热点思考 | AI资本开支：美国经济的“支柱”？——“无尽前沿”系列之二（申万宏观·赵伟团队）

Group 1 - The article discusses the significant rise in AI capital expenditure in the US over the past few years, which has strongly supported the US capital markets and economic growth. It questions whether this investment boom signals a "bubble" and how long the capital expenditure expansion cycle can continue [2][7]. - From a micro perspective, the capital expenditure of major US tech companies is approaching $100 billion by Q2 2025, doubling from three years ago with a year-on-year growth rate of 64.8%. The public's interest in AI technology and investment has surged dramatically [8][11]. - From a macro perspective, AI investment is expected to contribute 1.0 percentage points to economic growth in the first half of 2025, nearly on par with consumer spending, but the negative impact of imports on net investment cannot be overlooked [3][22]. Group 2 - The article highlights that while AI has shown some potential to boost productivity, the overall effect remains limited. The probability of the US being in a "low growth" phase for productivity is still high at 85% as of Q2 2025 [4][26]. - Historical comparisons indicate that the current AI investment, productivity, and cost performance are lagging behind the internet revolution of the 1990s, suggesting that the AI revolution is still in its early stages [4][46]. Group 3 - The article raises concerns about the sustainability of the current AI capital expenditure cycle, noting that while the financial metrics of major tech companies are stronger than during the internet bubble, potential headwinds include declining free cash flow, pressure on profits, and rising electricity demands [5][60]. - It suggests that the macroeconomic environment remains favorable for AI capital expenditure, with expectations of continued Federal Reserve rate cuts and potential economic recovery supporting the capital expenditure cycle [5][69].