US AI Infrastructure Spending to Hit $1 Trillion by 2027, Leaving China Behind

Wall Street’s top forecasting firms are aligned: American companies will funnel more than $1 trillion into AI infrastructure by 2027. The projection, driven by private sector hyperscalers, places US spending at roughly ten times the level forecast for China.

Goldman Sachs projects US spending will reach $1.1 trillion, with potential to hit $1.4 trillion. JPMorgan has concurrently raised its cumulative global forecast for AI-related capital expenditure to $5.5 trillion through 2030.

The core spending is concentrated among Microsoft, Amazon, Alphabet, and Meta. Goldman Sachs forecasts hyperscaler capital expenditures of $757 billion in 2026, an 84% year-over-year increase, with chips and data centers as primary targets. Nvidia CEO Jensen Huang has noted that $1 trillion will be needed for AI chips alone by 2027.

In contrast, Chinese hyperscalers, including Alibaba and Tencent, are projected to invest a combined $84 billion in 2027. Chinese firms remain constrained by US export controls on advanced chips and a strategic shift toward operational efficiency.

JPMorgan’s $400 billion upward revision signals strong confidence in scaling AI workloads. Some Bitcoin mining firms are exploring a pivot toward AI workloads for more stable revenue, though this is separate from the core hyperscaler forecasts.