AI Capital Spending May Reach $1.73 Trillion in 2026-2027
Six companies would need $1.42 trillion in revenue from 2028 through 2030 to meet a 15% annualized return target on AI computing investments.
U.S. hyperscale cloud providers are projected to spend about $1.73 trillion on AI capital expenditures in 2026 and 2027, raising the revenue required to justify the investment over the following three years.
Alphabet, Microsoft, Amazon, Meta, Oracle and SpaceX would need combined revenue of about $1.42 trillion from 2028 through 2030 to reach a 15% annualized return on invested capital from this AI computing investment. That equals roughly $11.6 billion in annual revenue for each gigawatt of computing capacity.
AI infrastructure capital expenditures totaled about $633 billion from 2023 through 2025. The figure is projected to rise to $1.73 trillion in 2026 and 2027, then reach about $4.14 trillion from 2028 through 2030.
Demand commitments have also expanded. AWS, Azure and Google Cloud had combined backlog of about $1.69 trillion as of the second quarter of 2026, an increase of about 152% from a year earlier.
The three cloud providers are expected to spend about $1.22 trillion on capital expenditures in 2026 and 2027. They would need roughly $1 trillion in revenue from 2028 through 2030 to meet the 15% return threshold, equal to about 59% of their backlog.
The scale of the upfront investment is expected to weigh on near-term returns, but the figures do not indicate a structural failure of the AI business model. Across return targets from 0% to 30%, required cumulative revenue for 2028 through 2030 ranges from about $908 billion to $1.89 trillion.
AI infrastructure capital expenditures are projected at about $1.3 trillion in 2027 and $2 trillion in 2028. New AI data-center deployments are expected to total 35 gigawatts in 2027 and 57 gigawatts in 2028.
Wider enterprise AI deployment and continued growth in cloud backlogs are expected to support future demand for AI computing capacity.
