When Jamie Dimon makes predictions, the C-suite pays attention. 

And at the America Business Forum in Miami on Thursday, the JPMorgan Chase CEO sketched a provocative horizon line: AI will optimize our work schedules within our lifetime. 

“It’s going to affect every application, every job, every customer interface,” Dimon said. “My guess is the developed world will be working three-and-a-half days a week in 20, 30, 40 years, and have wonderful lives.”

He isn’t speaking in hypotheticals. Dimon has turned JPMorgan into a live-fire AI lab. Roughly 2,000 people are building AI systems; about 150,000 employees use large language models every week on internal documents; and the bank has hundreds of use cases running, from fraud detection to legal review, reconciliations, and marketing optimization.

Dimon’s three-and-a-half-day forecast hinges on that cumulative productivity. As AI absorbs routine work, the same output may require fewer hours. But he’s adamant the transition won’t be painless. 

“It will eliminate jobs. People should stop sticking their heads in the sand,” he warned at Fortune’s Most Powerful Women conference, arguing companies and governments must plan for retraining, income assistance, redeployment, and, in some cases, early retirement to avoid a social backlash. The bank, he said, is building with that redeployment mindset.

He also stresses AI’s economics aren’t the same as the internet’s. The buildout is both capital- and power-hungry; some hyped projects “won’t get the power they need,” he said. Investors should underwrite data-center and AI infrastructure deal-by-deal, he added, noting who has revenue, who takes construction and technology risk, what happens if the chips or plants don’t perform—rather than buying the theme wholesale. In his words: Some AI efforts will be “in a bubble,” but in total, the technology “will probably pay off.”

Dimon has a consistent message to operators: Stop overintellectualizing model theology and deploy. 

“Use it… in any business,” he said. JPMorgan is even running AI master classes for senior managers after finding many leaders at the company simply didn’t know what current tools could already do (One reaction Dimon quoted was: “I didn’t know it could read 100,000 documents.”).

All in all, the future of work may be shorter in hours but richer in value, if leaders do the hard parts now. That means modernizing data so AI can actually use it (“We spend a lot of money getting data into the proper format… We’re not measuring how much it costs,” Dimon said), investing through the power constraints, and building humane off-ramps for roles that will disappear. 

Dimon’s bet is that what machines remove, well-led institutions can remake: And that’s how you get to a three-and-a-half-day week without blowing up the social contract.

“You know, technology has downsides. It’s used by bad people,” Dimon said at the Forum. “But embrace it.” 

This story was originally featured on Fortune.com