Goldman Sachs Bombshell: Hundreds of Billions in AI Spending Delivered Zero GDP Growth
A bombshell Goldman Sachs research note concluded that despite hundreds of billions of dollars invested in AI infrastructure throughout 2025, the technology's measurable contribution to US GDP growth was effectively zero — shaking confidence in the AI investment supercycle.
Goldman Sachs Says AI's Trillion-Dollar Bet Has Not Moved the Economic Needle — Yet
It is perhaps the most uncomfortable finding in Silicon Valley's recent memory. Goldman Sachs, in a research note distributed to institutional clients Monday and quickly leaked to the wider financial press, concluded that the unprecedented wave of artificial intelligence investment that defined 2025 produced essentially no measurable contribution to US GDP growth.
Zero. After all of it.
The finding does not mean AI is a fraud or that the technology lacks promise. What it means, the analysts wrote carefully, is that the productivity gains from AI adoption have not yet appeared in the macroeconomic data — and that the timeline for those gains to materialize may be far longer, and the path far less linear, than the stock market has assumed.
The Gap Between Hype and Hard Data
The numbers behind the Goldman analysis are striking. In 2025, US technology companies, cloud providers, and AI-focused startups collectively invested an estimated $400 billion in AI infrastructure — data centers, GPU clusters, energy systems, and talent. The figure represents the largest single-year technology capital expenditure surge in American economic history.
And yet, when Goldman's economists ran the numbers against productivity data, GDP growth figures, and sector-by-sector output metrics, the AI signal was statistically indistinguishable from noise. The gains that companies report — efficiency improvements, cost savings, enhanced output — have not aggregated into the kind of economy-wide productivity surge that would show up in national accounts.
This is not entirely surprising to economic historians. The introduction of electricity took decades to show up in productivity statistics. The internet investment boom of the 1990s preceded measurable economic gains by years, and was accompanied by a spectacular market crash in between. But investors have not been pricing AI stocks for a decade-long adoption curve.
According to Dr. Daron Acemoglu, MIT Institute Professor and co-author of the landmark paper on AI's limited near-term economic potential, the Goldman findings are consistent with what the academic literature has been suggesting for two years. We have been automating a relatively narrow set of tasks. The transformative economic impact requires much broader capability deployment, which is still years away.
What This Means for the Tech Trade
The market implications are significant. The so-called Magnificent Seven technology stocks — Apple, Microsoft, Alphabet, Amazon, Meta, Nvidia, and Tesla — have collectively justified stratospheric valuations on the premise that AI monetization would arrive quickly and at scale. The Goldman note challenges that premise at its foundation.
Nvidia, whose GPU chips power the vast majority of AI training workloads, has seen its stock come under pressure despite record revenues. The concern investors are now voicing is not about Nvidia's near-term earnings — those remain strong — but about whether the companies buying those chips will ever generate returns sufficient to justify the investment. If AI spending plateaus or disappoints, the ripple effects through the tech sector will be substantial.
Goldman stopped well short of calling AI a bubble. The firm's analysts maintained that the technology remains likely to deliver significant economic value over a multi-year horizon. But the note's conclusion was sobering: the payoff is not as close as the market priced it, and the patience required to wait for it may test investors in ways that 2025's relentless AI rally obscured.
