- Chip curbs cut near-term available compute across both leased and self-built capacity, likely delaying the next Hunyuan model
- Tencent expects a long-term shift to domestic chips, accelerating China's supply chain and R&D concentration
- AI infrastructure capex is reusable, and with falling equipment prices and efficiency gains, pressure should ease by 2028
AI CURATED HOT STORY · 11/18/2026
Tencent Confirms Chip Shortage Delays Next-Gen Hunyuan Model Training
Tencent has explicitly confirmed during its Q3 2026 earnings call that chip supply restrictions will extend the training period for its next-generation Hunyuan large model. This marks a shift from industry-wide concern to a tangible constraint affecting the model iteration pace of a leading Chinese AI player. Despite the short-term compute crunch, Tencent reported quarterly revenue of 207.3 billion yuan, up 8% year-over-year, with cloud services growth accelerating to 35% and AI-related businesses emerging as key growth drivers.
Multiple sources indicate that the chip restrictions have reduced available compute for Tencent, impacting both leased and self-built clusters, and potentially delaying the next Hunyuan model's launch. However, management offered a long-term optimistic view: the constraints will accelerate the shift toward domestic chips and supply chains, concentrating R&D investment. AI infrastructure capex is seen as reusable, and with falling equipment prices and efficiency gains, operating pressure is expected to ease by 2028. Tencent also disclosed that WorkBuddy surpassed 10 million daily active users, making it the most token-intensive AI product in China, underscoring robust application demand.
The development highlights compute supply as a strategic resource shaping the pace of the large-model race. Tencent's candid disclosure reveals short-term challenges while pointing to a long-term path of domestic substitution. For the industry, the capex pivot and supply chain restructuring by a top player will profoundly impact China's AI ecosystem, channeling more resources toward domestic chips and infrastructure.
Hourly heat curve
22 fully observed hoursThe line compares only sources observed throughout; gap hours are interpolated to keep the trend continuous.
Coverage timeline
Total 2 reports · Latest first- IT之家 · 智能时代T2·58 pts
- IT之家 · IT业界T2·62 pts
- Tencent confirms chip restrictions will lengthen next-gen Hunyuan training, showing compute constraints now affect frontier model cadence
- Q3 revenue reached RMB 207.3B, up 8% YoY, with cloud growth accelerating to 35% as AI becomes a growth driver
- WorkBuddy passed 10 million DAU, becoming China's highest-token-volume LLM-powered AI product