In its latest research report, JPMorgan Chase stated that with AI companies' revenue exceeding expectations in the second quarter, the economic sustainability of the AI capital expenditure cycle has significantly improved compared to six months ago.
The investment bank's internal credit research projects AI-related capital expenditures to reach approximately $5.5 trillion between 2026 and 2030, with external forecasts reaching as high as $10 trillion. Based on the median estimate of approximately $7.5 trillion, AI cloud, model providers, and new AI cloud service providers (Neocloud) could generate $1.6 trillion in revenue by the end of 2026. If this figure increases by 10%-20% annually thereafter, revenue could rise to $2.5 trillion-$3 trillion by 2030.
Demand is primarily driven by enterprises. Large enterprises' AI spending as a percentage of their total expenses plus capital expenditures is expected to rise from 4.5% in the past 12 months to 5.8% in the next 12 months. Considering the approximately $30 trillion in expenses and capital expenditures of large global enterprises, this corresponds to approximately $1.7 trillion in AI spending.
Therefore, JPMorgan Chase believes that the pessimistic argument that "AI infrastructure far exceeds its monetization potential" has weakened, but the actual returns after considering profit margins, depreciation, utilization rates, pricing power, and payments within the industry chain still need to be verified. (The above views are from JPMorgan Chase's report dated August 20.)
