Inside Hudson River Trading's Blistering Token Burn

Hudson River Trading’s head of AI, Ian Dunning, opened his live Odd Lots appearance with a confession: his last tweet before the show read, “Worry it’s some sort of AI-induced delirium, but then again, various empirical measures are exponential looking. So feels best to assume we’re hurtling towards some sort of end game.” On stage, he admitted he still can’t tell if that’s feverish or clear-eyed. But the numbers keep piling up.

The exponential feeling from inside a trading firm

Dunning has been working on AI since his time at DeepMind, and he describes himself as a “pragmatic engineer archetype” who was a natural skeptic for years. That skepticism has eroded not because of any grand vision, but because of what he sees inside his own firm. The amount of compute Hudson River Trading uses has grown exponentially year over year. The same technology now works across every equity, future, crypto, and option market globally in a unified approach. And Dunning notes that many of HRT’s peers are investing just as heavily. The question that keeps him up at night is what happens when everyone is pouring more compute into the same problem simultaneously. “Surely you can’t just keep getting better at predicting markets,” he says. “There’s got to be some sort of forcing function where your margins go to zero as you keep investing more in computers.” He calls this the “Highlander problem