AI资本开支太狂热了？高盛：这才到哪呢

Core Viewpoint - Despite record nominal investments in AI infrastructure, the current investment level is not excessive compared to historical technology cycles, with AI investment in the U.S. accounting for less than 1% of GDP, while peaks in past cycles like railroads and IT reached 2-5% [1][7]. Group 1: AI Investment Sustainability - Recent capital expenditures in the AI sector have raised concerns about sustainability; however, Goldman Sachs' latest report indicates that the current scale of AI investment is not overheated and remains sustainable [3]. - Since mid-2023, AI infrastructure investments have accelerated, with U.S. companies projected to generate an additional $300 billion in AI-related infrastructure revenue by 2025 [5]. - AI-related spending has seen an annualized growth of $277 billion compared to 2022 [5]. Group 2: Productivity and Computational Demand - The report highlights two main reasons supporting continued AI capital expenditure: significant productivity gains and increasing computational demand [6]. - Goldman Sachs estimates that the widespread application of generative AI will enhance U.S. labor productivity by 15% over the next decade, with AI applications potentially delivering a 25-30% average productivity increase [6]. - The demand for computational power is growing at an annual rate of 400%, outpacing the cost decline of computational resources at 40% per year, indicating sustained investment motivation in AI infrastructure [6]. Group 3: Economic Impact of AI - Goldman Sachs projects that productivity improvements from generative AI could create a present value of $20 trillion for the U.S. economy, with $8 trillion flowing as capital gains to U.S. companies [7]. - Even under pessimistic or optimistic scenarios, the projected economic impact ranges from $5 trillion to $19 trillion, significantly exceeding current and future AI investment totals [7].