In August 2026, after a period of sharp correction in global tech stocks, capital markets have refocused on the artificial intelligence sector. Several foreign institutions have recently released intensive analyses of the second half of the year. Although their views vary in emphasis, they provide a highly consistent answer regarding one core issue: AI remains the central theme throughout the year, but the market is shifting from a phase of "mass euphoria" to one of "earnings validation."
Foreign Institutions Speak Up Unusually Frequently: Fundamentals Are Supported, Volatility Is Increasing
According to Xinhua Finance, affiliated with Xinhua News Agency, around August 6th, multiple foreign institutions, including UBS, Credit Suisse, LBP, and Morgan Stanley, concentrated on releasing their latest views on the recent market. These institutions generally believe that geopolitical situations and AI applications remain the two major themes throughout 2026. The global economy maintains resilience, but the characteristics of "K-shaped divergence" are evident in both the Chinese and US markets. Against the backdrop of slowing AI capital expenditure growth and global monetary policy divergence, foreign institutions have given a "cautiously optimistic" assessment of the equity markets in the second half of the year.
The LBP Asset Allocation Committee stated in its Q3 outlook that the AI capital expenditure cycle is continuously accelerating, having evolved into the core engine driving the current macroeconomic cycle. Its ripple effect is extending from data centers to electricity, utilities, industrial manufacturing, and semiconductors. This judgment implies that the investment value of AI has long surpassed the scope of mere technology sectors, becoming an "infrastructure-level" variable in macroeconomic growth—which is one of the most fundamental answers to "why invest in AI."
The "Gold Rush" Metaphor: Why Are the Companies Selling the Shovels Making Money First?
Li Changfeng, Head of Market Strategy at UBS, described the current AI investment landscape using the "gold rush" metaphor. The market has not yet found the optimal use case for AI "gold mines"; instead, companies that "sell the shovels"—such as those providing chips, power, and cloud computing power—have made significant money, driving a notable rally in the Asia-Pacific stock market in Q2.
Data supports this assessment: The ratio of capital expenditure to operating cash flow for large US technology companies may reach 98% by the end of this year; the year-on-year growth rate of capital expenditure is expected to slow from 78% this year to 41% next year and 9% the year after. In the view of UBS, the core reason for increased volatility in tech stocks in the second half of the year is the slowing "rate of improvement" in fundamentals—this is not a recession, just that the rate of ascent is no longer as strong as before. In other words, the fundamental base of the AI sector is still solid, but the slope of marginal improvement is moderating, and the market needs to digest valuations based on earnings rather than expectations.
The "Valuation Gap" in US-China AI Investment: Who Is More Sustainable?
Zhu Liang, Investment Director at UBS, pointed out that as of the end of June 2026, the internal valuation dispersion in A-shares has reached a historical extreme, stating, "This valuation divergence is like a rubber band; it cannot be stretched forever." In terms of AI investment, the differences in paths between the US and China are particularly noteworthy: the ratio of capital expenditure to cash flow for related US companies has risen to about 92% or even higher, whereas for related Chinese companies, it is only about 50%, which is a relatively healthier ratio and possesses stronger sustainability. More critically, the Chinese AI industry has developed a path of high cost-performance models, with stable capital expenditure expectations. Xing Ziqiang, Chief Economist at Morgan Stanley China, revealed that the token cost of domestic large models in China is about one-tenth of that in the US. By renting this cheap digital infrastructure through the national computing power network to tech enterprises, it is expected to replicate the development advantages of the mobile internet era. This cost advantage is becoming a crucial underlying logic supporting the large-scale implementation of AI applications in China, and it provides solid support for the medium-to-long-term investment value of the domestic AI industry chain.
From "Infrastructure" to "Application": The AI Investment Logic is Shifting
Xing Ziqiang believes that AI investment has entered a "half-time break" phase, and the investment logic should shift from "AI enabler" (building infrastructure capacity) to "AI adopter" (using AI to reduce costs and increase efficiency). In the second half, investment will focus on a broader AI ecosystem, including "HALO" assets (heavy assets with low obsolescence rates) such as energy transformation.
