AI Just Solved One of Math’s Millennium Problems – What Does It Mean for Us?
3. Scientific discovery is becoming a question of computing power

There is another number worth considering: 10,000 AI agents.
The Navier–Stokes proof reportedly required 88 hours of coordinated work, nearly five million messages and approximately 130 billion output tokens. This represents an extraordinary amount of computational infrastructure. Most university mathematics departments cannot independently operate systems on that scale. The ability to tackle certain frontier problems may increasingly belong to organisations with access to enormous computing resources, specialised AI models and the money to run them.
Scientific discovery has always required resources, from laboratories and particle accelerators to powerful supercomputers. AI is introducing another expensive requirement: the ability to deploy thousands of specialised agents working together on a single intellectual problem.
This creates a new concentration of scientific power. A relatively small number of technology companies could gain an enormous advantage in determining which problems get solved, how quickly they are solved and who gets access to the results. For universities and independent researchers, the challenge will be finding ways to participate in a scientific environment where computational resources increasingly determine the boundaries of what is possible.