Looking ahead to 2026, CITIC SEC forecasts that computing power development holds high certainty, with hyper-node technology set for continuous upgrades and major CSPs maintaining robust capital expenditure growth, presenting attractive investment opportunities in computing chips and system-level manufacturers. Large model iteration is entering a more complex phase, with capabilities in areas like AI Agents and multimodality steadily improving, indicating that the inflection point for AI application adoption has arrived; concurrently, supported by overseas expansion and domestic demand policies, AI application companies are expected to see accelerated growth in orders and revenue. Regarding structural highlights, the construction of computing satellites is projected to gradually accelerate starting in 2026, speeding up the industrialization of space-based computing, while innovations in AI-powered pharmaceuticals and e-commerce continue to emerge, hastening real-world implementation. The firm recommends focusing on four key themes: 1) AI Computing Power; 2) AI Applications; 3) Computing Satellites; and 4) AI Healthcare. CITIC SEC's primary views are as follows: Computing Power Trend: Capitalizing on the inflection point of hyper-node technology, where system-level capability becomes the next critical competitive factor in computing power. The hyper-node era facilitates enhanced cluster performance, and leading overseas server companies have demonstrated the ability to capture超额 market share and profits through superior hyper-node system-level capabilities. Domestically, the competitiveness of China's homegrown hyper-node computing solutions is rapidly increasing, providing comprehensive support for local models, which is expected to lead to greater procurement in the hyper-node era. By 2026, competition in the domestic computing power sector is anticipated to shift gradually from a focus on single-card performance to system-level competition involving hyper-nodes, making system-level capability a crucial differentiator for manufacturers. Furthermore, against a backdrop of sustained capital expenditure acceleration by major CSPs and the ongoing release of token demand, the trajectory for computing power development remains highly certain. AI Applications: Enhanced model capabilities, combined with new overseas expansion opportunities, signal that the AI application inflection point has been reached, warranting a potential value reassessment. Next-generation large models (e.g., Gemini 3.0, GPT-5) offer substantial benefits for complex reasoning and multimodal scenarios, driving the AI industry toward large-scale, practical implementation. Domestically, major model developers are continuously enhancing their capabilities, while independent model providers with strong R&D prowess are distinguishing themselves. Previously, AI applications were primarily concentrated in areas like Chatbots, Coding, search, and customer service; the report suggests that as model capabilities improve, future applications will increasingly expand into broader areas such as multimodality, AI Agents, and embodied intelligence. From an international demand perspective, since 2025, overseas AI application Annual Recurring Revenue (ARR) data has shown consistent growth, indicating accelerated commercialization, prompting domestic AI application firms to expedite their global market presence to capture international market share, with overseas operations contributing significant new commercial value. Domestic Demand Support: Monitor ongoing policy support for domestic demand technologies in sectors like satellites, healthcare, and consumer goods. Since 2025, policies have consistently supported technology-driven domestic demand, and the firm expects this support to intensify further in 2026. In the satellite internet sector, increased policy backing for areas like satellite computing is driving demand for satellite technology companies; in healthcare, policies continue to focus on AI for public welfare and medical convenience, boosting demand for healthcare technology firms; in the consumer and fiscal sectors, policies aimed at stimulating consumption and standardizing fiscal systems are generating demand for consumer technology companies. Structural recovery in domestic demand is poised to be a key area of focus for the computer sector in 2026. Risk factors include potential disruptions in the computing chip supply chain; risks of insufficient chip production capacity; the risk of capital expenditure by major internet companies falling short of expectations; slower-than-expected iterations in chip technology; delays in the progress of large AI models and their commercial deployment; intensifying industry competition; lower-than-anticipated IT investments by governments and enterprises; risks associated with unsuccessful overseas expansion; slower development of satellite internet; and policy changes affecting the AI pharmaceutical e-commerce industry.
Comments