China Securities Co., Ltd.: AI Investment Focus in 2027 to Shift Toward ROI Realization, Domestic Chips and Computing System Upgrades Remain Key Themes

Stock News11-10

According to a research report released by China Securities Co., Ltd. (601066), the A-share AI industry in 2026 experienced a rally driven by the popularity of "crayfish," rapid model iteration, and continuously increasing global capex. The computer sector went through a first-half trend of "rising capital expenditure driving expectations for sustained hardware demand growth," followed by a second-half trend of "rising application expectations."

Currently, domestic intelligent computing center construction is accelerating, the penetration rate of domestic AI chips in both training and inference continues to rise, AI applications are moving from proof of concept to vertical scenario deployment, model ARR and token growth are showing exponential increases, computing service prices are climbing, and the AI industry maintains a high level of prosperity.

Looking ahead to 2027, the main investment theme for the AI industry will shift from input expectations to ROI realization. The core industry variables will be the pace of domestic substitution and the commercialization validation of applications. The report recommends focusing on four directions: domestic computing power, AI applications and models, Xinchuang (domestic innovation), and future industries.

The main views of China Securities Co., Ltd. are as follows:

Computing Power Side: Seize Marginal Increments and Business Model Changes; Domestic Chips and Computing System Upgrades Remain Key Themes.

In 2027, overall domestic computing power investment is expected to maintain high growth rates. Three main themes deserve attention: First, mass production and delivery of domestic training chips, large-scale commercial deployment of super nodes, and comprehensive improvement in domestic substitution rates. Second, computing services shifting from "selling cabinets and GPUs" to "selling tokens," with token factories continuously driving down unit costs. Third, supporting segments such as high-speed interconnect, liquid cooling, power supplies, advanced packaging, and enterprise-level storage benefiting from the ramp-up of super nodes.

AI Applications and Models: Accelerating Penetration, Moving from Models to Deepened Industry Applications.

In 2027, four directions warrant attention: Personal and Work Agents landing in both B-end and C-end markets, with preference given to targets where AI-native business revenue accounts for over 15%; AI for Science entering pharmaceuticals, chemicals, chips, and other fields, driving R&D cost reduction and efficiency improvement; content creation such as animated dramas benefiting from multimodal and AI tools, with rapid scale expansion; AI Infra and security compliance becoming rigid enterprise demands, with the cybersecurity market concentrating toward leading players and AI penetration expected to trigger a new round of security demand.

Xinchuang (Domestic Innovation): Sprint Assessment Year, Industry-Wide Xinchuang Scaling Up.

Basic Xinchuang is expanding comprehensively from party and government sectors to central state-owned enterprises. Eight major industries—finance, telecommunications, power, petroleum, transportation, education, healthcare, and aerospace—are transitioning from pilot programs to large-scale procurement, with bidding and order confirmation concentrated in 2027–2028. Attention should be paid to the performance elasticity of database, industry software, and complete machine and peripheral companies.

Future Industries: The Arena of Technological Competition, with Capital Fueling Accelerated Industry Development.

Quantum technology is entering industry scenario pilots, with attention on commercialization opportunities in finance, pharmaceuticals, power grids, and other fields. Physical AI and world models are enhancing end-effector capabilities, with robots integrating with industrial and household scenarios, generating demand for simulation platforms, industrial software, edge inference, and robot bodies. In 2027, the focus will be on breakthroughs in real-world scenarios and commercial iteration, the improvement of planning and execution efficiency brought by large model companies entering the space, and the support of the data industry for world models.

Risk Warnings: (1) Macroeconomic downturn risk: The computer industry's downstream covers all sectors of the economy. Under macroeconomic downward pressure, lower-than-expected IT spending will directly impact computer industry demand. (2) Accounts receivable bad debt risk: Most computer companies operate on a project-based contract model, requiring acceptance before payment collection. Extended payment cycles from downstream customers may lead to increased accounts receivable bad debts and could further result in asset impairment losses. (3) Intensified industry competition: While computer industry demand is relatively certain, intensifying supply-side competition may lead to changes in the industry landscape. (4) Impact of changes in the international environment: The current international situation is volatile, which may impact companies with high overseas revenue exposure. Additionally, the United States continues to exert pressure on Chinese technology, creating supply chain security risks.

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Comments

We need your insight to fill this gap
Leave a comment