ByteDance, Alibaba, Tencent order $35B in Nvidia chips

A $35 billion signal from China’s AI giants

ByteDance, Alibaba and Tencent have ordered more than 3 million Nvidia RTX PRO 5500 graphics cards with a combined sticker value above $35 billion, according to reporting published on October 20, 2026, citing people familiar with the matter [1]. The scale of the orders is striking not only because of the dollar amount, but because the chip involved is not Nvidia’s most advanced data-center processor. It is a workstation-class graphics card that has become important because it remains one of the most powerful Nvidia products still available to Chinese buyers under current U.S. export controls [1].

The reported orders show how China’s largest internet platforms are still organizing their AI road maps around Nvidia supply, even after years of Washington restrictions and Beijing’s push for domestic semiconductor alternatives [2]. ByteDance needs computing capacity for recommendation systems, video products and generative AI services; Alibaba is building cloud and model infrastructure; Tencent has AI demand across gaming, social platforms, cloud services and enterprise tools. The common thread is simple: for large-scale AI deployment, access to reliable acceleration hardware remains a strategic constraint.

The three companies did not respond to Reuters requests for comment, and U.S. and Chinese regulatory agencies cited in the report also did not provide comment [1]. That silence matters. It means the story is best understood as a supply-chain report rather than a confirmed corporate announcement. Still, the details reported across several outlets point to a major scramble for Nvidia chips that are powerful enough to be useful but limited enough to fit within current export-control boundaries [1], [2], [3].

Why the RTX PRO 5500 matters

The RTX PRO 5500 is described as a workstation graphics card designed for powerful professional computers, not as a data-center accelerator built primarily for vast AI clusters [1]. That distinction is important. Nvidia’s top-tier AI platforms are built to be linked together in large numbers for training and serving demanding models. The RTX PRO 5500 is less powerful than Nvidia’s H200 and B300 products, but it is still valuable because it sits in the narrow category of Nvidia hardware that Chinese buyers can seek under current rules [1], [3].

The economics of the order underline the urgency. The cards are reportedly priced around $11,400 each, which puts the three companies’ combined orders above $35 billion [1]. The reported total is a sticker-price figure, not a guarantee of delivered revenue. Nvidia has reportedly warned customers and partners that it cannot guarantee full delivery because demand from Chinese technology companies has surpassed planned production [1]. In other words, the headline number reveals appetite, but not necessarily near-term shipments.

The supply gap appears severe. One source cited by Reuters said partners might receive only 5% to 10% of the quantities they requested, while another said Nvidia had indicated that it could produce no more than 3 million RTX PRO 5500 cards globally in 2027 [1]. Total orders from Chinese customers have reportedly exceeded 6 million units, meaning demand from China alone could be roughly double the global production figure discussed in the report [2]. If those figures hold, allocation rather than pricing may become the central issue.

Export controls are shaping the product mix

The reason a workstation chip has become the focus of such large orders is the structure of U.S. export controls. Nvidia’s Blackwell-based B300 data-center chip is reported to be unavailable for sale to Chinese buyers under U.S. rules, while exports of the Hopper-based H200 are subject to case-by-case review and licensing conditions [1]. That leaves Chinese companies looking for alternatives that are weaker than Nvidia’s top AI products but still more capable or more compatible than many substitutes.

This is the practical effect of technology restrictions: demand does not disappear; it moves to the next available channel. ByteDance, Alibaba and Tencent are not simply buying chips for prestige. They are trying to secure compute for AI products that must run at commercial scale. If the most advanced data-center accelerators are restricted, a large installed base of workstation-class GPUs can still support inference, experimentation, engineering work and some model-development tasks [3].

Reuters reported that Chinese companies are seeking the RTX PRO 5500 primarily for AI inference, meaning the use of trained models to generate outputs such as chatbot responses [1]. Inference generally requires less computing intensity than the original training of very large models, but it can require enormous aggregate capacity when millions of users interact with AI services. For companies with social, shopping, cloud, gaming and content platforms, inference demand can scale quickly.

The report also suggests that expectations have changed among Chinese buyers. Earlier, some major Chinese technology firms reportedly showed limited interest in reduced-capability Nvidia products because they considered them expensive relative to performance and hoped access to more powerful processors might eventually reopen [2]. Now, with restrictions still in place and AI capacity still scarce, the RTX PRO 5500 looks more attractive as a practical substitute [2].

Nvidia’s dominance, with limits

For Nvidia, the orders are a reminder of the company’s extraordinary position in the AI hardware market. Even when export controls block its leading data-center products, Chinese technology companies still appear willing to place massive orders for less powerful Nvidia hardware [2]. That reflects the strength of Nvidia’s software ecosystem, developer familiarity, performance reputation and supply-chain credibility.

At the same time, the report highlights Nvidia’s constraints. A $35 billion order book is valuable only if the company can manufacture, allocate and deliver the products. Nvidia has reportedly told customers to expect early samples in December, with commercial deliveries beginning in the first quarter of 2027 [1]. That schedule suggests the current story is not about immediate mass deployment, but about reserving scarce future capacity before competitors do.

