Goldman Sachs: AI capital expenditure surges toward $1.7 trillion, requiring $1.42 trillion in revenue over three years to achieve a 15% return.

动察 Beating AI News Flash: According to the latest Goldman Sachs research, U.S. hyperscale cloud service providers' AI capital expenditure for 2026 to 2027 is projected to reach approximately $1.73 trillion. Based on a 15% annualized return on invested capital (ROIC) benchmark, six companies—Alphabet, Microsoft, Amazon, Meta, Oracle, and SpaceX—would need to generate cumulative revenue of approximately $1.42 trillion in 2028 to 2030 to cover the return requirements corresponding to this round of AI computing power investment, equivalent to approximately $11.6 billion in revenue per gigawatt of computing power per year.

Goldman Sachs divides AI computing power buildout into three phases: capital expenditure of approximately $633 billion in 2023 to 2025; approximately $1.73 trillion in 2026 to 2027; and a projected further rise to approximately $4.14 trillion in 2028 to 2030.

On the demand side, AWS, Azure, and Google Cloud had a combined backlog of approximately $1.69 trillion as of the second quarter of 2026, up approximately 152% year-over-year. Goldman Sachs estimates that the three cloud service providers' approximately $1.22 trillion in capital expenditure for 2026 to 2027 would require approximately $1 trillion in revenue for 2028 to 2030 to reach the 15% ROIC threshold, equivalent to only approximately 59% of the current backlog.

Goldman Sachs believes that the current phased pressure on AI capital expenditure returns stems mainly from large-scale upfront investment and does not mean there is a structural profitability problem in the AI economic model. Its estimates show that with ROIC targets ranging from 0% to 30%, the cumulative revenue required for 2028 to 2030 would be approximately $908 billion to $1.89 trillion.

In addition, Goldman Sachs expects AI infrastructure capital expenditure of approximately $1.3 trillion in 2027, rising further to approximately $2 trillion in 2028; the corresponding newly deployed AI data center capacity is expected to reach 35GW and 57GW, respectively. Goldman Sachs believes that the acceleration of enterprise AI applications from the experimental stage to actual deployment, as well as the continued growth of cloud service backlogs, will become important demand support for the monetization of AI computing power in the future.