Wall Street’s six largest banks cut 15,000 jobs and posted $47 billion in profits. The CEOs stopped pretending.

The six largest American banks shed 15,000 employees in the first quarter of 2026 while posting 47 billion dollars in collective profits, up 18 per cent year on year. The chief executives are no longer hedging. Jamie Dimon said artificial intelligence will eliminate jobs and that people should stop sticking their heads in the sand. Wells Fargo’s Charlie Scharf said anyone who claims AI will not reduce headcount either does not know what they are talking about or is not being totally honest. Goldman Sachs circulated an internal memo warning of job cuts and a hiring freeze under a programme called OneGS 3.0. Citigroup’s headcount fell by 2,000 in a single quarter. Bank of America’s Brian Moynihan credited AI directly for workforce reductions four months after telling employees they did not have to worry.

The hedging is over. The earnings calls are the data.

The numbers

JPMorgan Chase is spending 19.8 billion dollars on technology in 2026, up two billion from the previous year. The bank reports that productivity in divisions using AI has risen to approximately 6 per cent, double the rate before deployment. Operations roles have been trimmed by 4 per cent and support functions by 2 per cent, while positions tied to client engagement and revenue generation rose 4 per cent. Dimon described it as a huge redeployment. The bank’s headcount has held roughly steady at 318,500, but the composition has shifted. The humans who process are being replaced by the humans who sell.

Bank of America’s AI assistant Erica now saves the equivalent of 11,000 full-time positions. Nearly 90 per cent of the bank’s 213,000 employees use the Erica for Employees internal tool. The bank has also used AI to remove 30 per cent of the labour from its coding process, a saving Moynihan quantified as 2,000 engineering positions. Quarterly profit reached 8.6 billion dollars, 1.6 billion ahead of the previous year, and the bank explicitly attributed part of the gain to headcount reduction and automation.

Citigroup is executing the most aggressive restructuring of the group. The bank plans to cut roughly 20,000 positions by the end of 2026, approximately 8 per cent of its global workforce. When the Banamex retail banking business in Mexico is listed separately, total headcount will fall by approximately 60,000 to around 180,000. CEO Jane Fraser told staff in a January memo that the bank is not graded on effort and that AI and automation would allow it to run middle-office and operational functions with fewer people. Citi’s AI tools now reach 182,000 employees across 84 countries, with adoption exceeding 70 per cent.

Goldman Sachs launched OneGS 3.0 in late 2025, an AI-driven overhaul of its cross-bank operations programme. The initiative targets sales, client onboarding, lending, regulatory reporting, and vendor management for AI deployment. Annual staffing cuts have been moved to the second quarter, targeting a 3 to 5 per cent reduction. CEO David Solomon has predicted that Goldman will have more employees, not fewer, in the long term, but the internal memo contradicts the conference rhetoric. In the near term, there will be fewer.

Wells Fargo has shrunk from 275,000 employees when Scharf joined in 2019 to just over 210,000 by late 2025. AI-powered code generation tools have made engineering teams 30 to 35 per cent more efficient. The bank has not reduced the number of coders, Scharf said, but is getting substantially more output from the same headcount. Internal budgets already assume a smaller workforce for 2026. AI now generates instant borrower creditworthiness memos and pitchbooks for merger deals, front-office work that was historically performed by mid-level investment bankers earning six-figure salaries.

The pattern

The pattern across all five banks is identical. Profits are rising. Headcount is falling or being recomposed. AI productivity gains are measurable and specific, not hypothetical. And the CEOs, after two years of reassuring employees that AI would augment rather than replace, are now publicly acknowledging that replacement is happening.

Oracle is cutting up to 30,000 employees to pay for AI data centres, a restructuring that frees up an estimated 8 to 10 billion dollars in annual cash flow for infrastructure construction. The banks are running a version of the same trade. They are spending billions on AI technology and recouping the investment by reducing the workforce that the technology makes redundant. The difference is that Oracle is building data centres. The banks are building nothing. They are buying software that makes their existing operations require fewer people.

