OpenAI Claims AI Solved Navier-Stokes Millennium Problem
By @sharedot · · 6 pages
- AI
- Navier Stokes
- Openai
- Mathematics
OpenAI says an internal multi-agent AI system proved the Navier-Stokes equations can develop a finite-time singularity, a claimed Millennium Prize first now under scrutiny.
What OpenAI Announced
On September 8, 2026, OpenAI announced that an internal multi-agent AI system produced a proof of the Navier-Stokes Existence and Smoothness problem, one of the seven Millennium Prize Problems selected by the Clay Mathematics Institute in 2000 with a $1 million prize each.
According to i-programmer.info, the system demonstrated that smooth three-dimensional fluid motion can develop a mathematical singularity in finite time with finite energy — a so-called "blow-up" where velocity and energy grow infinitely.
The company released a paper titled "Finite Time Blowup for Navier-Stokes" alongside a formalization checked in the Lean proof assistant, and stated it does not intend to claim the million-dollar prize.
The Massive AI Effort Behind It
The scale of the computation was extraordinary.
yourweather.co.uk reports that OpenAI deployed 10,000 autonomous AI agents coordinated through a state-of-the-art model running in parallel for about 88 consecutive hours, exchanging millions of messages, with estimated computing costs in the millions of dollars.
According to i-programmer.info, the effort consumed roughly 130 billion output tokens, equivalent to $23 million in a single week, and was spurred by rumors that rival Anthropic was close to a similar result.
The paper acknowledged the analytical framework for singularity formation established by Spanish mathematicians Diego Córdoba and Luis Martínez-Zoroa, i-programmer.info notes.
Controversy: Stolen Research Allegations
The announcement immediately ignited an authorship dispute. According to i-programmer.info, New York University mathematics professor Tristan Buckmaster publicly claimed that OpenAI stole unpublished research he had been developing with Levent Alpöge, a mathematician at Anthropic.
Buckmaster alleged that information about their private, AI-assisted work — run through tools like Codex — was prematurely absorbed or leaked into OpenAI's training or search pipelines, enabling OpenAI to close in on the Navier-Stokes proof and announce it first.
The accusation echoes earlier friction in the AI research race and remains unadjudicated.
What It Means and What Comes Next
If confirmed, the result would be conceptual rather than practical: it would establish an intrinsic limitation of the classical Navier-Stokes model, which has powered weather forecasting, engineering, and hydraulics for over a century.
yourweather.co.uk reports that no real fluid could actually reach infinite velocity — microscopic collisions and quantum fluctuations would dissipate energy first — so the proof shows the equations are not universal.
Skeptics also note the solution applies only to the case with external forcing, leaving the isolated-fluid case open. Academic verification of the Lean-checked formalization will decide whether this becomes the second.