Oct 20 – ByteDance, Alibaba and Tencent have placed orders for more than 3 million Nvidia RTX PRO 5500 graphics cards, with a combined listed value exceeding $35 billion, as Chinese technology companies scramble to secure artificial intelligence computing power amid continuing US restrictions on advanced semiconductor exports.
The orders reflect growing demand for Nvidia hardware that remains eligible for sale in China, even as Washington limits access to some of the American chipmaker’s most powerful processors. The RTX PRO 5500 is less powerful than Nvidia’s leading data-centre chips, but its availability makes it an attractive alternative for Chinese companies developing and operating increasingly sophisticated AI systems.
Three people familiar with the situation said Nvidia had warned customers and business partners that it could not guarantee delivery of the requested quantities. Demand from Chinese technology companies has significantly exceeded the production capacity planned for the product, raising questions about how much hardware the companies will ultimately receive.
The scale of the purchases highlights the difficulties facing China’s technology industry as it competes in the global AI race. Companies are investing heavily in computing infrastructure to develop new models, improve existing services and introduce AI tools across businesses, while restrictions on foreign technology have narrowed their options.
Export Restrictions Drive Demand for Nvidia’s Available Chips
The RTX PRO 5500 is designed primarily for professional workstations rather than the large data-centre systems typically used to train and operate advanced AI models. Nevertheless, it is among the most powerful Nvidia products that Chinese buyers can currently access without encountering the restrictions applied to the company’s leading processors.
The three technology companies have ordered cards priced at approximately $11,400 each, according to two people familiar with the purchases. At that price, orders exceeding 3 million units would carry a combined listed value of more than $35 billion.
However, the requested quantities are far greater than the supply Nvidia is expected to make available. One person said the company had indicated that customers might receive only 5% to 10% of the amounts they requested. Another said Nvidia had suggested it could manufacture no more than 3 million RTX PRO 5500 cards worldwide in 2027. Orders from Chinese customers alone have exceeded 6 million units, two people said.
Those figures suggest that even a substantial increase in production would leave many buyers competing for limited supplies. They also indicate that the announced order volumes should not be interpreted as confirmed deliveries or completed purchases.
Nvidia has not guaranteed that it can fulfil the orders in full, leaving the three Chinese companies facing uncertainty over the timing and scale of their planned hardware acquisitions.
The restrictions are particularly significant because Nvidia’s most advanced products remain difficult or impossible to obtain in China. The company’s Blackwell-based B300 data-centre processor is barred from sale to Chinese customers under current US export controls. Exports of its Hopper-based H200 chip are subject to case-by-case review and licensing requirements.
These limitations have encouraged Chinese companies to consider less powerful alternatives that can still support parts of their AI operations. The RTX PRO 5500 is expected to be used mainly for inference, the process through which an already-trained AI model generates responses, produces content or performs other tasks in response to user requests.
Inference is generally less computationally demanding than training a large AI model on enormous datasets, although the resources required depend on the model, workload and scale of deployment. This makes workstation graphics cards a potential option for companies seeking to expand AI services without relying exclusively on restricted data-centre processors.
The new product also appears to benefit from changing market expectations. Nvidia introduced the RTX 6000D last year as a lower-capability product intended to comply with US export restrictions. Initially, major Chinese technology companies showed limited interest in the card, viewing its price as excessive relative to its performance and anticipating that Washington might eventually permit sales of more powerful processors.
Those expectations have weakened as restrictions have persisted and computing capacity has remained difficult to secure. The RTX PRO 5500, which has performance characteristics similar to the RTX 6000D according to product specifications reviewed by people familiar with the matter, has consequently become a more attractive option.
The shift illustrates how export controls can influence purchasing decisions beyond the products directly prohibited. When access to more powerful hardware becomes uncertain, companies may be willing to pay higher prices for less capable equipment that can meet at least some of their operational needs.
An Nvidia spokesperson said Chinese manufacturers of workstation chips and graphics cards had experienced unprecedented growth since 2022, when the United States first introduced the relevant restrictions.
