The AI Boom Is Really Two Bets in One
Former regulator Gary Gensler argues tech spending must pay off twice, in sales and in fast economywide growth.
Freakonomics Radio Network · 684. He Helped Clean Up the Last Crash. Does He See Another One Coming? | Freakonomics RadioWhen Stephen Dubner asked Gary Gensler, the former Goldman Sachs executive who led two federal market agencies and now teaches at MIT, what he sees as "the big financial risk in the economy," Gensler answered with a gambler's phrase.
"We have a parlay bet right now," Gensler said.
A parlay is a bet that pays only if every leg wins. Gensler used it to describe artificial intelligence as two linked wagers that must both come through.
The first leg is revenue. He said the capital spending by model companies like OpenAI and Anthropic, and by large cloud companies like Microsoft and Google, "will lead to enough revenues."
The second leg is broader and faster economic payoff. He said that spending will also have to "lead to sufficient productivity gains in the economy in the near term."
By near term, Gensler said he did not mean 10 or 20 years from now. He said the gains would need to arrive soon enough "to make up for what is going to happen," a phrase he left unexplained in this part of the discussion.
The force of his idea is the doubling. In Gensler's framing, it would not be enough for AI investment to produce sales for AI companies. It would also have to make the rest of the economy more productive, and do it quickly.
Questions people ask
What agencies did Gensler lead?
He chaired the Commodity Futures Trading Commission under President Barack Obama and the Securities and Exchange Commission under President Joe Biden. Both oversee markets, with the first focused on derivatives and the second on stocks and public companies. He now teaches finance at MIT.
What is an AI model company?
It is a business that builds the large AI systems other people use, such as OpenAI and Anthropic. These firms spend heavily on computing, data, and researchers to train their systems. They earn money by selling access to those systems directly or through cloud providers.
What does productivity gain mean in this context?
It means workers and businesses produce more value with the same time and resources, often because a tool saves labor or reduces waste. Economists care about it because sustained gains tend to raise living standards over time. Gensler is emphasizing gains that arrive soon, not over decades.