In partnership with
OpenAI published a Navier Stokes result on September 8. It described the result as a solution to a Clay Mathematics Institute Millennium problem. OpenAI says an internal system produced a proof of finite time breakdown.
Navier Stokes equations describe fluid motion through velocity, pressure, viscosity, and applied force. The Clay formulation asks whether smooth, physically reasonable solutions remain smooth for all time, or whether a permitted construction can break down.
200+ Proven Ways to Make Money With AI in 2026
The next wave of millionaires will be people who figured out how to make AI work for them.
The window to get ahead is still open. But not for long.
Here are 200+ proven ways to make money with AI in 2026.
Sign up for Superhuman AI, the free daily newsletter read by 1M+ professionals, and get instant access to all 200+ ways to profit from AI this year.
OpenAI says its result establishes alternative C and alternative D. Both alternatives allow a smooth external force. The company’s account says an initially smooth fluid can develop a singularity while its energy remains finite. It released a written proof and a formalization in Lean, a proof assistant.
That distinction matters. A formal proof assistant checks that a coded argument follows from the definitions and lemmas accepted by the system. It does not by itself show that the coded statement matches the statement people think they are reading. The formalized proposition must still be inspected.
The Clay Institute responded on September 11. It said the problem had apparently been settled and noted that its evaluation rules require a deliberately unhurried process. The response is recognition of a public claim, not an award or an independent certification of every step.
The event has a second layer involving research provenance. OpenAI says it began its effort after hearing reports that mathematician Tristan Buckmaster and Anthropic researcher Levent Alpöge had made progress on related fluid equations. The company says its investigation found no access to their specific user data and that the approaches differ.
Axios reported that Buckmaster questioned whether unpublished research influenced the effort. It also described disagreement over authorship and access to private coding prompts. Those allegations remain disputed. They should not be presented as established findings.
AI made PMs faster. Multiplayer mode is still broken.
A PM can summarize research, draft a PRD, and mock up a prototype before lunch. The hard part starts when the team has to decide what actually gets built.
Jira Product Discovery gives product teams one place to capture insights, prioritize ideas with consistent frameworks, and build living roadmaps stakeholders can rally around.
And because it’s connected to Jira, the context behind every decision stays with the work—so developers and their agents know not just what to build, but why.
AI helps PMs move faster. Jira Product Discovery helps the whole team build with confidence.
OpenAI’s technical account gives unusual operational detail. It says agents began work on September 1. They reached the Navier Stokes result about 88 hours later, then spent another 17 hours on Lean formalization and verification. The company reports roughly 10,000 concurrent agents and large message and token totals.
Those numbers describe OpenAI’s own account. They are not an independent measure of proof quality, mathematical novelty, or cost efficiency. The announcement also says the system used a model more capable than GPT 6 Astra. That model is not publicly available for replication.
The underlying mathematics is easy to misread. A forced breakdown result is not the same as proving that every unforced three dimensional solution becomes singular. Clay’s original problem lists several alternatives. A valid result must match one of those permitted statements and satisfy its regularity, force, and energy conditions.
The released Lean project therefore becomes a central research object. Reviewers need to examine its assumptions, imported libraries, theorem statements, and connection to the Clay formulation. They also need to compare the written proof with the formal code. A green compilation result would not answer every mathematical or attribution question.
The episode shows why AI research claims need more than a dramatic output. Reproducible artifacts, precise problem definitions, independent examination, and a record of source access are separate parts of the evidence. Each answers a different question.
For now, the defensible description is narrower than a settled headline. OpenAI reports a formalized construction aimed at Clay alternatives C and D. Clay says the problem appears settled but has not finished its process. Independent mathematical review and credit assessment remain open.
This case will be useful even before its final status is known. It tests whether research organizations can disclose enough technical and provenance information for others to evaluate machine generated mathematics.
10 AI Stocks to Lead the Next Decade
AI isn’t a tech trend – it’s a full-blown, multi-trillion dollar race, and 10 companies are already pulling ahead.
These are the innovators driving real revenue, attracting institutional attention, and positioning for massive growth.
Get all 10 tickers in The 10 Best AI Stocks to Own in 2026, free today.
