Deutsche Bank says a large portion of U.S. GDP growth is propped up by AI spending with no guaranteed return

A fresh estimate of third-quarter U.S. GDP growth is set to be released today. Analysts’ consensus forecast calls for a 3.2% year-over-year increase—solid growth by any measure. It’s little surprise, then, that the S&P 500 climbed another 0.88% yesterday, moving within half a percentage point of its all-time high, and futures are slightly higher this morning as well. Traders appear optimistic about the direction of the U.S. economy.

Yet a number of analysts are growing concerned about the extent to which this growth is concentrated in AI.

A recent report from Pantheon Macroeconomics stated that private fixed investment—an indicator of corporate spending—“is increasing solely because of AI-related expenditures.” This morning, analyst Oliver Allen shared a chart demonstrating that all non-AI private fixed investment is actually on the decline:

“Capital expenditure (capex) plans continue to be sluggish, indicating that investment in sectors unrelated to AI stays weak,” he informed clients in a note reviewed by .

echoed similar sentiments in a recent note exploring whether AI constitutes a bubble. Analysts Adrian Cox and Stefan Abrudan wrote: “AI sector investment is vital to GDP growth—without tech-related spending, the U.S. would be on the brink of recession this year, given that other forms of spending have stagnated post-pandemic.”

The volume of capital expenditure (capex) flowing into AI is enormous. Bank of America analysts Justin Post and Nitin Bansal project that AI capex from just five major “hyperscalers” (, , , , and Oracle) will reach $399 billion this year and exceed $600 billion in subsequent years.

More and more, this AI capex is expected to be financed through debt. According to BofA, large tech firms boast strong cash flows and sturdy balance sheets, making it feasible for most to take on additional debt without negatively impacting their profitability.

This debt is already setting records. Spencer Rogers and his team at Goldman Sachs recently told clients: “Net new debt issuance from AI-related entities in the U.S. dollar credit market surpassed $200 billion in 2025—more than twice last year’s total. AI-related issuances account for 30% of all U.S. dollar credit net supply this year.” Rogers anticipates this figure will increase further next year.

BofA notes that these companies are targeting an additional $1 trillion in revenue over the next five years. Approximately $500 billion of this is expected from cloud services, $400 billion from increased digital advertising, and $200 billion from AI subscriptions for both consumers and enterprises.

“Over the historical period 2021–2024, every dollar of capex contributed an average of $0.90 in additional revenue and $0.42 in extra EBITDA in the subsequent year,” the analysts stated.

Let’s cross our fingers this holds true. Deutsche Bank reports that hyperscalers will invest a total of $4 trillion in AI data centers by 2030—surpassing the U.S. government’s 1960s moon-landing initiative: “10 times the inflation-adjusted cost of the Apollo program, with no guaranteed payoff.”

Below is a quick look at market conditions ahead of the New York opening bell this morning:

  • S&P 500 futures are slightly higher this morning; the previous session ended with a 0.64% gain.
  • STOXX Europe 600 rose 0.18% in early trading.
  • The U.K.’s FTSE 100 remained unchanged in early trading.
  • Japan’s Nikkei 225 stayed flat.
  • China’s CSI 300 climbed 0.2%.
  • South Korea’s KOSPI gained 0.28%.
  • India’s NIFTY 50 held steady.
  • Bitcoin dropped to $87,000.