OpenAI Says Its AI Cracked a 90-Year-Old Math Problem. A Mathematician Says It Took His Work to Do It.

Tuesday was the kind of day AI coverage doesn’t usually get: a genuine mathematical breakthrough, immediately followed by a fight over who actually deserves credit for it. Alongside that, two very different companies made very different bets on the AI buildout, and in India a wicketkeeper spent two days rewriting a record book that has stood for decades.

OpenAI says it cracked a 90-year-old math problem. Not everyone is applauding.

On September 8, OpenAI announced that an unreleased internal model, working with a swarm of up to 10,000 AI agents over roughly 88 hours, had produced a proof for the Navier-Stokes existence and smoothness problem, one of the seven Millennium Prize Problems set by the Clay Mathematics Institute in 2000. The equations describe how fluids like water and air move, and mathematicians have spent nearly a century unable to say for certain whether smooth, well-behaved fluid flow can suddenly break down into a singularity. OpenAI’s model found that it can: a proof describing a spinning vortex that tightens on itself in finite time even though the fluid’s total energy stays bounded. GPT-6 Astra, the company’s public model, then spent another 17 hours formalizing the argument in Lean, software mathematicians use to machine-check every logical step. OpenAI says it does not intend to claim the $1 million prize, and outside mathematicians still need to independently confirm the result meets the Clay Institute’s own criteria.

The announcement landed one day after Tristan Buckmaster, a mathematician at NYU, posted online that he and his collaborator Levent Alpöge had made their own breakthrough on a related equation, using AI coding tools extensively along the way. Buckmaster has since alleged that OpenAI became aware of the pair’s unpublished method and adopted a similar approach for its own proof, and that an OpenAI researcher, Sébastien Bubeck, pushed to have Alpöge, who works at Anthropic but was not doing this research for the company, dropped from co-authorship. Buckmaster says that when he threatened to make the dispute public, Bubeck told him, “Why would you ruin your career?” Bubeck has publicly called the allegations “false and inflammatory” and says he acted according to normal academic practice throughout. Separately, Fields Medalist Terence Tao praised Buckmaster and Alpöge’s own results as a genuine breakthrough, while cautioning days earlier that if an AI’s route to a proof stays a black box, its value to the field of mathematics would be close to zero even if the proof itself checks out.

Two things can be true here: the Lean-verified proof is a real technical achievement, and the story of how it was produced is now genuinely disputed. Expect the arguments over both to run for a while.

Qualcomm’s deal with Amazon could be worth $60 billion. Today, it’s worth $4 billion.

Qualcomm and Amazon Web Services announced a multi-generation partnership on September 8 to co-design custom AI inference chips and high-speed optical connections for AWS data centers. As part of the deal, Qualcomm issued Amazon warrants to buy 25 million shares of Qualcomm stock at $161.26 each, a stake worth roughly $4 billion if Amazon exercises the whole thing.

The much larger number attached to this deal, up to $60 billion, is not a guaranteed order. It is the ceiling: the warrants vest in tranches as Amazon actually places and completes purchase orders for Qualcomm’s server chips and related technology, and Qualcomm has said 3.75 million of the 25 million shares, 15 percent, have already vested based on Amazon’s initial purchase commitments. So the deal is real and partly funded today, but the headline $60 billion figure describes what Amazon could spend over multiple future chip generations, not what it has spent. Qualcomm’s stock rose more than 3 percent on the news, a sign investors read even the partial commitment as a meaningful diversification away from the smartphone chip business that has defined the company for decades.

Mistral just raised the largest funding round in European tech history

French AI company Mistral closed a €3 billion Series D funding round on September 8, at a post-money valuation of more than €21 billion, which the company and several outlets tracking European funding describe as the largest single equity round any European technology company has ever raised.

Samsung Electronics led the round, with the Scaleup Europe Fund, managed by EQT, and existing backer PSG Equity serving as co-leads. New investors include Advent, funds managed by BlackRock, and the government of Luxembourg, joining a long list of existing backers that includes Andreessen Horowitz, ASML, General Catalyst, Lightspeed, Nvidia and Salesforce Ventures. Mistral says the money will go toward expanding its own computing infrastructure, including data centers it owns outright rather than only rents, as well as its research and international commercial push.

Mistral has positioned itself as Europe’s answer to the American and Chinese labs dominating frontier AI, built around the idea that European governments and companies want AI systems they can host and control on their own soil. A round this size, barely three years after the company’s founding, is the clearest evidence yet that investors, including a major hardware maker like Samsung, are willing to bet heavily on that pitch.

An Indian wicketkeeper just rewrote a cricket record book that has stood for decades

At the Duleep Trophy final in Chennai, East Zone captain and wicketkeeper Ishan Kishan scored 270 runs off 334 balls against South Zone, an innings that included 30 fours and seven sixes and helped his side post a total of 708. He briefly retired hurt after passing 250 before returning to the crease to keep batting.

The knock is the eighth-highest individual score in the tournament’s history and the highest ever by a wicketkeeper in a Duleep Trophy final, passing Dinesh Karthik’s 183. Kishan is only the third East Zone batter to score a double century in the competition, and the first wicketkeeper-captain to do it in a final. Along the way he also passed 4,000 runs in first-class cricket. Former India star Yuvraj Singh publicly praised the innings as a special one, and Kishan later said his best cricket tends to come under pressure.

Individual milestones like this one do not always translate into team success, but a 708-run total built partly on an innings this dominant gives East Zone a commanding position in the final as it heads into its later days.

None of these four stories share an industry, but they share a question worth sitting with: who gets to claim credit when a breakthrough happens, and how much do we actually know about how it happened. OpenAI wants credit for a math proof it says an AI produced largely on its own. A mathematician says the truth is messier than that. Qualcomm wants credit for a $60 billion deal that is, for now, a $4 billion one. The only claim in today’s roundup nobody is arguing about is the 270 next to Ishan Kishan’s name in the scorebook.