Nvidia plans to generate $26B annually from RTX Pro 5500 shipments to China

Nvidia appears to have found a very expensive loophole. The company is planning to ship roughly 500,000 units per quarter of its new RTX PRO 5500 Blackwell workstation GPU into China starting late December 2026, a move that could generate approximately $6.5 billion in quarterly revenue, or about $26 billion annually.

The key word in that sentence is “workstation.” By classifying the RTX PRO 5500 as a professional workstation GPU rather than a data center chip, Nvidia may be able to thread the needle on US export restrictions that have choked off its most lucrative China sales for years.

The numbers behind Nvidia’s China play

China’s Ministry of Industry and Information Technology is reportedly assessing purchase plans from major tech firms before potentially greenlighting the deals. ByteDance, the parent company of TikTok, is said to be considering an order of approximately 1 million units alone. Alibaba is also among the companies whose purchase proposals are under review.

The RTX PRO 5500 is not a modest piece of hardware. It packs 84 GB of GDDR7 ECC memory and 21,760 CUDA cores, with a thermal design power of up to 600 watts.

Pricing in China lands between 85,000 and 90,000 yuan per unit, roughly $13,000. Multiply that by 500,000 units per quarter and the math gets interesting fast. At $6.5 billion per quarter, China alone would represent one of Nvidia’s largest revenue streams globally.

The GPU is optimized for AI inference workloads, meaning it’s built to run trained AI models rather than train them from scratch.

Why the workstation label matters

US export controls on advanced semiconductors to China have been tightening since late 2022. Those restrictions have primarily targeted data center GPUs, the high-performance chips used to train and run large-scale AI models. Nvidia’s flagship data center products have been progressively locked out of the Chinese market.

The RTX PRO 5500’s classification as a workstation GPU could place it outside the scope of current export control frameworks. Nvidia previously designed China-specific chips like the H20 to comply with export thresholds, only to see those thresholds tightened further.

What this means for Nvidia and the broader market

For ByteDance, acquiring a million inference-optimized GPUs signals serious ambitions in deploying AI products at scale. Alibaba’s interest follows a similar logic, as the company continues expanding its cloud and AI service offerings across Asia.

Late December 2026 shipments mean revenue wouldn’t start flowing until Nvidia’s fiscal Q1 2028 at the earliest.