A glass of water looks simple: if you pour it, it flows; if you stir it, it swirls. Yet describing exactly how that water moves leads to one of mathematics’ most famous unsolved problems.
The story begins with Newton’s second law, F = ma. For a ball, applying it is relatively straightforward: identify the forces, and calculate the acceleration. However, a fluid is much harder because it is not one object. Every tiny part of it is moving while simultaneously pushing and pulling against the parts around it. Liquids are much harder to map out than solid objects.
Instead of tracking every molecule, mathematicians treat a fluid as continuous and describe its velocity at every point in space. This creates a velocity field: a map of how the fluid is moving.
Leonhard Euler was among the first to develop equations describing this motion. When he mapped it out, he assumed an ideal fluid with no viscosity. In the nineteenth century, Claude-Louis Navier and George Gabriel Stokes developed the equations further by accounting for viscosity, the internal friction that causes neighboring layers of fluid to resist sliding past one another. These became the Navier-Stokes equations.
At their core, they are simply Newton’s second law applied to a fluid: pressure differences accelerate the fluid, viscosity smooths out differences in its motion, and gravity can provide another source of acceleration. The difficult part is that the fluid’s movement also changes what happens next. A piece of water moving into a region where the velocity is different will itself accelerate, causing changes to the surrounding flow, which then changes its motion again. This feedback mechanism is what makes the equations nonlinear.
Being able to use the equations is one thing, but interpreting them can be a whole different beast.
Consider smoke rising from a candle. At first, it forms a relatively smooth column, but eventually the flow twists and breaks into increasingly complicated structures. This is turbulence. The motion still follows Navier-Stokes, but the interactions between different scales of motion become extremely difficult to control mathematically.
The fundamental question is simple to state: if a three-dimensional fluid starts out smooth, must its solution remain smooth forever?
There are two possibilities. The equations might always produce a well-behaved solution, or the velocity could become infinitely large in a finite amount of time, producing what mathematicians call a singularity. We can simulate enormous numbers of fluid flows on computers, but a simulation can only examine particular cases. A proof has to establish what happens for every case covered by the problem.
That distinction is the heart of the Navier-Stokes existence and smoothness problem.
In 2000, the Clay Mathematics Institute named it one of the seven Millennium Prize Problems, offering a $1 million prize for a correct solution. The problem asks whether smooth solutions in three-dimensional space always exist and remain smooth. For decades, mathematicians could not prove either possibility.
A Singular Vortex
For a single vortex, the difficulty is easiest to picture through turbulence. Imagine a whirlpool becoming increasingly concentrated. As its central region shrinks, the fluid can spin faster and faster. The mathematical question is whether this process can continue until the velocity becomes unbounded in finite time, or whether viscosity must always prevent such a breakdown.
That distinction is why Navier-Stokes has remained so important. We know that the equations can accurately model real fluids, but mathematical understanding requires proving exactly what their solutions can and cannot do.
AI Enters the Problem
In September 2026, OpenAI announced that an internal AI system had produced a proof of a finite-time singularity for a three-dimensional Navier-Stokes system. The reported construction involves a vortex that spirals inward and becomes increasingly elongated while its velocity grows without bound.
The proof was produced by a system of cooperating AI agents. According to OpenAI, the group working on Navier-Stokes involved around 10,000 concurrent agents and reached its reported result about 88 hours after the project began. Formalization and verification took another 17 hours.
The announcement does not mean the mathematical community has simply accepted the problem as finished. Clay described the problem as “apparently been settled” on September 11 and said its process for evaluating Millennium Prize solutions is deliberately unhurried.
Whether or not this particular proof ultimately receives the mathematical community’s full acceptance, the story illustrates something remarkable about Navier-Stokes. An equation that began with Newton’s simple idea of force and motion eventually became a problem about whether a fluid can mathematically tear itself apart.
We have spent more than a century learning how to use these equations, but understanding everything they are capable of doing is another matter entirely.
Bibliography
Britannica, The Editors of Encyclopaedia. “Navier-Stokes Equation.” Encyclopaedia Britannica, 23 Aug. 2024, https://www.britannica.com/science/Navier-Stokes-equation
Clay Mathematics Institute. “Navier-Stokes Announcement.” Clay Mathematics Institute, 11 Sept. 2026, www.claymath.org/news/navier-stokes-announcement/.
Clay Mathematics Institute. “Navier-Stokes Equation.” Clay Mathematics Institute, www.claymath.org/millennium/Navier-Stokes-Equation/.
“Navier-Stokes Equations.” Encyclopedia of Mathematics, Encyclopedia of Mathematics.
NASA Glenn Research Center. “Navier-Stokes Equation.” NASA, NASA Glenn Research Center.
OpenAI. “On the Navier-Stokes Millennium Prize Problem.” OpenAI, 8 Sept. 2026, openai.com.
Leave a comment