Sources: Chinese tech companies bought 1M+ Nvidia gaming and workstation chips, mostly RTX 5090 and RTX Pro, for basic AI workloads amid an AI chip shortage
Context & Ripple Effects
Chinese companies had already been repurposing Nvidia gaming chips for AI in 2024. In 2026, a reported plan to permit limited H200 purchases for major AI companies underscored that specialised compute remained scarce rather than broadly available.
The reported million-plus purchases extend that workaround to RTX 5090 and RTX Pro hardware for basic workloads. It sits alongside confirmed evidence that Nvidia's China data-centre AI-chip share and shipments declined in the first half of 2026, making non-data-centre products a more important route to serving demand.
First-order effects
- Chinese tech companies can add capacity for basic AI workloads with gaming and workstation chips when dedicated AI accelerators are constrained.
- Nvidia gains demand for RTX 5090 and RTX Pro units from buyers whose needs would otherwise be directed toward data-centre AI chips.
Second-order effects
- Demand from Chinese AI buyers reaches into the gaming and workstation supply chain, tightening the connection between consumer-oriented GPU availability and AI infrastructure demand.
- Limited access to dedicated Nvidia capacity makes workload placement more heterogeneous: basic jobs move to RTX hardware while scarce specialist chips are reserved for workloads that require them.
Third-order effects
- If this purchasing pattern persists, AI compute shortages will be measured across the broader GPU market rather than only by data-centre accelerator supply.
- The episode points to an AI infrastructure market in which product segmentation matters less than usable compute capacity, with supply restrictions determining which hardware buyers can deploy.
The trend: AI demand is spilling from dedicated accelerators into gaming and workstation GPUs as buyers assemble heterogeneous compute fleets around available supply.