Sam Altman has taken up an online dare to hunt a room-temperature superconductor, replying to a satirical post with “you know what, let’s try.”
The remark landed days into a dispute over OpenAI’s Millennium Prize claim. The company says its agents solved one of seven famous math problems worth $1 million each.
A Viral Dare Follows the Math Fight
A developer needled OpenAI, arguing that cracking a few more Millennium problems would silence every skeptic. Sam Altman replied with two words, “ok fine.”
Hours later, another user floated a satirical rumor that Anthropic had quietly found a room-temperature superconductor. The joke warned that a rival might throw 10,000 agents at the problem for 88 hours to beat the press release. The post lifted both figures straight from OpenAI’s account of its disputed Navier-Stokes proof.
That account is already contested. Two mathematicians, Tristan Buckmaster of New York University and Anthropic’s Levent Alpoge, say the work may lean on their unpublished research. OpenAI denies touching it.
Altman leaned into the banter instead of brushing it off. His five-word reply turned a running gag around the GPT-6 Astra launch into a public challenge.
Why the Sam Altman Superconductor Dare Matters
Charge a phone, and the back gets warm. That warmth is electricity going to waste, lost as current pushes through ordinary wires. Superconductors waste none of it, because they carry current with zero resistance.
The catch is temperature. Today’s superconductors only work under extreme cold or crushing pressure. One that works at room temperature would cut waste from power grids, chargers, and chip design alike. Physicists have therefore chased that material for decades.
No evidence supports the room-temperature superconductor rumor, and Anthropic has announced nothing. Meanwhile, one of its own researchers quit this week over superintelligence safety warnings, not superconductivity. The researcher had spent three years on pre-training work at Anthropic and OpenAI.
Altman’s answer reads as banter rather than a roadmap. Still, OpenAI now carries a public dare it took on itself. Whether anything concrete follows the joke should become clear soon enough.