Analysis: The AI investment cycle has entered "overtime", with four indicators to alert to capital expenditure bubbles.

BCA Research’s investment strategist noted that the current AI investment cycle has entered "extra time", with the AI bubble more pronounced on the earnings front rather than the valuation side. Pressures stemming from profit growth, demand structure and supply expansion are likely to converge around 2027, at which point AI capital expenditure growth may slow significantly. Investors should closely monitor four key bubble warning indicators: GPU rental rates, AI storage chip prices, AI application penetration and corporate spending, as well as token prices and AI programming agent downloads. Among these, token price declines and stagnating downloads of AI programming tools are the most notable early signals, reflecting that enterprises are gradually shifting from pursuing the most advanced models to cost control. If the AI bubble eventually bursts, its impact on the U.S. economy and capital markets could exceed that of the 2000 dot-com bubble. According to estimates, U.S. stocks could see a 30% to 50% correction at that time, which would drag down U.S. consumption and economic growth via the wealth effect.