JPMorgan: AI capital expenditures of 870 billion accelerate, with funding shifting from hardware to cloud giants

Deep Tide TechFlow news, according to Chao Research, JPMorgan Chase’s U.S. stock market strategy report expected on July 20 said that in 2026, global AI-related capital expenditures are expected to be close to $870 billion, up 77%. Of that, hyperscale vendors are expected to contribute about $750 billion. In 2027, the growth rate remains as high as Google +54% (about $300 billion), Amazon +42% (about $300 billion), and Meta +42% (about $200 billion). Bond financing by the five largest tech giants rose from $40–$50 billion in 2022 to about $190 billion in 2026, and Google completed an $85 billion equity financing in June.

JPMorgan believes the AI supercycle has not ended, but the investment logic is shifting from “competing on capital expenditures” to “competing on business monetization.” In the medium term, the risk-reward profile for hyperscale vendors is better than that of AI hardware, with the free cash flow inflection point occurring earliest in 2027. Semiconductor fundamentals remain healthy: this year, with growth expected to exceed 30% excluding memory revenue. However, the hardware sector has recently pulled back by about 15%, and although valuations are not yet cleared, volatility may persist. The energy sector’s market value makes up only 4% of the S&P, yet it contributes 23% to earnings growth.

On the timeline, Google will release earnings on July 22, Microsoft and Meta on July 29, Amazon on July 31, and Nvidia on August 26; capital expenditure guidance remains the core directional indicator.