AI investment and the AI-driven stock rally now account for roughly one-third of US economic growth, a concentration that cuts both ways.
AI investment and the AI-driven stock rally now account for roughly one-third of US economic growth, a concentration that cuts both ways.

AI investment and the AI-driven stock rally now account for roughly one-third of US economic growth, a concentration that cuts both ways.
AI investment and the AI-fueled stock rally account for roughly one-third of recent US economic growth, leaving the expansion exposed if the boom cools, Oxford Economics estimates.
"It is very much an AI-driven economy right now," said Jonathan Millar, an economist at Barclays. "It's hard to imagine that we would be anywhere near as resilient without that impetus."
Business investment in AI-related categories — software, data-center construction, and computer and communications equipment — reached an annual rate of about $1.5 trillion, up from about $1 trillion two years earlier. June construction outlays on data centers hit $68.3 billion at an annual rate, up $21.5 billion from a year earlier, while outlays on all other private construction fell by $101.6 billion over the same period, the Commerce Department reported.
The concentration cuts both ways. If the boom fizzles, the economy's resilience would be threatened, and the debt markets funding the build-out are a potential vulnerability — Barclays projects bond issuance at five hyperscalers and SpaceX will reach $285 billion this year, up from $109 billion last year.
Not all tech spending went to AI — some went to computers on white-collar desks. But the build-out also pours money into other capital-spending categories, such as cooling systems and backup generators. A chunk of what goes into the build-out, including memory chips, is imported, so some AI spending doesn't directly lift US production. In the first five months of this year, the US imported about $90 billion in computer equipment, semiconductors and other electronic components from Taiwan, versus about $20 billion in the same period of 2024. Inflation also eats into the value of AI spending, as surging chip prices mean buyers get less for their money. Even so, Oxford's Pearce estimates AI investment alone has been behind nearly a quarter of GDP growth recently.
The AI-boosted stock market is the other big source of support. In the first quarter, US household net worth reached $174 trillion, up $13 trillion from a year earlier, mostly due to stock gains, according to the Federal Reserve. Since then, the broad S&P 1500 index has risen about 15 percent, so net worth is likely trillions of dollars higher now. When net worth rises, people spend more freely — economists generally think that for every dollar of stock-market wealth gained, people spend a few cents more. Higher-income consumers, who hold the bulk of stocks, benefit most. "Without this investment boom, I think it's pretty clear the economy would be running cooler," Pearce said.
For now, despite recent selling in some AI-related stocks, the spending boom looks set to continue. Analysts estimate capital spending at five hyperscalers — Alphabet, Amazon.com, Meta Platforms, Microsoft and Oracle — will reach nearly $4 trillion over the four years ending 2029, according to FactSet, more than $300 billion above estimates from about a month ago. Real trouble in the stock market could disrupt that story, as could problems in the debt markets funding the boom.
AI might also be sucking up resources that would otherwise go elsewhere. "You can't just take out AI and leave everything else unchanged," said Michael Feroli, an economist at JPMorgan Chase. "The activity and the financial exuberance associated with AI might be squeezing out activity." AI is also feeding into inflation, cutting into consumers' spending power — the iPhone 18 Pro coming in September might cost less if not for the jump in memory-chip prices.
The last time the economy leaned this heavily on a single investment cycle was the housing boom of the mid-2000s, which ended in a deep recession when it collapsed. The comparison is imperfect — AI spending is backed by corporate balance sheets rather than household mortgages — but it shows the stakes. "AI is touching so many things," Millar said. "A lot of the economy is kind of banking on continued strength from the build-out."
This article is for informational purposes only and does not constitute investment advice.