Dimon: Corporate AI Spending Is Becoming More Rational
Claire Weston
JPMorgan CEO Jamie Dimon says companies are tightening AI budgets the way they manage any other resource, with systems already routing queries to the cheapest available tokens — a sign the era of uncapped AI spending is fading.
What exactly did Dimon say?
In a CNBC interview, Dimon was blunt: AI costs are rising fast, and companies "will of course treat it rationally, like any other resource."
This means → JPMorgan now manages AI spending under normal budget discipline — no more "special category, no ceiling."
He added that JPMorgan continuously negotiates with vendors when assessing AI value — big buyers are already using leverage to push prices down.
What is "routing to the cheapest token"?
Dimon observed that companies are already routing queries to the lowest-cost token and solution available.
In plain terms = companies used to give staff an "AI all-you-can-eat buffet." Now they read the menu — whichever model is cheap enough and good enough gets the job.
He expects the trend to extend to compute and data centers, covering the entire AI supply chain with cost scrutiny.
Are other leaders saying the same thing?
Palantir CEO Alex Karp said this month that AI models have been "oversold" and many U.S. firms are paying for worthless tokens.
Cerebras CEO Andrew Feldman put it bluntly last month: giving employees unlimited tokens is "like driving a Ferrari to the grocery store — stupid from the start." He urged a shift to low-cost open-source models.
This reflects a converging industry view: AI itself isn't the problem — spending on AI without discipline is.
How does JPMorgan view its own data?
Dimon stated clearly: JPMorgan is "very careful" with its data and intellectual property, and will do everything possible to protect client interests.
This means → data sovereignty is JPMorgan's red line in any AI vendor relationship — it won't trade core assets for AI access.
What does this mean for the market?
As corporate AI budgets shift from "spend freely" to "spend wisely," the most direct pressure falls on premium model providers — pricing power is moving to the buyer side.
In plain terms = profit distribution across the AI supply chain is being reshuffled. Vendors offering low-cost, good-enough solutions benefit; those relying purely on selling expensive tokens face margin compression.
Whether corporate AI spending can truly achieve efficiency gains is the key proof point for this round of AI investment returns.
Content is for reference only, not financial advice.