Dual Engines: AI Application Commercialization Wave and Semiconductor Industry Self-Reliance Enter New Phase

Bottom Line

This week's investment theme in China's IT sector is clearly defined as the "Acceleration of the Dual Engines: AI Application Monetization and Semiconductor Localization." Investors should immediately increase allocations to AI application-layer companies that have demonstrated commercialization capabilities and are generating actual revenue, while also focusing on hardware leaders in the semiconductor supply chain with clear capacity expansion plans and substantive technological breakthroughs. Avoid chasing AI concept stocks that rely solely on market sentiment without real revenue support.

Key Developments

1. AI Applications Achieve 'Nationwide Penetration,' Commercial Inflection Point Emerges — During the Spring Festival holiday, user numbers for leading tech companies' AI applications surged (e.g., Tencent's "Yuanbao" DAU exceeded 50 million), marking the first time China's large language model technology has achieved large-scale, society-level application, transitioning from technical validation to the stage of creating user value. Portfolio implication: Prioritize increasing holdings in internet platforms with massive user bases that can effectively convert AI traffic into revenue (e.g., companies within Tencent's and ByteDance's ecosystems), while reducing or avoiding pure-play AI technology companies still in the "cash-burning" R&D phase with unclear commercialization paths. Value capture at the application layer will precede the foundational layer.

2. Semiconductor Manufacturing Leader SMIC Sets Revenue Record, Capacity Expansion Continues — SMIC reported 2025 revenue of $9.327 billion, a year-on-year increase of 16.2%, and added 16,000 units of 12-inch wafer capacity in Q4, indicating strong domestic demand and steady progress in capacity construction despite a complex external environment. Portfolio implication: Increase holdings in leading companies in upstream core segments such as semiconductor manufacturing, equipment, and materials. SMIC's performance and expansion plans provide clear demand guidance for the entire domestic supply chain, benefiting its suppliers and peers with technological synergies. Focus on companies with healthy capital expenditure and capacity utilization in their financial reports.

3. AI Chips and Domestic LLMs Complete Deep Adaptation, Key Progress in Localized Ecosystem Closure — Domestic AI chipmaker Taichu Electronics announced that its chips have completed deep adaptation for mainstream large models like Zhipu AI's GLM-5.0 and Alibaba's Qwen3.5. Previously, Loongson also achieved localized deployment of AI models on purely domestic chips. Portfolio implication: This is a landmark event in building the "domestic computing power + domestic models" ecosystem. It is advisable to explore and position in AI chip design companies and system integrators with unique technological barriers in hardware-software co-optimization. Such companies will hold an advantage in future government and critical industry procurement.

4. Unprecedented AI Frenzy in Capital Markets, Soaring Valuations of Zhipu AI & MINIMAX Hint at Bubble Risk — Zhipu AI's stock price surged 500% within a month post-IPO, its market cap surpassing that of JD.com and Kuaishou; MINIMAX's year-to-date gain exceeded 450%. Extremely optimistic market sentiment has driven valuations severely detached from current fundamentals. Portfolio implication: This is a strong risk signal. Adopt an extremely cautious stance towards such star AI stocks with massive short-term gains; avoid chasing highs. Consider establishing hedge positions in the portfolio or shifting focus to "AI + traditional industry" transformation companies with relatively reasonable valuations and solid businesses.

5. Semiconductor Industry Consolidation Accelerates, M&A Targets Capacity & Tech Synergy — Huahong plans to acquire a 97.5% stake in Huali Microelectronics, and AMEC intends to acquire a 64.69% stake in Hangzhou Zhonggui Electronics. The trend of industry consolidation to optimize resources and capacity layout is clear. Portfolio implication: Monitor medium-sized semiconductor equipment, materials, and design companies that could become M&A targets or industry consolidators. M&A activity typically increases industry concentration and overall valuations. Study the industry chain map in advance to position in potentially attractive "small but beautiful" tech companies.

Sector Pulse

Risk Watch

• AI Concept Stock Valuation Bubble Burst Risk — Stock prices of some AI companies have seen enormous short-term gains, driven entirely by market sentiment and narrative, lacking profit support. A sharp correction is possible if performance fails to materialize or market risk appetite shifts. Probability: Medium. Impact: High.
• Execution Risk in Hardware-Software Co-development — Despite progress in adapting domestic chips to large models, large-scale commercial application still requires validation of their performance, stability, and cost competitiveness. Risks exist that ecosystem development lags behind promotional progress. Probability: Medium. Impact: Medium.

Outlook

Key events and indicators to monitor next week:

• Watch for more negative news from AI companies, such as product execution failures or commercialization setbacks similar to "Zhipu AI's apology letter," serving as leading indicators of cooling market sentiment.
• Pay attention to quarterly earnings or business updates from other semiconductor supply chain companies (e.g., equipment, materials) following SMIC's results, to verify the diffusion effect of industry prosperity.
• Monitor the subscription response and post-IPO performance of upcoming tech IPOs (e.g., Lyyun Software, Rongxin Software) in the Hong Kong capital markets, as an important window to assess primary market enthusiasm for the IT sector.

Positioning consideration: Adopt a "barbell strategy." One end should be allocated to leading companies with solid fundamentals benefiting from clear industry trends (e.g., semiconductor capacity expansion, AI+government services). The other end can involve small, exploratory positions in small-to-mid-cap companies with technological uniqueness within the domestic computing power ecosystem. Strictly avoid overvalued pure concept stocks.