The productivity gains are concentrated in specific functions. Code generation, document processing, creditworthiness analysis, anti-money-laundering investigations, pitchbook assembly, regulatory reporting, and client onboarding are all partially or fully automatable with current AI tools. These are not entry-level tasks. They are the work of analysts, associates, compliance officers, and mid-level technologists, the professional-class roles that have historically been the economic backbone of financial services employment.

The spending

Projected AI spending per bank reached 177 million dollars in the first quarter of 2026, a 33 per cent increase from the previous quarter. Forty-two per cent of US financial firms plan to increase internal AI investment by 50 per cent or more. JPMorgan’s technology budget alone exceeds the total annual revenue of most enterprise software companies. Goldman Sachs has estimated that 25 to 30 per cent of its code could eventually be written by AI. The tools are not pilots. They are production systems operating at institutional scale.

More than 100,000 technology jobs have been eliminated globally in 2026, with AI cited as the driving force in at least one in five cases. The banking sector is not an outlier. It is the leading edge. Financial services operates in a regulatory environment that demands audit trails, compliance documentation, and risk controls, exactly the structured, rule-governed tasks that large language models handle well. The banks adopted AI faster than most industries not because they are more innovative but because their work is more automatable.

SAP unveiled its Autonomous Enterprise platform with more than 200 AI agents at Sapphire last week, targeting the same back-office processes that banks are automating internally. The enterprise software industry is building the tools. The banks are deploying them. The workforce implications are identical regardless of whether the AI runs on proprietary bank infrastructure or on a vendor’s platform. The functions disappear.

The question

The question is whether the 15,000 jobs lost in one quarter are the beginning of a sustained contraction or a one-time adjustment. The projections suggest the former. Industry estimates place potential global banking job losses at 200,000 within three to five years. Citigroup’s own restructuring plan implies 20,000 cuts at a single institution. Wells Fargo’s headcount has already fallen by 65,000 since 2019. The trajectory is clear and the CEOs are no longer pretending otherwise.

Dimon has suggested that AI could eventually reduce the working week to three and a half days. He has also said that JPMorgan has huge redeployment plans for displaced workers, offering them other positions within the bank. The redeployment narrative is genuine at JPMorgan, which has the scale and revenue diversity to absorb displaced staff. It is less plausible at smaller institutions that lack JPMorgan’s range of business lines and client segments.

Microsoft has offered voluntary retirement to US workers as part of its own AI-driven workforce restructuring, an approach that frames the displacement as a benefit rather than a loss. The banks are using a combination of attrition, redeployment, and targeted cuts. The language differs. The outcome is the same. Fewer people doing the same work, or more work done by the same number of people whose roles have been redefined around AI tools they did not choose and cannot opt out of.

The precedent

Banking is the test case that matters. If AI can replace human judgement in the most regulated, highest-stakes information industry on the planet, it can replace it in industries with less regulatory scrutiny, lower compliance burdens, and simpler workflows. The Q1 2026 earnings calls are the first quarterly data point where AI-driven workforce reduction moved from conference speculation to audited financial results. The profits are real. The headcount changes are real. The productivity metrics are specific and attributable.

Blackstone is preparing the first AI-era data centre REIT, with Goldman Sachs, JPMorgan, and Morgan Stanley underwriting the listing. The banks are simultaneously financing the AI infrastructure that is replacing their own workforce. They are on both sides of the trade, profiting from the capital markets activity that AI investment generates while using the same AI to reduce the number of people needed to execute that activity. It is the most Wall Street outcome imaginable.

The six largest American banks will spend more on AI this year than many countries spend on their entire technology sectors. They will also employ fewer people than they did last year. Those two facts are now connected by earnings call testimony, quarterly headcount disclosures, and specific productivity metrics that analysts can track. The era of speculating about whether AI will change financial services employment is over. The data is in. It already has.