JPMorgan Spends $2 Billion on AI, Says It's Already Broken Even. Goldman Puts Up $6 Billion.

JPMorgan Chase runs a nearly $20 billion annual technology budget. A chunk of that now goes toward making sure the bank isn't caught flat-footed on AI, according to Business Insider. CEO Jamie Dimon has said he wants to win the AI arms race outright, and he told analysts on the bank's second-quarter earnings call that a $2 billion AI investment has already matched its cost in savings.

Dimon didn't lay out a line-item accounting of where those savings show up, and no independent audit of the figure has been published. Treat it as a bank's self-reported number until JPMorgan or a regulator says otherwise.

What's not in dispute is the scale of the rollout. JPMorgan has pushed a proprietary generative AI platform out to more than 200,000 employees, according to Business Insider, touching everyone from coders to portfolio managers. Dimon said the bank has nearly 1,000 internal AI use cases running, covering fraud detection, marketing, and note-taking.

The bank is also watching its own engineers closely. One internal dashboard tracks GitHub Copilot usage and sorts developers into "light," "heavy," or "non" user categories, Business Insider reported. Engineers now have updated performance objectives that expect them to "drive excellence" by adopting AI tools, according to posts on the company intranet.

Leadership churn has followed the buildout. Teresa Heitsenrether, JPMorgan's AI chief who spent decades at the bank, announced her retirement in July. The bank said it's restructuring its firmwide chief data and analytics office as it shifts from building AI infrastructure to applying it to actual business problems. JPMorgan had already reorganized its commercial and investment bank back in February specifically to "maximize the impact of AI," according to an internal memo reviewed by Business Insider, with major business lines now reporting to newly named chief operating officer Guy Halamish.

The AI push is reaching outside JPMorgan's own walls, too. The bank's asset management arm plans to drop its long-standing practice of using outside proxy advisors for U.S. shareholder votes this year. In its place: an in-house AI platform called Proxy IQ, which will support decisions on how the firm votes shares on behalf of clients. JPMorgan's asset management arm votes shares on behalf of millions of retail and institutional investors, giving the shift significance for corporate governance across corporate America.

Goldman Sachs is in the game too, committing $6 billion to technology spending this year, according to Business Insider's reporting, though details on Goldman's specific AI deployments were thinner in available coverage compared to JPMorgan's.

Dimon has been blunt that AI adoption alone isn't a competitive edge anymore. He told analysts his bank doesn't "uniquely benefit from AI since everyone is now using it." That's a notable admission from a CEO who's spent two years talking up his firm's AI ambitions: the technology has gone from differentiator to table stakes, by his own account.

Banks spending billions while their own CEOs admit the tech doesn't provide unique advantage anymore points to a broader problem showing up outside finance. Stephen Messer, co-founder of Collective[i] and Intelligence.com, argued in Fortune that most companies claiming an AI strategy are really just running what he calls the "AI Shuffle": swapping technology logos while keeping the same workflows, hierarchies, and approval chains intact. Messer's point is that bolting a chatbot onto a broken process doesn't fix the process, it just makes the same mistakes faster and more expensive to unwind.

Applied to Wall Street, that's the open question analysts keep raising on earnings calls: are these banks actually restructuring how work gets done, or mostly layering AI dashboards on top of the same org charts? JPMorgan's February reorganization and its Proxy IQ launch suggest at least some structural change. Its GitHub Copilot usage-tracking dashboard, on the other hand, sounds a lot like measuring adoption for its own sake, exactly the kind of activity Messer warns can look like progress without producing one.

None of this comes with independent verification of returns. No outside auditor has confirmed Dimon's $2 billion breakeven claim, and no regulator has weighed in on whether AI-driven proxy voting through Proxy IQ creates new conflicts of interest for a firm managing trillions in client assets while also voting those same clients' shares. Those are the numbers and disclosures worth watching as more of this spending shows up in year-end filings and next earnings season.