AI investment logic is reshaping; the new main investment theme is gradually becoming clear

Securities Times Reporter Wu Qi, Liu Yiwen, Tan Chudan

Recently, investment narratives around artificial intelligence (AI) applications have shown clear divergence.

Since the beginning of this year, stocks in U.S. sectors such as cybersecurity, fintech, and human resources services have generally declined sharply. The market’s extreme divergence expectations for AI have been transmitted outward, dubbed as “AI ghost stories.”

As a result, the Hang Seng Tech Index has experienced significant adjustments, while funds in A-shares have concentrated in AI industry chain beneficiaries and traditional industries less affected by shocks, showing obvious risk-avoidance and structural groupings.

AI Panic Sentiment Spreads to Global Markets

Many investors are concerned about how much impact the “AI ghost stories” in the U.S. will have on other global markets.

Liu Yang, Deputy General Manager of Shenwan Hongyuan Research, and Huang Zhonghuang, Chief Computer Analyst, told reporters that the impact of “AI ghost stories” on the U.S. market is relatively greater, while China’s industry complexity is high, with many vertical, semi-customized, and semi-general scenarios. The influence brought by AI Agents is more beneficial than harmful.

Lin Qingyuan, fund manager of Ping An Xin’an Hybrid Fund, believes that the panic in U.S. stocks is not only not negative for A-shares but is actually a long-term relative positive. The underlying asset structure of A-shares’ tech sector mainly revolves around “hardware manufacturing,” which naturally aligns with the current macro trend of risk-averse investment in “heavy assets and hard technology.”

Hong Jiajun, Assistant General Manager of the Overseas Research Center at Industrial Securities’ Economics and Financial Research Institute and Chief TMT Analyst, believes that the recent “AI ghost story” style adjustment in U.S. stocks is a re-pricing of sentiment and valuation, not the end of the AI industry cycle. This fluctuation mainly transmits short-term sentiment in China’s AI industry, while in the medium to long term, it reinforces the logic of independent controllability. For A-shares and H-shares, it results in valuation differentiation and sector selection.

Redefining the Valuation System of Listed Companies

2026 is seen as the decisive year for global artificial intelligence. Since the start of the year, the market’s AI investment landscape has shown extreme divergence. The deep reshaping of business models by AI and the redefinition of the valuation system for listed companies are becoming increasingly clear.

Wang Dongsheng, Fund Manager at Dinsa Investment, analyzed that some industries’ business models have been significantly squeezed by AI, and the benefiting industries are gradually becoming clear. For investment institutions, the current focus is on targeting high-certainty directions and making clear strategic allocations.

From a research perspective, considering AI’s differentiated impact on vertical software moats, Yang Shangdong, Chief Analyst of Media Industry at Industrial Securities’ Economics and Financial Research Institute, divided the AI benefit chain into three: the benefit chain, the fragile chain, and the immune chain. The benefit chain mainly includes two types of companies: one is those with exclusive, scarce data that AI further enhances their pricing power; the other is AI infrastructure and ecosystem builders, as well as native vertical AI tool providers, which are expected to become industry aggregators and earn core profits. The fragile chain consists of traditional vertical software “search layer” companies, whose moats are directly destroyed by AI, with core value commodified and pricing power collapsing. The immune chain includes companies with regulatory compliance locks, embedded in transaction processes, and possessing strong network effects, which have structural moats that AI cannot break through in the short term, and AI cannot replace their core value.

However, it is worth noting that many institutions believe that the current AI panic-driven adjustment involves both sectors that have been mispriced and industries facing fundamental logical disruption.

New Investment Themes Are Becoming Clear

After this round of AI panic, the new investment themes in the market are gradually becoming clear, and the investment logic is also undergoing reshaping.

Zhai Sen, Manager of Ping An Technology Innovation Hybrid Fund, emphasized that the impact of AI is not the disappearance of industries but a profound restructuring of the value chain. Demand still exists, but value will shift from intermediary and execution links to upstream computing power, models, data, and downstream applications with real scene and closed-loop capabilities.

In response to the disruptive effects of AI, Lin Qingyuan suggests abandoning illusions about fragile business models and focusing on entities with truly high reset costs and strong moats. AI can instantly rewrite SaaS code and replace consulting analysis but cannot create transformers, power plants, or copper mines out of thin air. Therefore, the best way to avoid risks is to reduce allocations to lightweight assets whose moats are based solely on “labor-intensive” or “shallow digital arbitrage.” The dividends of the AI revolution are shifting toward upstream and foundational physical support in the industry chain.

Wu Mingyuan, Chief Computer Analyst at Huachuang Securities, is optimistic about directions such as large model companies, MaaS as a disruptor—recently proven to be an attacker rather than a defender—, as well as sectors that serve as model infrastructure and benefit from model development, such as AIDB (vector databases), data elements, and vertical AI applications including AI + industry, AI + fintech, AI + healthcare/AI4S (scientific AI), AI + education, and AI + ERP (Enterprise Resource Planning).