How many more Millennium Prize Problems will AI solve in 2026?
I think there is a strong chance that no qualifying Millennium Prize Problems will be counted for 2026 under this market’s rules, even though OpenAI’s Navier-Stokes claim makes the broader “AI solved a Millennium Problem” narrative look more plausible. The exclusion of Navier-Stokes from resolution is the key reason the zero-solved outcome still looks likelier than not.
Analysis
The biggest input here is that the most prominent recent AI-math headline does not help the “Yes” side on this specific market. OpenAI’s announcement about Navier-Stokes, while potentially historic in a broader sense, is explicitly excluded from qualification. That means the market still resolves to 0 unless one of the five counted problems is clearly announced as solved by an eligible AI lab before year-end. On the current record, there is no equally strong public evidence that any of Birch and Swinnerton-Dyer, Hodge, P versus NP, the Riemann Hypothesis, or Yang-Mills has actually been solved in a qualifying way.
The calendar also matters a lot. With only part of September, October, November, and December left, the window for a new, clean, eligible announcement is short. Even if AI capability on frontier math is improving rapidly, these problems are extremely deep and the standard for this market is not just progress but an explicit claim of solution from an eligible lab. Historically, genuinely accepted breakthroughs on Millennium problems are rare even for human mathematicians, and the subset that can be credibly packaged as a definitive lab announcement is even smaller. That pushes the distribution heavily toward zero qualifying wins this year.
There is still some nontrivial upside to the “No” side if one believes the Navier-Stokes episode signals a broader internal capability jump that could spill into another problem quickly. OpenAI’s reported use of a large multi-agent system suggests the possibility of highly accelerated theorem-proving workflows, and other leading labs may be racing to produce comparable demonstrations. Still, the jump from producing a formal proof in one problem to publicly announcing a clean solution to one of the remaining five is enormous. The market-implied probability of 18% for at least one qualifying solve already reflects some tail risk, but I think the evidence supports an even lower chance that anything counted under the market rules will land before the deadline.
Arguments
For
- Arguments for Yes: The current news cycle shows how close frontier AI systems may be to producing formal mathematical breakthroughs.
- Arguments for Yes: The bar for resolution is an announcement, not independent acceptance, so a lab could satisfy the market faster than the broader math community would endorse.
Against
- Arguments against Yes: The only major reported AI proof this year is for Navier-Stokes, which the market explicitly excludes.
- Arguments against Yes: The remaining five qualifying problems are extraordinarily hard, and there is no verified public claim that any of them has been solved.
Key drivers
- Navier-Stokes is excluded from resolution, so the biggest AI-math headline does not count toward this market.
- There is no strong public evidence yet that any of the five qualifying Millennium Prize Problems has been solved by an eligible AI lab.
- The remaining time in 2026 is short relative to the difficulty and validation burden of a real Millennium Prize breakthrough.
Risk factors
- An eligible AI lab could still make an unexpected qualifying announcement on one of the five counted problems before year-end.
- The Navier-Stokes claim may indicate a capability jump that increases the odds of a second breakthrough faster than expected.
Scenarios
Best case
An eligible AI lab announces a convincing solution to one of the five qualifying Millennium Prize Problems before the end of 2026, which would force the market to resolve to No.
Most likely
The Navier-Stokes claim remains the main headline, but because it is excluded and no additional qualifying announcement emerges, the market resolves to Yes for zero more solved problems.
Worst case
No eligible AI lab makes a qualifying announcement on any of the five counted problems, and the year ends with only excluded or non-qualifying claims.
More from this day
- FinancialsKalshi13y
Will OpenAI or Anthropic IPO first?
AI22%MKT95%Edge-73HypedAnthropic now looks better positioned to list first, with OpenAI’s timeline slipping toward 2027 and multiple reports pointing to Anthropic as the earlier mover. I assign only a modest chance that OpenAI is the first of the two to IPO.
- pop culturePolymarket3mo
Where will 2026 rank among the hottest years on record?
AI15%MKT76%Edge-61Hyped2026 can still finish as an exceptionally warm year, but it is hard to justify it overtaking every prior year and ending as the single hottest on record. The market price looks too optimistic given the lack of a strong present signal that 2026 will surpass the recent peak years.
- CompaniesKalshi1y
Starbucks total global stores in 2026
AI64%MKT11%Edge+53Hidden GemStarbucks has a credible path to clear 41,800 global stores by its 2026 reporting date, and I think the market is pricing in too little growth. My independent estimate is a 64% chance of Yes, not 11%.