US forecasts $1T AI infrastructure spending by 2027, outpacing China by a factor of ten
Goldman Sachs and JPMorgan project hyperscaler capital expenditures will surge past $1 trillion, with Chinese rivals investing roughly one-tenth of that figure
Wall Street’s two most influential forecasting shops agree: US companies will pour more than $1 trillion into AI infrastructure by 2027. That figure, driven almost entirely by private sector hyperscalers, would dwarf Chinese spending by an order of magnitude and reshape the competitive landscape for everything from chip manufacturing to data center real estate.
Goldman Sachs pegs the number at $1.1 trillion in 2027, with upside potential stretching to $1.4 trillion. JPMorgan recently raised its cumulative global AI-related capex forecast through 2030 to $5.5 trillion, up from $5.1 trillion as of June 2026.
The numbers behind the trillion-dollar thesis
Goldman Sachs forecasts hyperscaler capital expenditures of $757 billion in 2026, which represents an 84% increase year over year. The spending is concentrated among Microsoft, Amazon, Google’s parent Alphabet, and Meta, with the primary targets being chips and data centers.
US entities account for approximately 80-85% of global AI and data center capex, along with a similarly dominant share of venture capital investments in the space.
Nvidia CEO Jensen Huang has estimated that at least $1 trillion will be needed in AI chips alone by 2027.
China’s spending gap is widening, not closing
Chinese hyperscalers, including Alibaba and Tencent, are projected to invest a combined $84 billion in AI infrastructure for 2027. That’s a 60% increase from 2025 levels, but roughly one-tenth of what US companies plan to spend in the same year.
Chinese firms face constraints including US export controls on advanced chips and a strategic pivot toward operational efficiency rather than raw expansion.
What this means for investors
JPMorgan’s decision to raise its cumulative forecast from $5.1 trillion to $5.5 trillion through 2030 signals growing confidence that AI workloads will continue scaling. That $400 billion upward revision alone is larger than most countries’ annual GDP.
Some Bitcoin mining firms have begun exploring pivots toward AI workloads, attracted by potentially higher and more stable revenue streams. That said, there is no direct connection between these mining pivots and the trillion-dollar infrastructure forecasts from Goldman and JPMorgan, which are squarely focused on hyperscaler spending.