In August 2026, as the US stock Q2 earnings season enters its intensive disclosure phase, global investors' eyes are once again focused on the tech giant headquarters in Silicon Valley and Seattle. Over the past few years, the technological revolution represented by generative AI has not only reshaped the industrial landscape but also become the core engine driving the US stock market to repeated record highs. However, with the S&P 500 Index and the Nasdaq Composite Index approaching historical highs, the market narrative is undergoing a subtle and profound shift: investors are no longer satisfied with grand narratives and vision portrayals, but are demanding companies prove the rationality of their massive capital expenditures with hard revenue data and clear profitability paths. In this major test of the earnings season, the allocation of US core assets is facing the severe challenge of how to bridge the valuation and profitability gap.

1. The Earnings Season Spotlight: The "Hard Landing" Test of AI Capital Expenditure

Since the outbreak of generative AI in early 2023, the US tech giants' investment in AI infrastructure has been sparing no effort. From the frantic procurement of high-end GPUs to the global layout of hyperscale data centers, the capital expenditure curve has shown a steep upward trend. Entering 2026, the market has long become accustomed to capital expenditure budgets routinely in the tens of billions of dollars, but the accompanying voices of doubt have also grown louder: when will these investments translate into substantial free cash flow?

From the currently disclosed financial data and industry forward-looking guidance, the AI industry chain is experiencing a hard landing from "expectation-driven" to "performance-verified". For computing power providers at the core of the industry chain, although their data center business still maintains strong growth inertia, the marginal change in growth rate has become the focus of market gaming. Although some chip giants delivered better-than-expected revenue results, their next-quarter performance guidance failed to satisfy the appetites of some aggressive investors, causing stock prices to fluctuate violently after the earnings release. This volatility does not purely reflect a deterioration in fundamentals, but rather the market reassessing the margin for error in a high-valuation environment. When valuations embed perfect expectations, any minor flaw will be amplified by the market.

A deeper test lies with the cloud service giants. As the "landlords" of AI computing power, these enterprises bear the heavy responsibility of transforming high hardware costs into cloud service revenue. Financial reports show that while cloud business growth remains at a high level, the proportion of AI computing power in total revenue and its gross margin performance have become key indicators measuring AI monetization capabilities. Investors need to see not only overall cloud revenue growth but also evidence of scaled profitability in AI-related services. If massive depreciation and amortization begin to erode operating profit margins, the market's valuation models will inevitably face downward revision pressure.

2. Deep Deconstruction of the Industry Chain: Valuation Reconstruction from Computing Infrastructure to Application End

To understand the current allocation dilemma of US AI assets, we must deconstruct the entire AI industry chain. We can roughly divide it into three tiers: the infrastructure layer, the model and platform layer, and the application and service layer. Under the magnifying glass of the earnings season, these three tiers exhibit entirely different fundamental characteristics and valuation logics.

1. Infrastructure Layer: Coexistence of High Prosperity and High Gaming

The infrastructure layer mainly includes AI chip manufacturers, server assemblers, network equipment suppliers, and data center REITs. This tier represents the "water sellers" of the AI wave, and its performance has extremely high certainty during a high capital expenditure cycle. However, the 2026 earnings season revealed a new trend: the growth rate of capital expenditure is transitioning from an explosive phase to a stable phase. As the basic computing power reserves of hyperscale cloud clients reach a certain scale, future hardware procurement will rely more on upgrade and replacement demand rather than pure incremental expansion. This means that the valuation logic for infrastructure companies needs to shift from "growth stocks" to "cyclical growth stocks". When allocating such assets, investors must closely monitor companies' inventory turnover rates, new product iteration cycles, and customer concentration risks. Pure shipment volume growth is no longer sufficient to support high valuations; product mix optimization (such as an increased proportion of high-margin premium chips) and pricing power have become key to screening quality targets.

2. Model and Platform Layer: The Capital Expenditure "Gold Eater" and Moat

The model and platform layer is mainly composed of tech giants with self-developed large model capabilities. These enterprises not only need to procure massive computing power for model training but also need to maintain huge R&D teams for algorithm iteration. The capital expenditure scale of this tier is the most massive and is also the core of market debate. Data from the earnings season shows that although these giants' overall revenue maintains growth, the drag of AI R&D investment on short-term profits has begun to show. From the perspective of fundamental valuation, the traditional price-to-earnings (P/E) ratio appears inadequate when measuring these enterprises, as massive non-cash depreciation and stock-based compensation expenses distort current-period profitability. Investors need to pivot to metrics like Free Cash Flow Yield (FCF Yield) and Enterprise Value Multiple (EV/EBITDA). The core allocation logic is: only those enterprises capable of using AI technology to significantly improve their original core business efficiency (such as search ad conversion rates, e-commerce recommendation precision) and build a dual moat of data and computing power are worth continuing to hold at current valuation levels.

3. Application and Service Layer: The "Last Mile" of Profitability Realization

The application and service layer is the "last mile" of AI monetization, covering enterprise SaaS software, AI security, automated office tools, etc. This field is considered the biggest beneficiary of future AI valuation diffusion. However, the reality of the earnings season is relatively stark. Although many software companies have announced AI feature integrations, customers' willingness and ability to pay for AI add-ons still need verification. Financial reports from some enterprises show that while AI features have improved user retention rates, they have not yet translated into significant ARPU (Average Revenue Per User) increases. For assets in this tier, capital flows show clear differentiation: funds are concentrating toward leading software companies with massive customer bases and complex application scenarios, while small and medium-sized enterprises relying solely on "AI concepts" and lacking differentiated products face the pressure of capital withdrawal. Value investing stock selection is particularly crucial in this segment; investors need to look for enterprises with clear commercialization paths and replicable business models.

