Dimon: Hyperscale AI Capex Could Reach $1 Trillion Next Year
nashnova research
JPMorgan CEO Jamie Dimon says hyperscale AI capital spending has surged from roughly $300 billion to $700 billion this year and could top $1 trillion next year — making AI infrastructure a macro-scale force driving GDP growth and stoking inflation simultaneously.
How big is this spending?
Dimon's numbers: hyperscale cloud capex on AI climbed from about $300 billion to roughly $700 billion this year, with $1 trillion possible next year.
This means → AI infrastructure alone could add roughly 1% to annual GDP growth — a macro-level force, not just a tech-sector story.
In plain terms = this is no longer Silicon Valley spending on itself; it is enough money to shift the temperature of the entire economy.
Why does it boost growth and inflation at the same time?
Building data centers requires hiring workers, constructing factories and power plants, and buying equipment and raw materials — all demand-side pressure.
This means → in the short run, AI capex acts as both a growth engine and an inflation accelerator; the two effects are inseparable.
Dimon added that infrastructure buildout, rearmament, and persistent government deficits are also competing for capital, likely pushing interest rates higher.
Long-term deflation or inflation — what does Dimon think?
He called AI "an incredible technology" and said its rapid expansion "looks like it will continue," with a long-term deflationary potential.
But on near-term inflation, he was cautious: he hopes price pressures ease, yet "there is also a chance they don't ease or even rise slightly."
He stressed the Fed should hold to its 2% inflation target — this reflects his view that inflation risk has not been cleared.
How do you measure the return on AI investment?
Dimon argued that AI investment decisions cannot always be reduced to a direct return calculation — "sometimes it is just the price of admission."
In plain terms = think of a bank launching online banking — you cannot pinpoint exactly how much revenue it generates, but not doing it means falling behind.
He cited hard-to-quantify gains like customer-experience improvements and said deployment efficiency will rise over time.
Who will be the ultimate winners? Dimon says "too early to tell"
He invoked the dot-com bubble: many prominent companies of that era ultimately failed, while obscure newcomers emerged as winners.
This means → he is signaling that today's AI front-runners may not be the endgame winners — it is too early to bet.
He also warned that "markets could see a correction," though he was uncertain whether AI would be the trigger.
市场有风险，内容仅供研究参考，不构成投资建议。