Goldman Sachs: AI capital expenditure surges to $1.7 trillion, requiring $1.42 trillion in revenue over three years to achieve a 15% return According to the latest research by Goldman Sachs, the AI capital expenditure of US super scale cloud service providers is expected to reach approximately $1.73 trillion from 2026 to 2027. Based on an annualized return on investment capital (ROIC) of 15%, Alphabet、 Microsoft, Amazon Meta、 Oracle and SpaceX need to achieve a cumulative revenue of approximately $1.42 trillion between 2028 and 2030 to meet the return requirements of 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 construction into three stages: capital expenditures of approximately $633 billion from 2023 to 2025; Approximately $1.73 trillion from 2026 to 2027; It is expected to further rise to approximately 4.14 trillion US dollars from 2028 to 2030. On the demand side, AWS, Azure, and Google Cloud have a total backlog of orders of approximately $1.69 trillion as of the second quarter of 2026, a year-on-year increase of approximately 152%. Goldman Sachs estimates that three cloud service providers will have a capital expenditure of approximately $1.22 trillion from 2026 to 2027, corresponding to a revenue of approximately $1 trillion from 2028 to 2030, which can reach a 15% ROIC threshold, equivalent to only about 59% of the current backlog of orders. Goldman Sachs believes that the current temporary pressure on the return on investment of AI capital expenditures is mainly due to large-scale early-stage investments, and does not mean that there are structural profitability issues with the AI economic model. According to its calculations, with an ROIC target of 0% to 30%, the required cumulative revenue from 2028 to 2030 is approximately $908 billion to $1.89 trillion. In addition, Goldman Sachs expects AI infrastructure capital expenditures to be around $1.3 trillion in 2027, further rising to around $2 trillion in 2028; The corresponding new deployment scale of AI data centers 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 continuous growth of cloud service backlog orders, will become important demand support for the monetization of AI computing power in the future. [Original link]