Report: Chinese AI firms buy over a million Nvidia GPUs to fill compute gap
Financial Times sources say Chinese technology companies bought more than 1 million Nvidia gaming and workstation GPUs this year to address AI compute shortages. Domestic AI chip production is projected at 3.5 million units, leaving a gap of nearly 1.5 million against demand above 5 million. The report says the export-compliant chips, including RTX Pro 6000D, RTX Pro 5000, and RTX 5090, are being used for simpler AI tasks.
According to Financial Times reporting cited by industry sources, Chinese technology firms purchased over one million Nvidia gaming and workstation GPUs during the year. The purchases were intended to ease shortages in AI computing capacity. Domestic AI accelerator output was projected at 3.5 million units, while demand exceeded five million, leaving an estimated shortfall of nearly 1.5 million chips.
The named models include the RTX Pro 6000D, RTX Pro 5000, and RTX 5090. These Blackwell-based cards reportedly handle less demanding AI workloads, such as prefilling, and cannot match Huawei's Ascend 910C. Nvidia's CUDA ecosystem is also described as more favorable than Huawei's Cann alternative.
The reliance on Nvidia hardware may affect Chinese AI developers by shaping access to compute and costs. Consumers, especially gamers, could face higher GPU prices and reduced availability if demand diverts cards from retail. Nvidia and domestic chipmakers may see shifting incentives, while broader AI progress could depend on how quickly alternative ecosystems mature. The outcome may influence global supply chains and the pace of AI services, though effects remain uncertain.