The Strange Capitalism of AI
AI’s frontier is becoming more governed and expensive in the West, even as China and open-weight models commoditize intelligence, lower prices and force competition back into the market.
By Vishnu Rajkumar · Marginalia
Peter Thiel borrowed from Anna Karenina to make his case against competition. Successful companies are different because each has found a way to escape it. Competition is good for the economy. It is rarely good for the capitalist.
Now look at AI. Dario Amodei says the frontier needs to be paced. Sam Altman agrees. Elon Musk says Dario is right. Demis Hassabis agrees with the direction. Four men building some of the most powerful AI systems have arrived, unusually, at roughly the same place; capability is moving faster than our ability to control it. The safety argument is serious. But pacing also brings standards, independent evaluators and coordination, making an already expensive frontier harder to enter.
Except China exists. And open-weight models exist. Just as the Western frontier becomes more expensive, closed and governed, another is making intelligence cheaper and more available. Silicon Valley is learning to govern scarcity just as intelligence itself is becoming downloadable. There is something almost embarrassingly capitalist about what communist China is doing; commoditizing the product, collapsing prices and forcing everyone else to compete.
Anna Karenina has acquired a strange new ending. The capitalists escaped competition. The communists brought it back.