3. Undercurrents in Capital Flows: From Crowded Concentration to Barbell Rebalancing

In the face of earnings season uncertainty, US capital flows are undergoing profound changes. Over the past few years, the "crowded trade" characteristic of funds being highly concentrated in a handful of tech giants has been prominent. However, as the valuation and profitability gap emerges, institutional investors' behavioral patterns are shifting toward caution and balance.

First, actively managed funds are reducing their exposure to high-valuation AI core assets, instead increasing their proportion of cash or short-term Treasury bonds to cope with potential market volatility. Although passive index funds continue to see inflows, their internal structure is also undergoing minor adjustments. As some index compilers adjust weights, passive rebalancing of funds objectively mitigates the risk of over-concentration.

Second, capital exhibits a "barbell-style" flow characteristic within the tech sector. On one end are extremely robust tech giants with abundant free cash flow and strong shareholder return capabilities (such as massive buybacks and stable dividends); on the other end are sub-industry leaders whose valuations are at the bottom but whose AI applications have substantive breakthrough potential. The middle ground—i.e., those enterprises with non-low valuations, mediocre growth,勉强 told AI stories, and persistently unrealized profitability—is facing the risk of being marginalized by the market.

Furthermore, as the market repeatedly games the Federal Reserve's interest rate cut expectations, some funds are starting to flow out of high-valuation growth stock sectors into rate-sensitive industries and value sectors, such as utilities, healthcare, and consumer staples. This sector rotation does not mean the end of the AI narrative, but rather an instinctive reaction of the market guarding against risks at high levels, reflecting the self-protection mechanism of asset portfolio management in coping with cyclical volatility.

4. US Core Asset Allocation Strategy: A Practical Guide to Bridging the Gap

Facing the dense data and complex market gaming of the earnings season, how should investors optimize US core asset allocation? The following are practical strategies proposed based on current fundamentals and the valuation environment:

1. Adhere to the Core-Satellite Strategy, Build a Robust Base Position

Against the backdrop of the ongoing AI wave, completely stripping out tech stock allocation is unrealistic. Investors should adopt a "core-satellite" strategy. Core positions should be allocated to tech giants with deep moats, diversified revenue sources, and strong shareholder return capabilities. Even if these enterprises face short-term valuation pullbacks, their long-term free cash flow creation ability can provide investors with a margin of safety. Satellite positions can be moderately allocated to high-elasticity AI infrastructure or application-layer startups with disruptive potential (participated via ETFs or specialized funds) to capture excess returns.

2. Strengthen the Fundamental Valuation Anchor, Beware of the "P/E Trap"

During the earnings season, investors cannot rely solely on surface-level Earnings Per Share (EPS) figures. High capital expenditure means high depreciation and amortization, which will significantly depress EPS in later periods. Therefore, we must return to the essence of fundamental valuation, emphasizing free cash flow and Return on Invested Capital (ROIC). Vigilance should be maintained against enterprises with massive capital expenditure but continuously declining ROIC. Only when an enterprise's capital expenditure can translate into higher future profitability is the current investment worthwhile. Investors should look for turning-point enterprises where the growth rate of capital expenditure begins to slow, while revenue and profit margins begin to rise in sync.

3. Monitor Capital Flows, Implement Dynamic Asset Allocation

Capital flows are a thermometer of market sentiment and fundamental expectations. At the current stage, investors should closely monitor institutional holding changes and capital flow data, avoiding fighting against the main market trends. Dynamic asset allocation requires us to rebalance across different industry tracks. For example, if the earnings season shows AI hardware growth peaking while software applications begin to scale, the allocation focus should shift timely from hardware to software. Meanwhile, using derivatives like options to hedge the downside risk of core positions during the earnings disclosure window is also an important part of position management.

4. Broaden Horizons, Unearth Hidden Winners Empowered by AI

Besides directly participating in the AI industry chain, investors should also cast their gaze on traditional industries utilizing AI technology to reduce costs and increase efficiency. For instance, the fintech sector using AI for risk pricing and anti-fraud, the retail industry using AI to optimize supply chain management, and the manufacturing sector achieving smart manufacturing upgrades through AI. These "AI enablers" might not be on traditional tech stock lists, but their fundamental improvements tangibly benefit from the AI revolution. Including such enterprises in an investment portfolio can not only effectively reduce the portfolio's valuation bubble risk but also share the dividends of AI development across a broader dimension.

5. Conclusion: Staying True to Long-Term Value Investing Amid Short-Term Volatility

The 2026 Q2 earnings season is a critical review for US AI core assets. The valuation and profitability gap will not be bridged overnight, and market volatility will inevitably intensify with the collision of expectations and reality. However, from the long river of industrial development, the productivity leap brought by generative AI has just begun. For true value investors, earnings season volatility is not only a test of holdings but also an excellent opportunity to re-examine asset allocation logic and separate the wheat from the chaff.

Bridging the valuation and profitability gap requires investors to possess cycle-penetrating insight. Amidst the noise of short-term performance guidance, grasping those enterprises with genuine technological barriers, clear business models, and rationally capital-allocating management teams is the winning path for US asset allocation. In an uncertain market environment, fundamental valuation and long-term value investing remain investors' most reliable moats. Facing the major test of the earnings season, maintaining patience, adhering to discipline, and dynamically optimizing are the only ways to progress steadily and go far in the next AI-driven US bull market.