Skip to content
Advertisement
Goldman Sachs warns AI’s boost to US GDP growth will fade

Goldman Sachs warns AI’s boost to US GDP growth will fade

Marcus by Goldman Sachs / Wikimedia Commons (Public domain)

AI investment has carried most of America's recent growth, but Goldman expects that tailwind to weaken next year

Goldman Sachs puts current AI-related capital expenditures in the US at around $600 billion per year. That covers the physical buildout behind the AI boom: chips, hardware and the facilities that house them.

The bigger checks are expected from the hyperscalers. These are the cloud giants, including Amazon, Microsoft and Alphabet, that rent out computing power at enormous scale.

Goldman projects their total capital expenditures could reach between $725 billion and $800 billion by 2026. Spending is not the issue. Growth in spending is.

The bank expects the increase in hyperscaler investment as a share of GDP to peak at approximately 1.5% in Q4 2026. After that, Goldman sees it declining.

Advertisement

Goldman’s forecast puts the peak of the growth impact from hyperscaler AI investment somewhere between late 2026 and early 2027. The money may keep flowing at elevated levels, but the incremental push to GDP is expected to shrink.

That distinction matters. GDP growth responds to changes in spending, not just levels of it. A company that spends the same huge sum two years running adds plenty to output, but it adds little to growth, because growth measures acceleration.

The import problem and the skeptic’s view

Goldman’s chief economist, Jan Hatzius, has voiced skepticism about how much AI capex actually feeds into US GDP. His argument centers on where the money goes.

A large share of AI spending is tied to imports of chips and hardware. When a US company buys equipment made abroad, the investment shows up in the accounts, but the import subtracts from GDP on the other side of the ledger.

On that measure, Goldman has estimated that AI spending lifted US GDP growth by only about 0.1 percentage points, with some periods showing basically zero impact.

Earlier claims had suggested AI capex accounted for as much as 30% to 90% of US economic growth.

A long game, not a sugar rush

Goldman describes the buildout as a multi-decade structural shift rather than a temporary spike. Significant productivity gains are expected to arrive as businesses actually adopt the technology across their operations over the coming years.

What to watch next: hyperscaler capex guidance from Amazon, Microsoft and Alphabet, which will show whether spending tracks the $725 billion to $800 billion range Goldman projects for 2026.

The other signal is the GDP data itself. If the AI share of growth starts slipping toward the Q4 2026 peak Goldman has flagged, the question shifts from how much AI is adding to what, exactly, replaces it.

Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.