JPMorgan Cuts 30% to 40% in Specific Areas with AI; Goldman Says It Won't Imitate
Dimon detailed the cuts in the Q2 conference call. Solomon from Goldman went on air saying that productivity does not need to reduce headcount. The question of who pays remains open.
Jamie Dimon told analysts on Tuesday, July 14, that artificial intelligence has already cut between 30% and 40% of the workforce in specific areas of JPMorgan. The revelation came during the second quarter earnings conference call, where the bank reported investment banking revenue of $3.3 billion, a 30% increase year-over-year, and a 41% jump in net income.
The CEO did not name the affected divisions but stated that most of the laid-off employees have been internally reassigned. "We fully expect to have enormous efficiency in specific parts of the company," Dimon said, according to the transcript of the call. He added, however, that the productivity gain should primarily benefit the client, not the bank's margin.
Jeremy Barnum, JPMorgan's CFO, elaborated on the real savings from the adoption. Costs associated with border models' tokens are still "trivial" in 2026 and are expected to grow significantly in the second half. Barnum pointed out the indiscriminate use of the more expensive models: "You really don’t need the absolutely expensive cutting-edge model to summarize an analyst report."
The Contrasting Voice: Goldman
On the other side of the table, David Solomon told CNBC on the same day that AI boosts employee productivity but should not affect headcount. Goldman delivered revenue of $20.3 billion for the quarter, a record for the institution, with investment banking at $3.4 billion and an announced M&A pipeline of $1 trillion for the semester. Solomon attributes the results to demand for financing data centers, semiconductors, and energy infrastructure. It’s the portrait of a bank selling tools while the other speaks of cuts.
The divergence has consequences. Solomon is currently the most visible voice on Wall Street arguing that additional productivity does not imply fewer people. His thesis rests on a reading of expansion: if client activity grows, AI absorbs the additional volume rather than eliminating jobs. Dimon takes the opposite view and acknowledges the substitution effect, although he minimizes the impact by saying that the gain will be passed on to the client.
Citi Falls 5% Despite Record Quarter
The market made its own verdict on the economics of AI in the sector. Citigroup surpassed all profit estimates for the quarter, reporting $25 billion, the highest revenue in a decade, along with a new $30 billion buyback program. The stock fell 5%. The reason was Jane Fraser's comments about anticipating cuts and investments in technology, which dilute the promise of operational efficiency in the short term. Fraser has been public about what she calls "two races of AI": an offensive one, for revenue and customer experience, and a defensive one, against fraud and financial crime.
Wells Fargo repeated the pattern. It exceeded consensus on revenue and profit, with $22.6 billion and an EPS of $2.00 against an expected $1.72, but the stock retreated 3% after Charlie Scharf signaled more investment in generative AI and automation.
What the Rest of the World Reads into This
In Mumbai, TCS and Infosys are observing closely. India hosts the largest captive BPO centers of major global banks, and the shrinkage of specific areas in New York, although offset by internal reallocation, tends to reach delivery hubs before it reaches headquarters. In London, the same thesis weighs on HSBC and Barclays, which are reporting results in the coming weeks with the same dilemma regarding where to declare the impact of AI. In Tokyo, MUFG and Mizuho have yet to publicly touch on the headcount issue, and that is the next conversation that sell-side analysts are anticipating.
What Dimon admitted and Solomon denied is the same question Deutsche Bank, UBS, and Nomura will have to answer in their upcoming conference calls. If the response aligns with Dimon’s, the next round of cuts in the financial sector has a defined target: back-office, operations, junior credit analysis. If it aligns with Solomon’s, the sector grows in volume without increasing headcount. Neither of the two scenarios offers comfort to research analysts who ask Barnum whether it’s worth investing in the more expensive model to summarize a note.
The paradox is that who pays the most for this story today is not the payroll but the shareholder. Dimon promises the gain to the client, Fraser anticipates expenses, Scharf reinforces capex. The margin Wall Street envisioned for 2025 in the 2027 balance sheet remains in the lap of those writing the next check to OpenAI and Anthropic.