AI资本开支太狂热了？高盛：这才到哪呢

Core Insights - The current scale of AI investment is sustainable and not overheated, indicating a robust macro story for AI infrastructure development [1][4] - AI-related investments account for less than 1% of the US GDP, significantly lower than historical peaks in other technology cycles [4] - The productivity gains from AI are projected to generate $8 trillion in capital income for US companies, far exceeding current and foreseeable AI investment totals [1][4] Investment Trends - Since mid-2023, there has been a significant acceleration in AI infrastructure investment, with an estimated $300 billion in revenue growth for US companies in AI-related infrastructure by 2025 [2] - AI-related spending has seen an annualized growth of $277 billion compared to 2022 [2] - Major investment agreements have been announced by OpenAI, including a $300 billion partnership with Oracle and a $100 billion investment from Nvidia [2] Supporting Factors for AI Capital Expenditure - Productivity improvements are expected to be substantial, with a projected 15% increase in US labor productivity due to the widespread application of generative AI over the next decade [3] - The demand for computing power is increasing rapidly, with AI model sizes growing at an annual rate of 400%, outpacing the 40% annual decline in computing costs [3] - The growth rates for training queries and cutting-edge models are 350% and 125% respectively, indicating sustained demand for AI infrastructure investment [3] Historical Context of AI Investment - Although nominal AI infrastructure investment has reached new highs, it remains modest compared to historical technology cycles, where peaks accounted for 2-5% of GDP [4] - The estimated present value of productivity gains from generative AI is $20 trillion, with $8 trillion expected to flow as capital gains to US companies [4] - Even under pessimistic or optimistic scenarios, the projected economic benefits from AI significantly exceed current and future investment totals [4]