OpenAI Solves 90-Year Math Problem in 88 Hours

Key Takeaways

• OpenAI claims an internal model solved the Navier-Stokes existence and smoothness problem in 88 hours using 10,000 AI agents.
• The solution describes a fluid vortex reaching infinite speed in finite time while keeping total energy finite.
• OpenAI will not accept the $1 million prize from the Clay Mathematics Institute for this result.
• NYU professor Tristan Buckmaster alleges OpenAI used data from his and Levent Alpöge's independent research.
• Buckmaster claims OpenAI researchers threatened his career if he did not agree to remove Alpöge's name from the credit.
• OpenAI admits it cannot rule out that de-identified data from human researchers influenced its model.
• Amazon plans to spend over $220 billion on capital expenditures in 2026, largely for AI infrastructure.
• Google's TPUv7 Ironwood chip offers 50% better performance per dollar than NVIDIA's Blackwell Ultra.
• Dell Technologies is refinancing $4 billion in debt following a sales surge driven by AI server demand.
• DI-Bench tests show current AI agents achieve only 32% accuracy on complex data intelligence tasks.

OpenAI claims AI solved math problem amid controversy

OpenAI announced that an unreleased AI model solved the Navier-Stokes existence and smoothness problem in about 88 hours. The company used roughly 10,000 AI agents to find a proof that a fluid vortex can reach infinite speed in finite time while keeping total energy finite. This problem is one of seven Millennium Problems set by the Clay Mathematics Institute in 2000, each offering a $1 million prize. OpenAI stated it will not accept the prize money for this result. However, the announcement sparked a debate about whether the AI used data from human researchers working on similar problems.

NYU professor questions how OpenAI solved math problem

OpenAI released a proof for the 90-year-old Navier-Stokes problem using 10,000 AI agents and about 130 billion tokens over four days. The company claims its internal model is significantly more capable than GPT-6 Astra. NYU professor Tristan Buckmaster says he and colleague Levent Alpöge were working on a related problem using OpenAI Codex and Anthropic Claude. Buckmaster alleges that OpenAI knew about their work before releasing its own solution and may have used de-identified data from their sessions. He claims OpenAI researchers offered to remove Alpöge's name from the credit in exchange for sharing the solution.

OpenAI AI agents solve million dollar math problem

OpenAI says an experimental system solved the Navier-Stokes existence and smoothness problem after 10,000 agents worked for 88 hours. The solution describes a fluid vortex that speeds up infinitely in finite time while maintaining finite energy. The proof requires verification by the mathematical community before it can be officially recognized. OpenAI states it will not claim the $1 million prize from the Clay Mathematics Institute. The company admits it cannot rule out that data from human researchers influenced its model, though it says no specific user data was accessed.

Researcher claims OpenAI copied his math work

Tristan Buckmaster from NYU says OpenAI solved the Navier-Stokes problem using a method his team had chosen independently. Buckmaster and Levent Alpöge spent a year working on a related Euler equations problem using AI tools like Codex and Claude. They found that OpenAI knew about their progress and reached out to offer a joint announcement. Buckmaster claims OpenAI researchers threatened his career if he did not agree to remove Alpöge's name from the credit. OpenAI says it did not see their work before publication but acknowledges de-identified data might have helped.

OpenAI AI leapfrogs mathematicians on Millennium Problem

OpenAI announced its AI solved the Navier-Stokes problem, one of the seven Millennium Problems chosen by the Clay Mathematics Institute. The company used an internal model and 10,000 AI agents to complete the task in 88 hours. Mathematician Tristan Buckmaster says he and his colleague were working on a similar problem when OpenAI contacted them. OpenAI stated it began its work after hearing a rumor about Buckmaster and Alpöge's progress. Some experts worry this competition turns mathematics into a meaningless game rather than a collaborative pursuit.

OpenAI claims to solve 90-year math problem in 88 hours

OpenAI announced it solved the 90-year-old Navier-Stokes math problem in just 88 hours. The company used a system of 10,000 coordinating agents powered by its internal AI model to reach this solution. These agents could read cached internet data and run code to complete the task. Tristan Buckmaster, a professor at New York University, questioned how quickly the team arrived at the answer. He suggested the group may have received tips about the progress made by Alpöge and Buckmaster. The Clay Mathematics Institute has not yet commented on the proposed solution.

Google TPUv7 Ironwood beats NVIDIA Blackwell Ultra on cost

SemiAnalysis found that Google's TPUv7 chip, codenamed Ironwood, offers 50% better performance per dollar than NVIDIA's Blackwell Ultra. This independent benchmark shows Google's chip is more cost-efficient for AI tasks. The advantage comes from specialized hardware called SparseCore that organizes data faster. Google also released new software tools called Pallas inference kernels to help the chip run efficiently. These improvements allow the chip to handle common AI tasks more quickly and cheaply.

Amazon plans to spend over 200 billion dollars on AI

Amazon expects to spend more than 220 billion dollars on capital expenditures in 2026, with a huge portion going to AI infrastructure. This massive investment aims to build the computing power needed for the next generation of AI services. The company believes it must invest early to meet future demand, similar to how it built its logistics network. However, investors worry that the industry might build too much capacity if AI demand does not grow as expected. Microsoft, Alphabet, and Meta are also spending billions on AI infrastructure this year.

Dell refinances debt after AI sales surge boosts forecast

Dell Technologies is seeking to raise about 4 billion dollars through a bond sale to refinance its existing debt. The company plans to use the money to repay notes due in 2026 and for general corporate purposes. This move follows a significant increase in sales driven by strong demand for AI servers equipped with Nvidia chips. Dell is also seeing renewed interest in traditional servers for managing AI agents. Several analysts have raised their price targets for Dell stock following these positive developments.

New DI-Bench tests enterprise agents on real data tasks

Researchers created DI-Bench, a new system for testing how well enterprise agents can handle data intelligence tasks. This benchmark generates realistic questions that require both retrieving business knowledge and performing calculations. The system builds a graph linking data tables, dimensions, and documents to create complex problems. When tested, four different models achieved only 32% accuracy on tasks where business rules changed the computation. This shows that current AI agents struggle to combine knowledge retrieval with analytical math.

Sources

• The Fight Over OpenAI’s Math Breakthrough Is a New Kind of Scientific Arms Race
• OpenAI solves 90-year-old Navier-Stokes maths problem with 10,000 AI agents; NYU professor who spent a year on the same problem using OpenAI's own AI says: "I asked again, about training, and I did…"
• OpenAI says 10,000 AI agents solved a $1 million 'Navier-Stokes' math problem
• OpenAI solution for the Navier-Stokes problem overshadowed by plagiarism controversy — researcher working on related proof claims his work with Codex influenced OpenAI's direction
• OpenAI says it has solved a 'Millennium Problem,' leapfrogging mathematicians
• OpenAI claims to have solved the 90-year-old Navier-Stokes math problem in 88 hours
• Google’s TPUv7 Ironwood Achieves 50% Better Performance Per Dollar Than NVIDIA’s Blackwell Ultra, Says Semi Analysis
• Amazon Plans to Spend More Than $200 Billion on AI. Is That a Smart Move?
• Dell's Debt Refinancing Follows Surge in AI Sales
• DI-Bench: Systematically Generating In-Domain Data Intelligence Benchmarks for Enterprise Agents