# Navier-Stokes background reading note

Date read: 2026-11-20. Status of this note: background reading for the
Millennium Lab blowup hunt, not an audit verdict and not a verification
of any other party's proof. The hunt, not this note, is the deliverable
of this move.

## What was read

1. OpenAI, "On the Navier-Stokes Millennium Prize Problem", announcement
page, read in full on 2026-11-20.
Source: https://openai.com/index/navier-stokes-solution/
2. Clay Mathematics Institute, "Navier-Stokes Announcement", dated
11 September 2026, read in full on 2026-11-20.
Source: https://www.claymath.org/news/navier-stokes-announcement/
3. navier-stokes.org, "Clay Navier-Stokes Problem: Official Statement
Explained", read on 2026-11-20, for the text of the four Clay
alternatives and for its quotation of the manuscript's Theorem 1.1 and
Corollary 10.6.
Source: https://navier-stokes.org/navier-stokes-existence-and-smoothness/
4. Search results consulted for the priority dispute and for the status
as of November 2026: the Wikipedia page on the Navier-Stokes priority
controversy, Brookings (29 October 2026), and APJ-RAISE (November 2026).

## What was NOT read or checked

The full 166 page manuscript "Finite Time Blowup for Navier-Stokes"
was not read line by line. The Lean formalization was not downloaded,
built, or run. No independent check of the proof was performed. Nothing
in this note verifies or refutes the OpenAI construction.

## The Clay problem, as stated by Fefferman

The official formulation asks for a proof of any one of four
statements. (A) is global smooth existence with no external force on
R^3, for every admissible smooth initial flow. (B) is the same, with no
external force, on the periodic torus T^3. (C) is breakdown on R^3:
some admissible smooth initial flow and some admissible smooth external
force admit no global smooth bounded energy solution. (D) is the
periodic analogue of (C). A forced breakdown under (C) or (D) does not
by itself decide the unforced statements (A) or (B), because (C) and
(D) let the solver choose a force. Proving any one of the four resolves
the problem as Clay posed it.

## What OpenAI announced, in its own words

On 8 September 2026 OpenAI announced an analytical proof and a Lean
formalization showing that an initially smooth fluid at rest, with a
smooth force applied to it and finite energy throughout, can develop a
singularity in finite time. OpenAI states that this establishes
statements (C) and (D) in the official formulation. The mechanism
described is a vortex that spirals inward and becomes increasingly
elongated while its central region shrinks and speeds up, with the
Navier-Stokes terms becoming large yet cancelling in a precise balance
so that the external force stays smooth. OpenAI also reports that its
agents first resolved the unforced Euler regularity problem, then moved
to Navier-Stokes, reaching the result on 5 September, about 88 hours
after the effort launched on 1 September, with Lean formalization and
verification taking a further 17 hours via GPT-6 Astra. OpenAI states
that it does not intend to claim the Millennium Prize.

## Clay's response

On 11 September 2026 Clay said it shares in the excitement of the
global mathematical community as it contemplates the announcement that
the Navier-Stokes problem has apparently been settled, and that the
process for evaluating what has been achieved and for assigning credit
is deliberately unhurried, with updates to follow. Clay awarded no
prize. As of the sources checked on 2026-11-20, Clay's problem page
still lists Navier-Stokes as Active, no refereed publication of the
proof has been confirmed in the sources read, and no independent review
has completed. Under Clay's rules a solution is considered only after
it appears in a refereed publication, at least two years pass, and it
wins general acceptance in the global mathematics community. That
two year interval has therefore not started on the evidence read here.

## Priority dispute, as of 20 November 2026

About twelve hours before OpenAI's announcement, Tristan Buckmaster
(NYU) posted a statement on behalf of himself and Levent Alpoege
(Anthropic) saying their recent advances on related fluid equations had
been leaked to OpenAI days earlier and may have inspired OpenAI's
prompts. Buckmaster and Alpoege posted preprints on 7 September 2026
proving finite time blowup with smooth forcing for the incompressible
porous medium equation, the two dimensional Boussinesq system, and the
three dimensional Euler equations, all formalized in Lean. They do not
claim Navier-Stokes. OpenAI denies seeing their work, says no specific
user data was accessed, and says its investigation confirmed that
Buckmaster's Codex prompts could not have influenced its system. OpenAI
recognizes the priority of Buckmaster and Alpoege on forced Euler.
OpenAI researcher Sebastien Bubeck posted on X disputing parts of
Buckmaster's account, and Sam Altman retweeted it. On 11 September a
declaration titled "A Severe Misalignment of AI in Mathematics", signed
by Fields Medalists, warned about AI in mathematics. As of 20 November
2026 the dispute is unresolved in the sources read: no party has
produced evidence that OpenAI read Buckmaster and Alpoege's drafts, and
no joint statement settling credit has appeared in the sources read.

## Relevance to the hunt

The OpenAI construction is a forced blowup, statements (C) and (D).
The Millennium Lab hunt is different in kind: it is a numerical search
for blowup in the unforced equations, statements (A) and (B) territory,
on a small grid. A forced construction, even if correct, says nothing
about whether an unforced flow can blow up, and a numerical hunt on a
32^3 or 48^3 grid cannot prove or disprove a continuum statement. The
two efforts answer different questions.
