Hyperscaler AI Capex Could Add Up to 3% to U.S. GDP

Former White House AI and crypto czar David Sacks posted on X Sunday in response to a Morgan Stanley update on capital spending by Amazon, Alphabet, Meta, Microsoft and Oracle.

Morgan Stanley raised combined capex forecasts for those five firms to $805 billion in 2026 and $1.1 trillion in 2027. The firm projects the $805 billion planned for 2026 would be roughly double the prior year’s expenditures.

Sacks wrote that the Morgan Stanley numbers imply AI capital expenditure could add about a 2.5 percentage-point tailwind to U.S. GDP growth this year and more than 3% in 2027. He also wrote, “The ROI on capex is likely to dwarf the capex itself, which is why investment continues to grow,” and added, “in Q1, AI was already 75% of GDP growth.”

The report’s capex totals mainly cover infrastructure to run large AI systems, including data centers and related hardware. The estimates do not include spending by other cloud providers, enterprise buyers or startups.

Analysts and industry observers note the Morgan Stanley figures likely omit additional AI investment outside the five hyperscalers, and they point to potential productivity gains from broader AI adoption as a source of economic growth that capex alone does not measure.

Some critics compare the current wave of infrastructure spending to past technology investment cycles, citing risks such as overbuilding capacity, slower-than-expected adoption of new tools by companies and workers, and the large energy demands of AI data centers.

The concentration of reported investment among five technology firms raises questions about how broadly any economic gains will be distributed beyond those companies.

The Morgan Stanley forecast and Sacks’s comments present the recent capex estimates and differing assessments of how that spending could affect GDP growth.