The artificial intelligence investment boom is showing few signs of slowing, with spending across the hyperscaler ecosystem potentially reaching $1 trillion next year, JPMorgan Chase CEO Jamie Dimon said, highlighting the growing influence of AI infrastructure on the broader economy.
Spending by hyperscalers and companies across their infrastructure ecosystem has more than doubled from about $300 billion last year to roughly $700 billion this year, Dimon said. If that pace continues, investment could approach $1 trillion in 2027.
“That’s like 1% increase to GDP each year,” Dimon told CNBC-TV18 on the sidelines of the 11th annual JPMorgan India Conference.
He said the investment boom was supporting economic growth but could also add to inflation as companies hire workers, build factories and power plants, and purchase equipment and materials.
The scale of the spending is increasingly making AI more than a technology-sector story. The buildout requires semiconductors, data centers, electricity generation, transmission infrastructure, construction and specialized equipment, creating demand across multiple parts of the economy.
That spending can boost economic activity in the short term, but Dimon said the longer-term effect could be different. He described AI as an “unbelievable technology” and said its rapid expansion appeared likely to continue, while potentially producing deflationary effects as companies become more productive and efficient.
The tension between those two effects is becoming central to the economic debate around AI. The initial construction boom can increase demand for labor, energy and equipment, while the technology itself could eventually reduce the cost of producing goods and services by automating work and improving productivity.
For investors, however, the size of the spending does not guarantee that every company benefiting from the AI buildout will generate attractive returns.
Dimon cautioned against trying to identify the winners too early, drawing a comparison with the internet boom. Many companies that appeared well positioned during that period ultimately failed, while some less prominent businesses emerged as major long-term winners. That history is relevant to the current AI cycle because enormous amounts of capital are being committed before the industry has established which business models will ultimately capture the largest share of the economic value.
Dimon also pushed back against the idea that every AI investment must produce an immediately measurable financial return.
“Sometimes it’s just table stakes,” he said, arguing that companies may have to invest in AI simply to remain competitive.
He pointed to improvements in customer experience as an example of a benefit that can be difficult to quantify and said businesses could become more efficient at deploying AI over time.
That is believed to have resulted in a more complicated investment equation. AI spending can be economically necessary even when the direct return on a specific project is difficult to isolate. Companies may be investing not only to generate new revenue but also to reduce operating costs, improve products and prevent competitors from gaining an advantage.
At the same time, the scale of investment raises questions about the durability of the current capital cycle. Data-center construction, advanced chips and power infrastructure require enormous upfront commitments, while AI models and computing hardware continue to evolve rapidly.
Dimon said he was also watching several other forces that could keep interest rates elevated, including heavy demand for capital from infrastructure projects, remilitarization and government deficits.
He warned that there “may be a market correction,” although he said he was not certain AI would be responsible for it.
His comments come as investors continue to assess whether the rapid expansion of AI infrastructure can translate into sustainable earnings growth. The spending itself is measurable. The eventual economic return remains much harder to establish.
Dimon also maintained a cautious view on inflation. He said he hoped price pressures would ease but warned that “there’s a chance it won’t, and it may even go up a little bit,” while arguing that the Federal Reserve should remain committed to its 2% inflation target.
The inflation question could result in a broader discussion if AI infrastructure investment remains at its current pace. Data centers require large amounts of electricity, while construction projects compete for labor, equipment and materials. Those pressures can raise costs before productivity gains from AI become large enough to offset them.
Beyond the AI economy, Dimon said the United States and China appeared to be making progress ahead of a summit between President Donald Trump and Chinese President Xi Jinping. He said the two countries should “fully engage” on trade, AI, and security, describing the discussions as important to the global economy.
On India-U.S. relations, Dimon called for the two countries to return to negotiations and complete a trade agreement.
“It obviously hasn’t moved forward,” he said. “I hope it’s not put on the back burner.”
Dimon also said he understood U.S. concerns about India’s purchases of Russian oil but argued that Washington should consider India’s refining requirements and avoid measures that could hurt India and global oil markets.
His broader outlook on India was more expansive. Dimon said the Indian economy could grow to three times its current size over the next decade and confirmed that JPMorgan plans to continue expanding its operations in the country.
“We’re going to keep on building,” he said.
For the AI industry, however, the more consequential part of Dimon’s assessment is the sheer scale of capital now being committed. Moving from $300 billion in hyperscaler ecosystem spending last year to $700 billion this year and potentially $1 trillion next year would make AI infrastructure one of the largest investment cycles in the global economy.
