Overview
On September 29, the Generative AI Application Development Report (2026) was released at the China Internet Fundamental Resources Conference. By mid-2026, China’s generative-AI users surpassed 700 million, a penetration rate above 50%. Smart Q&A leads usage (76.0%), followed by image/video (47.8%), text (37.6%) and office tasks (32.5%). In the first half of 2026, AI-related investment reached 1,255 deals worth about 250.1 billion yuan — 182% of 2025’s full-year amount. Intelligent computing hit 2,185 EFLOPS, up 177% year-on-year; the first fully domestic 100,000-GPU cluster went live. Over 400 humanoid robot models exist, with annual output projected above 100,000 units.
Background & Interpretation
1. Technology Background and Evolution
Large models are moving from “can chat” to “can act”, with multimodality, agentization and edge deployment as the main vectors. Multiple trillion-parameter open-source models launched; Chinese teams hold the top six spots on major global model-calling leaderboards.
2. Core Drivers and Mechanisms
User growth is driven by usability plus scenario necessity: AI assistant and office-tool usage more than doubled year-on-year. On the supply side, domestic super-clusters and 120,000+ high-quality datasets lower training and inference costs, feeding back into adoption.
3. Market Structure and Industrial Chain Effects
AI now permeates R&D, manufacturing and QA across manufacturing (30%+ adoption). Agri-models and embodied intelligence reach industry, healthcare and services. Heavy investment signals confidence in commercialization but also valuation-vs-delivery risk.
Implications & Outlook
Generative AI is becoming daily infrastructure. The next battleground is agent workflows, vertical models and edge inference; the compute-model-application flywheel accelerates.
Takeaways for Industry Participants
Firms should map high-ROI AI use cases in R&D, marketing and ops, starting with office and service agents. Compute and model suppliers must ensure sovereignty and a falling cost curve. Investors should watch application-layer cash flow, not just parameter counts.
This column compiles industry information and shares technical perspectives; it does not constitute investment advice.