Written by: Rita
Tide Guide
On July 20, 2026, J.P. Morgan’s strategy team released a report noting that the U.S. AI sector has recently shown clear differentiation: hardware stocks such as NVIDIA and Broadcom have experienced increased volatility with repeated price swings, while Microsoft, Google, Meta, and Amazon have exhibited more stable trends.
The market is concerned whether the AI rally has peaked. J.P. Morgan’s latest U.S. equity strategy report provides a clear assessment: the AI supercycle is not over, but capital is rotating. The investment thesis is shifting from “competing on capital expenditure” to “competing on commercial monetization.” While annual AI capital expenditures of $870 billion continue to accelerate, the market no longer settles for “scale of investment” and is now demanding “profit realization.” J.P. Morgan believes that hyperscale vendors with cloud platforms, data centers, and AI application ecosystems offer a better risk-reward profile than AI hardware providers.
Capital expenditures are still accelerating.
The key focus of this quarter's earnings report is not on profits, but on AI capital expenditures.
The development of the AI industry chain heavily relies on continuous investment by tech giants in data center construction. The market expects global AI-related capital expenditures to reach nearly $870 billion by the end of 2026, a 77% year-over-year increase, with hyperscalers contributing approximately $750 billion.
JPMorgan Chase's internet analyst believes the market's projections for 2027 remain overly conservative. Google's capital expenditures in 2027 are expected to increase by 54%, reaching nearly $300 billion. Amazon's spending is projected to grow by 42%, to approximately $300 billion. Meta's spending is forecast to rise by 42%, to about $200 billion.
AI infrastructure investment has not slowed; it is still accelerating. The upcoming quarterly earnings reports from the four major hyperscalers over the next few weeks will be the market’s key indicator.
The market is beginning to ask: When will we make money?
Over the past two years, the market focused solely on the scale of investment. Now, investors are turning their attention to the core question: When will large-scale investments deliver returns?
J.P. Morgan believes this will be a central theme in AI investment over the coming years. Positive signals are already emerging. Meta is selling its excess AI computing power, directly monetizing its data center resources. Anthropic’s progress in safety research suggests that security concerns may accelerate the migration of enterprise and government workloads to the cloud, replacing on-premises model deployments.
This means demand for cloud services such as Azure, Google Cloud, and AWS will continue to grow. The AI business model is gradually being validated, encompassing diverse revenue streams including model sales, computing power output, cloud services, and software licensing. J.P. Morgan believes that if management releases more signals regarding AI commercialization in earnings reports, profit forecasts and free cash flow could be revised upward.
Money is flowing from hardware to cloud giants.
Over the past two years, the biggest winners in the AI sector have been in the upstream segment, with nearly all categories—including GPUs, HBM, high-speed networking, switches, and optical modules—rising almost universally. However, J.P. Morgan believes a sector rotation is now underway.
AI hardware represents a classic momentum trade: highly concentrated positions and extremely crowded trades. Any shift in expectations will significantly amplify volatility. The semiconductor sector has recently corrected by approximately 15%; while crowding has eased, it has not been fully unwound. Market concerns center on four key points: the sustainability of capital expenditures, fundraising by AI companies diverting capital, the supply-demand dynamics of infrastructure, and whether return on invested capital can cover costs.
In comparison, Microsoft, Google, Meta, and Amazon offer a better risk-reward profile. These companies control computing power, cloud platforms, and end users. The true profit generators may not be the "shovel sellers," but rather the "mining operators."
Where does the money come from?
Funding sources have become a new concern for the market: Can tech giants’ cash reserves sustain annual investments of hundreds of billions of dollars?
JPMorgan Chase has specifically responded to this. The bond financing volume of the five major tech giants is expected to rise from $40–50 billion in 2022 to approximately $190 billion by 2026. Google completed $85 billion in equity financing this year, and Meta is also evaluating similar options.
However, JPMorgan believes this is merely a restructuring of financing, not an indication of deteriorating operations. It is expected that by 2027, tech giants will still generate over $900 billion in operating cash flow, with annual revenue growth of approximately 17% over the coming years and profit margins remaining high. The market is willing to provide financing support for AI development, as the core rationale lies in confidence in long-term profitability.
Regarding the free cash flow inflection point, JPMorgan expects it to occur as early as 2027. If the commercialization of AI slows down, significant improvement may not occur until after 2028.
The semiconductor logic has not changed.
JPMorgan is not bearish on AI hardware.
Demand from hyperscale manufacturers remains strong, with continued gaps in compute capacity for AI labs and rapidly growing demand for agent inference. Data center investments continue to rise, and new platforms such as Blackwell and Rubin further enhance visibility into demand over the coming years. The focus of market discussions has shifted: previously centered on the "authenticity of AI demand," it now centers on whether supply chains and power availability can keep pace.
Beyond GPUs, AI demand has spread to areas such as custom chips, HBM memory, high-speed networking, semiconductor equipment, and EDA. This year, global semiconductor industry revenue (excluding memory) is still expected to grow by more than 30%. J.P. Morgan continues to be bullish on the AI value chain, viewing only a shift in investment timing.
Second-quarter profits remain strong.
JPMorgan Chase is generally optimistic about the overall Q2 earnings season for U.S. equities. S&P 500 earnings for Q2 are expected to rise 23% year-over-year, or 19% excluding energy, with revenue growth of 12%. Revenue increased across nearly all sectors, and profits rose in most sectors, except healthcare.
However, the growth structure is highly concentrated. Just NVIDIA and Micron account for approximately 37% of the S&P 500’s earnings growth. Currently, U.S. stock earnings growth remains primarily driven by AI.
The energy sector, driven by rising oil prices, is expected to see a 122% profit increase, another key highlight of the quarter. Healthcare is the only sector forecast to experience a decline in profits, but J.P. Morgan still sees strong long-term allocation value in it, particularly given the advancements in AI-powered drug discovery and improved healthcare efficiency.
Tide View
This report examines the two key variables in the AI investment chain side by side: the capital expenditure slope and the time lag to monetization. An annual expenditure of $870 billion corresponds to a financing curve where debt issuance by the five largest hyperscale vendors has risen from $40 billion to $190 billion; the steepening of this curve itself is accumulating risk.
JPMorgan Chase estimates the free cash flow inflection point may emerge as early as 2027, but acknowledges that a more definitive recovery will likely require waiting until 2028 or later. This time window carries a key bet: if cash conversion lags behind expectations in 2027, market patience for a 2028 recovery could be exhausted prematurely.
The impact of private investment fair value changes on EPS is often overlooked. Approximately $4 of S&P 500 EPS in Q1 came from non-cash revaluations, and this magnitude may expand further in Q2. A significant portion of earnings revisions is driven by accounting effects, and the actual operational improvements should be assessed cautiously.
Disclaimer
This article is a compilation and interpretation by Chaoxiang Research of a third-party brokerage research report (JPMorgan, July 20, 2026). The ratings, price targets, earnings forecasts, and related judgments cited herein reflect the views of the brokerage’s analysts and represent the position of their respective institution, not the views of Chaoxiang Research, nor do they constitute any investment advice. The market carries risks; decisions must be made independently. This article should not be used as a basis for buying or selling any securities.