A 23-year-old with a ChatGPT subscription just solved a 60-year-old conjecture that world-class mathematicians couldn't crack. The problem sat unsolved for six decades. He asked the right question to an AI and got an answer. It's the kind of moment that gets framed as a threshold—proof that artificial intelligence has crossed from tool into collaborator, that the hierarchy between human intuition and machine capability has inverted.
The market did not move.
Tech stayed bid. Equities closed at records. Nobody recalibrated their portfolio based on a 23-year-old proving an Erdős problem. And that tells you everything about what's actually happening right now.
The AI narrative isn't about breakthrough capability anymore. It's about *validation theater*. We've moved past asking "can AI solve hard problems?" to "how do we price the companies selling the tools that let AI solve hard problems?" The solution to the math problem doesn't change valuations. The *existence* of the math problem solution does—because it's evidence that the companies you already own (OpenAI's backer Microsoft, the cloud providers training these models) are genuinely useful to people willing to pay for them.
Meanwhile, the real pressure is elsewhere. Mali just experienced coordinated attacks across multiple cities—jihadist groups and separatist forces hitting simultaneously in ways one analyst called "the largest in years." Geopolitical fragmentation is accelerating. And at the same time, Trump canceled the Witkoff-Kushner trip to Pakistan for Iran talks. The de-escalation trade is off. The risk-on bid that's been holding equities together depends partly on the idea that we're *avoiding* spiraling Middle East conflict. That trade just got shakier.
The Contrarian's nightmare scenario is hardening: geopolitical instability plus unsustainable valuations plus a realization that earnings can't keep pace with price multiples. The math problem proof doesn't move the needle on any of that. It's a fact that matters in five years. Today it's noise in a story about whether the world stays calm long enough for the tech-driven rally to mature into fundamentals.
Here's what bothers me: the insiders are still selling. Jarrod Patten at MSTR dumped $336K in stock over two trading days in late April. The sales happened at prices between $165 and $172 per share. Same pattern as before—directors and executives reducing exposure during a period when their companies are supposedly benefiting from AI tailwinds. Nobody's buying their own stock. Patten wasn't diversifying for lifestyle reasons; the transactions were immediate-turnaround option exercises followed by market sales. That's not confidence. That's liquidity.
The math problem is real. The market's indifference to it is realer.