Latest research reports from CITIC Securities also express similar views: the significant pullback in the global technology sector over the past month is not due to a fundamental change in the vitality of the AI industry, nor has there been a downward revision in company earnings forecasts; the fundamentals remain high and tightly balanced. The pullback is more of a concentrated release of negative narratives, such as concerns over price increases, doubts about the return on capital expenditure, and the disruption caused by open-source models, compounded by a sell-off triggered by the over-crowding of AI hardware trading and the reversal of leveraged trading.
Looking ahead to August, CITIC Securities believes the sector is likely to enter a phase of "awaiting a new narrative and rebalancing of technology hardware and software." The improvement in open-source model capabilities is driving "intelligent parity" and "token equalization." Valuations are still low, and leading internet companies and cloud vendors with rich AI application ecosystems are worth paying attention to. After the hardware "shrinking," the expansion chain is superior to the price increase chain. Domestic equipment, wafer fabs, and optical communication sectors with low valuations that are trending upwards also have configuration value.
Why Invest in AI? The Three Long-Term Logics Remain Unchanged
Based on the analysis of multiple institutions, although short-term volatility has increased, the long-term logic supporting "why invest in AI" has not fundamentally changed:
Technology breakthroughs are still accelerating: The pace of AI technology iteration has not slowed down, moving from large model capability leaps to the implementation of multimodal and intelligent agents. The prosperity of the open-source ecosystem has further lowered the barrier to application innovation, continuously opening up industrial imagination.
Commercial value is beginning to materialize: North American cloud vendors maintain high capital expenditure and abundant forward orders. AI is moving from a "burn money" phase to a "make money" phase. Revenue realization at the application end will provide continuous positive feedback to the industry chain, validating the commercial value of AI rather than just the narrative.
Policy and industry resonance: China continues to strengthen support for AI industry development. Related companies have shown stellar interim earnings. Domestic computing power, from hardware to chip design, semiconductor equipment, and materials, to downstream complete machines and switches, is expected to achieve both volume and price increases, significantly enhancing the certainty of the self-reliant and controllable logic.
How to Invest in the Second Half? Balanced Allocation and Performance Verification are Key In terms of specific strategies, multiple institutions have provided highly consistent recommendations: first, avoid betting on a single track by maintaining a balanced allocation between technology hardware and software applications, and domestic supply chains and overseas expansion chains, to diversify valuation fluctuation risks; second, use performance verification as the core metric for stock selection, focusing on interim order and cash flow performance, and increasing positions in batches after receiving definitive signals; and third, pay attention to the "water sellers" opportunities stemming from AI spillover, including supporting segments such as power, energy storage, optical communication, and semiconductor equipment, which also benefit from the AI capital expenditure cycle. Analysts cited by CNR noted that the trend in the technology sector is not over, and the long-term logic of the industry chain remains strong; however, the market will place greater emphasis on performance verification moving forward. After risks have been fully released, true leading enterprises in technology manufacturing are expected to reach new highs, and sectors such as electronics, communications, innovative drugs, and power equipment are worth watching. Compared to overseas chains, the domestic AI industry chain, which is stronger in terms of self-control logic and certainty, offers better configuration value.
Conclusion: AI is not a short-term trend, but a long-term trend
The landscape of AI investment is becoming increasingly three-dimensional, ranging from the computing infrastructure of "selling shovels" to the AI applications of "gold mining," and the "HALO" resource support. The intense commentary from foreign institutions sends a clear signal: although the high growth of the AI industry may shift from "acceleration" to "steady growth," the certainty of growth has not disappeared. For ordinary investors, understanding the underlying logic of AI investment—technological breakthroughs, cost reduction, application diffusion, and value realization—is more important than chasing short-term fluctuations. This is the most worthy answer to the question of "why invest in AI": it is not a trend, but a long-term transformation reshaping productivity and industrial structure, which deserves to be planned and waited for over a longer time horizon.