Goldman Sachs Expects Hyperscale Data-Center Investment’s US GDP Contribution to Peak in Q4 2026
DataNews Editorial Team
Goldman Sachs expects hyperscale data-center investment to contribute about 1.5% to the year-on-year change in US GDP at its peak in the fourth quarter of 2026. The forecast comes as Amazon, Microsoft, Alphabet, Meta Platforms, Oracle and SpaceX are reported to be planning combined 2026 AI-infrastructure CAPEX of roughly $725 billion to $800 billion.
- Goldman Sachs forecasts a peak contribution of about 1.5% to the year-on-year change in US GDP from hyperscale data-center investment in Q4 2026.
- Reported 2026 AI-infrastructure CAPEX plans for six major operators total roughly $725 billion to $800 billion, up 64% from 2025.
- AI-related high-grade bond financing has exceeded $300 billion globally since the start of 2026.
- The projected GDP peak points to potential downside risk for data-center orders, construction and financing if the investment cycle subsequently normalizes.
- Announced or forecast CAPEX should not be treated as commissioned data-center capacity.
Goldman Sachs expects the contribution of hyperscale data-center investment to the year-on-year change in US GDP to peak at approximately 1.5% in the fourth quarter of 2026, according to comments attributed to chief economist Jan Hatzius.
The estimate coincides with a reported step-up in planned AI-infrastructure spending. Amazon, Microsoft, Alphabet, Meta Platforms, Oracle and SpaceX are expected to direct a combined roughly $725 billion to $800 billion of CAPEX toward AI infrastructure in 2026, a forecast 64% above the 2025 level. These are projected expenditures, not confirmed spending or commissioned data-center capacity.
Debt markets are also supporting the build-out. AI-related financing in the global high-grade bond market has exceeded $300 billion since the start of 2026, according to the article. In the US, nine bond issues of at least $25 billion each have been recorded in 2026; apart from a Skydance Corp. acquisition-financing transaction, the article attributes the remaining issues to technology companies.
For data-center operators, infrastructure suppliers and investors, the key issue is whether this level of capital formation can be sustained beyond the expected peak in its GDP contribution. Goldman Sachs had previously assessed AI investment as making a virtually nil contribution to US GDP growth in 2025, citing the offsetting effect of equipment imports.
A deceleration in the investment cycle could affect the pace of future orders, construction activity and financing demand across the data-center supply chain. Goldman Sachs estimates that the end of the data-center investment boom could reduce the level of US GDP by about 2 percentage points. That is a scenario-based analytical estimate rather than an observed outcome, and should not be read as evidence that the projected CAPEX will translate directly into deployed capacity.
The figures underline the scale of the current AI-infrastructure investment cycle, while also highlighting the risk of a cyclical slowdown after the projected peak. For data-center markets, the durability of hyperscaler CAPEX, access to debt capital and the potential normalization of demand will be central considerations.
- — / October 27, 2026 · ChineseSource →
This article was prepared by the DataNews editorial team based on the sources listed above.