navier-Stokes equations the same day.
Why it's trending: The Riemann Hypothesis is widely regarded as the most important open question in pure mathematics, unsolved for more than 160 years, so a claim to have settled it with an AI model is the biggest mathematics story of the year. The reaction from mathematicians has been a mix of awe and disbelief. Brian Conrey, a leading specialist on the hypothesis, told AFP: It's a monumental moment in human history. It's really the biggest thing I think that's ever happened in mathematics. He added that he had not yet seen a manuscript, only the remarks Alpoge posted on his website. Sutherland, after compiling the proof himself, called it an historic moment and a human triumph, and a day for celebration. The claim also crowns a remarkable year for AI in mathematics. In July, Alpoge posted a Claude-generated solution to the Jacobian conjecture. In August, Anthropic said an unreleased version of Claude had improved a long-standing result related to the hypothesis. In September, OpenAI said its models had solved the Navier-Stokes problem, another Clay prize problem, and in October it released 722 AI-generated manuscripts. Alpoge himself struck a mournful note, writing that it is natural to feel that this is an end, in his case a loss of a principal guiding light, but that what he saw amazed him.
Background / context: Bernhard Riemann posed the hypothesis in 1859. It concerns the zeta function, which adds together an infinite series of fractions, and the values for which that function equals zero. Riemann hypothesised that all of the function's non-trivial zeroes can be expressed as a complex number whose real part is always equal to one half. That claim, if true, would tightly constrain how prime numbers are distributed along the number line, which is why so much of mathematics depends on it. The prime number theorem, proved in 1896, lets mathematicians estimate how many primes sit below a given number, but only loosely. Computers have checked the first ten trillion zeroes, but that proves nothing about the next one. The hypothesis is one of the seven Millennium Prize Problems selected by the Clay Mathematics Institute in 2000, each carrying a 1 million dollar bounty. Under the institute's rules, a solution must be published in a peer-reviewed journal and withstand two years of scrutiny, and a committee must then accept it. The Generalized Riemann Hypothesis extends the same claim to Dirichlet L-functions, a wider family of functions. Many results in mathematics assume the hypothesis is true, and a correct proof would turn them into theorems. But contrary to some popular accounts, a proof would not reveal the location of every prime number or break encryption systems.
What we know / what's unconfirmed: Confirmed: Alpoge's note and his exact words; the Friday, 20 November date; the Lean formalization of more than 17 million lines in a 14-part archive; that there is no preprint or journal submission; that he consulted Amodei before announcing; that Sutherland independently compiled the proof in Lean with only the three standard axioms; that two proof-checking systems accepted the result, according to The Hindu; and that a second claim by Masullo appeared the same day, shared by Vakil and vetted by Stanford's Brian Conrad after months of private review. One check is still open and needs flagging. As of Scientific American's Friday afternoon report, researchers still awaited a successful comparator run, which checks that the code does not redefine the meaning of any symbol in the statement and that it compiles in paranoid mode. A later report says mathematician Elliot Glazer posted on X that the formalized statement in the code is in fact the Generalized Riemann Hypothesis, but that is a social-media report, not a confirmed check, and it is flagged as such here. Unconfirmed: whether the community will accept the proof, which will take months at least; whether Alpoge's claim about the Generalized Riemann Hypothesis holds in full; and anything about the 1 million dollar prize, which cannot even begin its process until a paper is published.
What happens next: The next steps are technical and slow. The comparator run should settle within hours whether the Lean code really proves the standard statement, and mathematicians will then spend months working through 17 million lines of code to understand why the proof works, if it does. Alpoge has said he plans to write the argument up properly, a precondition for the Clay prize process, which itself takes at least two years after publication. Masullo's human-written draft will be read in parallel. If Alpoge's proof holds, hundreds of results that assume the hypothesis become theorems. If a flaw is found, the episode still marks a turning point: a machine produced a proof candidate for the most famous problem in mathematics, and independent experts took it seriously enough to compile it within hours.
Q: What exactly has Alpoge claimed?
A: Levent Alpoge, a mathematician at Anthropic, says he and Claude discovered a proof of the Riemann Hypothesis, and that the same method also settles the Generalized Riemann Hypothesis. He released the proof as more than 17 million lines of Lean code in a 14-part archive, but no traditional paper, preprint or journal submission.
Q: Has the proof been verified?
A: Not yet. MIT mathematician Andrew Sutherland independently compiled the Lean code and said the chance that anything is fishy is approximately zero, and two proof-checking systems have accepted the result. But a final comparator run to confirm the code really states the standard theorem was still awaited on Friday afternoon, and the mathematical community has not formally accepted the proof.
Q: What is the Riemann Hypothesis in plain terms?
A: Posed by Bernhard Riemann in 1859, it says that all the non-trivial zeroes of the zeta function have a real part of exactly one half. That claim, if true, tightly constrains how prime numbers are distributed. Computers have checked the first ten trillion zeroes, but a proof must rule out counterexamples forever.
Q: What is the second proof claim?
A: On the same day, American Mathematical Society president Ravi Vakil shared a draft by Stanford postdoc Alessandro Masullo claiming the Generalized Riemann Hypothesis by an entirely human route, with no AI involved. The draft had been under private review for months and was vetted by Stanford's Brian Conrad. According to DataCamp, the work took twelve years.
Q: Could Alpoge win the 1 million dollar Clay prize?
A: Not any time soon. The Riemann Hypothesis is one of the Clay Mathematics Institute's seven Millennium Prize Problems, each worth 1 million dollars, but under the institute's rules a solution must be published in a peer-reviewed journal, be accepted generally by the mathematical community over two years, and survive scrutiny by a bespoke committee. Alpoge has not yet published a paper.
Sources: Phys.org — Anthropic researcher says he and Claude proved the Riemann hypothesis, first posed in 1859, Scientific American — Anthropic mathematician claims to prove Riemann hypothesis, most famous problem in all of math, Scientific American — The AI Riemann hypothesis proof is almost certainly true, independent mathematicians attest, and The Hindu — Anthropic scientist claims proof of maths most famous problem, with AI help.

Anthropic opened at $111.50 on its Nasdaq debut, briefly touching a $2 trillion value, then slid to close at $97.17, still 17% above its $83 IPO price.
20 November 20267 min read

Google DeepMind's AlphaProof Code formally verified Signal's encryption down to compiled machine code — a first for a widely used protocol.
12 November 20264 min read

OpenAI's quarterly revenue crossed $10 billion for the first time in Q3 2026, up about 50% from Q2, while its operating loss widened to $10.9 billion.
12 November 20264 min read