Nvidia allocates 1% of $70B RTX chip order to China clients

3 hours ago 16

Chinese tech giants ordered millions of RTX PRO 5500 cards, but Bloomberg reports they will receive only a sliver of that demand

Chinese tech companies tried to place one of the largest graphics card orders in recent memory. According to Bloomberg, they will get about 1% of it.

Bloomberg reports that Nvidia’s Chinese clients will receive 1% of a $70 billion order for RTX chips. That is a striking gap between what buyers want and what Nvidia appears willing or able to ship.

Six million cards, three million slots

The chip at the center of this is the RTX PRO 5500, a workstation graphics card. On paper it is built for professional visual work. In practice, Chinese buyers want it for AI inference, which is the stage where a trained model actually answers queries rather than learning new ones.

Demand has been intense. Reports from October 20, 2026 said ByteDance, Alibaba, and Tencent together ordered more than 3 million units. Those orders were valued at approximately $35 billion, at a price of around $11,400 per card.

Those three companies are not the whole story. Total orders from Chinese customers have reportedly topped 6 million units. At list prices, that works out to potential revenue of around $70 billion.

The problem is the factory floor. Nvidia estimates global production capacity for the card in 2027 at 3 million units. Chinese orders alone come to more than double that cap, before anyone else in the world gets a turn.

Nvidia has told its partners it cannot guarantee these orders will be filled. Earlier reporting indicated that many Chinese clients could end up with only 5% to 10% of the cards they requested. Bloomberg’s 1% figure points to an even tighter squeeze.

Timing adds another layer. Sample units are expected to be available by December 2026. Production is set to begin in the first quarter of 2027, with commercial shipments expected to start in that same quarter.

Why Chinese buyers landed on this card

The RTX PRO 5500 did not become a hot commodity because it is Nvidia’s most powerful product. It became popular partly because it is one of the products Chinese firms can still buy.

Nvidia’s more advanced accelerators, such as the Blackwell B300, face restrictions under US export controls. That leaves Chinese companies working with whatever falls on the permitted side of the line. For inference workloads, the RTX PRO 5500 has become a preferred option.

The heavy concentration of orders on a single model shows how much Chinese tech giants still lean on Nvidia. Export rules have narrowed their choices. They have not erased the appetite.

There is also a question of priorities inside Nvidia. The company reportedly favors its next-generation platforms when allocating capacity for commercial shipments. That could leave even less room for the RTX PRO 5500, regardless of how many purchase orders pile up.

What this means for Nvidia, its buyers, and the AI hardware market

For ByteDance, Alibaba, and Tencent, the math is uncomfortable. They ordered more than 3 million cards collectively. If Bloomberg’s 1% figure holds, the hardware they actually receive will be a small fraction of their plans for AI inference capacity.

With orders far exceeding the 2027 production cap of 3 million units, existing supply of Nvidia products could command higher prices. Buyers who secure allocations early may find themselves holding an asset worth more than they paid.

US export controls already determine which Nvidia products can reach China. This episode shows that even permitted products face a second filter: production capacity and how Nvidia chooses to divide it.

What to watch next is fairly concrete. December 2026 is when samples are expected. The first quarter of 2027 is when production and commercial shipments are set to start. Those milestones will show whether the allocation lands closer to Bloomberg’s 1% or the earlier 5% to 10% range.

Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.

Read Entire Article