JPMorgan CEO: AI Has Cut Some Roles by 40%, but Won't Reduce Overall Operating Costs
Taylor Wilson
JPMorgan CEO Jamie Dimon said on the Q2 earnings call that AI has eliminated 30%-40% of positions in some business lines — yet he made clear the technology won't compress the bank's overall operating costs, because efficiency gains will flow to clients, not to margins.
Roles cut by a third — so why won't margins expand?
Dimon's core argument: in a competitive market, every bank will deploy AI to serve clients better. No one gets to pocket the efficiency gains alone.
His analogy: two decades of computerization should have pushed JPMorgan's margin to 80% by now if technology could unilaterally lift profitability — it obviously hasn't.
This means → AI in banking looks more like an arms race: fall behind if you don't invest, but investing doesn't guarantee extra profit.
Where did the displaced workers go?
Dimon said most employees replaced by AI were offered internal transfers, not outright layoffs.
In May he signaled the direction: the bank will hire fewer traditional bankers in some areas and recruit more AI talent instead.
In plain terms = the bank isn't shrinking — it's swapping traditional roles for AI roles.
Where does the nearly $20 billion tech budget go?
JPMorgan's tech budget stands at nearly $20 billion, with roughly 1,000 AI use cases deployed across fraud prevention, marketing, and meeting transcription.
A new cost line to watch: token fees — the per-use charges for calling large AI models. CFO Jeremy Barnum called the current figure "trivial" but forecast a meaningful acceleration in the second half.
This means → JPMorgan is shifting from "piloting AI" to "calling AI at scale," and token fees are the signal light for that inflection point.
How much did the bank earn this quarter?
Q2 net income hit $21.2 billion, up 41% year-on-year, partly boosted by gains on its Visa stake.
Every business line set a quarterly revenue record. Investment-banking fees rose to $3.3 billion, up 30% — the highest since 2021.
In plain terms = the earnings themselves are strong, but Dimon's remarks drew a ceiling for expectations: don't count on AI turning this growth into sustained margin expansion.
What does this signal for the market?
Dimon effectively capped expectations: AI in banking is defensive table stakes, not a differentiated profit driver.
This reflects a deeper point — when every competitor can access the same technology, it raises the industry's service level, not any single bank's moat.
The CFO framed the key challenge as "picking the right model for the right purpose." This means → the next frontier of bank competition isn't "having AI" — it's precision and efficiency in deploying it.
Content is for reference only, not financial advice.