Nvidia’s manufacturing priorities also matter. Reuters reported that from 2027 the company is expected to devote much of its capacity at Taiwan Semiconductor Manufacturing Co. to Rubin, its next-generation data-center platform, rather than to products such as the RTX PRO 5500 [1]. If Nvidia must choose between supplying export-compliant workstation cards and supporting its most advanced global AI platforms, the allocation problem becomes more complex.

The reported shortage also gives Nvidia leverage. When customers know supply may cover only a fraction of requested orders, they have an incentive to order early and order big. That can inflate headline demand, but it also reveals how much companies fear being left without compute. In the AI economy, a shortage of accelerators can delay model launches, limit product features and raise operating costs.

What this means for China’s AI race

The orders show that China’s AI leaders remain ambitious despite semiconductor constraints. ByteDance, Alibaba and Tencent are not retreating from AI; they are adapting procurement strategies to the rules as they exist now. They are pursuing export-compliant Nvidia hardware while also operating in a policy environment that encourages domestic chips [2]. That dual strategy is likely to continue.

Domestic Chinese suppliers, including Huawei, have gained relevance as U.S. restrictions pushed Chinese companies to consider local alternatives [2]. But the continued demand for Nvidia cards indicates that domestic options have not fully replaced Nvidia for many high-performance use cases [3]. Chinese chipmakers face their own constraints, including limits on access to advanced semiconductor manufacturing equipment, according to the Reuters-based reports [2].

The most important implication is that compute has become a geopolitical resource. The RTX PRO 5500 is not the most advanced chip in Nvidia’s lineup, yet it is reportedly drawing orders worth more than $35 billion from just three companies [1]. That tells us that the bottleneck is no longer only about cutting-edge training clusters. It is also about broad, dependable, deployable capacity for AI services.

If Nvidia can deliver only a small share of the requested quantities, Chinese companies may spread workloads across a mix of Nvidia cards, domestic processors and cloud arrangements. They may also prioritize inference workloads that can run efficiently on available hardware. The result could be a more fragmented AI infrastructure landscape in China, with companies optimizing around scarcity rather than ideal architecture.

The bottom line

The reported ByteDance, Alibaba and Tencent orders are best read as a snapshot of the AI hardware market under pressure. Three of China’s biggest technology companies want more than 3 million Nvidia RTX PRO 5500 cards worth over $35 billion, but Nvidia may not be able to meet anything close to total demand [1]. Chinese customer orders have reportedly exceeded 6 million units overall, while Nvidia’s indicated global 2027 production ceiling for the card may be no more than 3 million [2].

That imbalance explains the importance of the story. It is not only about a massive purchase order. It is about how export controls, AI competition and semiconductor manufacturing limits are reshaping the global chip market. Nvidia remains central, Chinese AI companies remain hungry for compute, and the available chips are becoming valuable precisely because the best chips are harder to obtain.

For now, the RTX PRO 5500 has become a symbol of compromise: not the most powerful Nvidia AI processor, but powerful enough to trigger a multibillion-dollar rush. In a market where access to compute can determine who ships AI products faster, that may be more than enough.

Developments

1. ByteDance, Alibaba, Tencent order over $35bn of Nvidia chipsTaipei Times · Oct 20, 2026, 6:00 PM · 8/10
2. Chinese Tech Giants Order $35 Billion Nvidia ChipsTIKR.com · Oct 20, 2026, 5:12 PM · 7/10
3. Chinese Tech Giants Order $35 Billion In Nvidia ChipsTIKR.com · Oct 20, 2026, 5:12 PM · 8/10
4. Nvidia Sees $35B Orders for RTX PRO 5500 Chips From Major Chinese Firms Amid Export CurbsTradingView · Oct 20, 2026, 12:48 PM · 8/10
5. Alibaba, ByteDance, Tencent order $35B in Nvidia RTX PRO 5500 chips amid export curbsTradingView · Oct 20, 2026, 12:48 PM · 9/10
6. ByteDance, Alibaba, Tencent order $35B Nvidia chips, signaling high demandLatest news from Azerbaijan · Oct 20, 2026, 12:15 PM · 8/10
7. ByteDance, Alibaba, Tencent place over $35 billion Nvidia chip ordersThe Economic Times · Oct 20, 2026, 11:05 AM · 9/10
8. ByteDance, Alibaba, Tencent order $35B worth of Nvidia chipsThe Economic Times · Oct 20, 2026, 11:05 AM · 9/10

Sources from the last 72 hours

1. [1]Exclusive-ByteDance, Alibaba, Tencent place over $35 billion worth of Nvidia chip ordersOct 20, 2026, 7:03 AM
2. [2]ByteDance, Alibaba, Tencent place over US$35b worth of Nvidia chip orders — ReutersOct 20, 2026, 11:10 AM
3. [3]Banned From Top AI Chips, China Just Dropped $35 Billion On This Nvidia GPUOct 20, 2026, 12:06 PM

AI-generated article based on recent web research, then preserved as a dated editorial snapshot.