BlockBeats News, July 3rd. According to JPMorgan's report "Fund Flows and Liquidity: The Demand for AI Rotation," since September last year, semiconductor stocks, namely AI chip and storage manufacturers, have continued to outperform and maintain a significant lead over hyperscale cloud service providers. This performance gap, in the long run, appears somewhat unsustainable. The report suggests that due to semiconductors being inherently part of a broader AI trade, the current differentiation has raised concerns in the market about its sustainability.

JPMorgan stated that this gap could narrow in two ways. The optimistic scenario is that as hyperscale cloud service providers, AI model providers, and users improve in commercialization, revenue, and profitability, their performance begins to catch up and gain a larger share in overall AI value-added. The pessimistic scenario is that if semiconductors continue to outperform at the expense of pressure on clients such as hyperscale cloud service providers, AI model providers, or end-users, it may dampen their willingness to spend capital and eventually create resistance to semiconductor product demand.

The report points out that JPMorgan's internal view leans more towards the optimistic scenario. However, analysts' consensus expectations indicate that the capital expenditure growth rate of hyperscale cloud service providers will significantly slow down starting from next year, which, if this expectation holds, aligns more with the pessimistic scenario. The report states that the capital expenditure growth rate of hyperscale cloud service providers is expected to reach 100% in 2026, but could drop to 22% in 2027 and further decline to 7% in 2028. If this deceleration path is confirmed, semiconductor trades may face considerable pressure, leading to a more significant and sustained correction in AI trades in the stock and bond markets.

JPMorgan also mentioned that the future AI computing power price will be key to whether hyperscale cloud service providers can commercialize AI capital expenditure. The higher the computing power price, the more capable cloud service providers are of maintaining or increasing profit margins. Additionally, the report stated that the U.S. money supply is expected to rise from $16 trillion in 2025 to $18 trillion in 2026, providing support for U.S. financial assets, especially U.S. stocks.