The spokesperson also argued that American companies faced a combination of US export controls covering older gaming products and China’s own restrictions on imports of US technology. The comments underscored the competing trade and technology policies affecting the semiconductor industry.
ByteDance, Alibaba and Tencent did not respond to requests for comment. The US Commerce Department’s Bureau of Industry and Security, which oversees American export controls, and China’s Ministry of Industry and Information Technology also did not respond to requests for comment.
Nvidia Faces Production Constraints as China’s AI Demand Grows
The surge in orders is creating a supply challenge for Nvidia at a time when the company is preparing to prioritise its next generation of computing infrastructure.
Two people familiar with the situation said Nvidia was expected to devote a substantial portion of its manufacturing capacity at Taiwan Semiconductor Manufacturing Co to its Rubin platform from 2027. Rubin is designed for large-scale data-centre computing, and the production shift could leave less capacity available for products such as the RTX PRO 5500.
Taiwan Semiconductor Manufacturing Co, commonly known as TSMC, declined to comment on the reported production plans.
Nvidia has told customers to expect early samples of the RTX PRO 5500 in December, with commercial deliveries scheduled to begin in the first quarter of 2027, the people said. The timing means Chinese companies could face an extended wait before receiving significant quantities of the cards, even if their orders are accepted.
Supply constraints are already affecting the market for comparable hardware. One person familiar with the situation said Nvidia had discontinued the RTX 6000D and that its market price in China had doubled since then. The increase reflects the pressure that limited availability can place on prices when buyers have few comparable alternatives.
The demand also comes as Chinese technology companies attempt to reduce their dependence on foreign semiconductor suppliers. Washington’s restrictions have encouraged domestic investment, while Beijing has promoted the use of Chinese-made chips in an effort to strengthen the country’s technological independence.
Huawei has gained market share in China’s semiconductor and AI-computing market. However, domestic manufacturers have generally struggled to match the capabilities of leading American products, particularly in the most advanced segments. Restrictions on access to foreign semiconductor manufacturing equipment have added another obstacle by limiting China’s ability to produce cutting-edge chips at scale.
The resulting gap between demand and domestic supply has left companies balancing two priorities: developing homegrown alternatives over the longer term and obtaining enough computing power to maintain their current AI development programmes.
The RTX PRO 5500 orders demonstrate that restrictions have not eliminated Chinese demand for Nvidia technology. Instead, they have helped redirect some purchases toward products that fall within the remaining boundaries of US export policy.
A separate disclosure by Shenzhen-listed Tansun Technology offers another indication of the competition for available hardware. In a filing dated October 13, the fintech and AI-computing services company said it had agreed to purchase 20,000 RTX PRO 5500 cards from an unidentified counterparty.
Tansun cautioned that shortages, strong industry demand and overseas technology controls could delay deliveries or reduce the number of cards ultimately supplied. The warning reflects the broader uncertainty facing Chinese buyers as demand continues to outpace the available supply of suitable processors.
For Nvidia, the orders represent a substantial potential market, even as restrictions have sharply reduced its access to China’s most advanced AI data-centre segment. Chief Executive Jensen Huang has said the company’s share of China’s AI data-centre chip market fell from near-total dominance to effectively nothing following the export curbs.
The interest in workstation graphics cards suggests that Nvidia may still be able to serve parts of the Chinese market through products with lower computing capabilities. However, the company’s ability to benefit will depend on export rules, manufacturing capacity and the extent to which buyers can secure the hardware they need.
For China’s AI industry, the immediate challenge is ensuring that computing resources keep pace with the rapid development of AI models and services. Domestic chipmakers may gradually close the performance gap, but the latest orders indicate that Chinese technology companies continue to see value in Nvidia products that remain available.
With requested quantities exceeding the potential global production of the RTX PRO 5500, the competition for supply is likely to remain intense. The outcome will help determine how quickly Chinese companies can expand their AI infrastructure while navigating restrictions that continue to reshape the global semiconductor market.