Key Notes
- ByteDance and Tencent have each received about 10,000 Nvidia H200 chips in recent weeks, the first meaningful movement of the chips since Trump approved their export to approved Chinese buyers, each firm is licensed to buy up to 100,000.
- Chinese regulators are directing companies to deploy them in Hong Kong, to protect China's domestic chip industry, even though neither firm currently has data centers there.
- Hong Kong's entire data-center capacity can't come close to powering the full 200,000-chip allowance for just these two companies.
ByteDance and Tencent have each taken delivery of roughly 10,000 Nvidia H200 processors in recent weeks, the Financial Times reported, marking the first meaningful movement of the chips into China since the Trump administration approved their export to vetted Chinese buyers last December.
Both companies were among 10 Chinese firms the US authorized to purchase the chips in May, with each permitted to buy up to 100,000 units. Reuters had separately reported in January that China agreed to import several hundred thousand H200s. The Financial Times said a few other Chinese technology firms could soon secure similarly sized shipments.
The H200, while no longer Nvidia’s most advanced chip, remains significantly more capable than China’s domestic alternatives for training frontier AI models, even as Huawei leads the country’s homegrown chip market. Nvidia is reportedly sitting on an inventory overhang of roughly 500,000 H200 units built largely for Chinese customers, so the deliveries also help clear that backlog.
The more consequential detail is where Beijing wants the hardware to actually sit. Chinese regulators are directing ByteDance and Tencent to keep most of their allotted chips out of mainland China, instead shipping them to Hong Kong, which operates under a separate customs regime and technically counts as outside the mainland.
The goal is protecting China’s push toward domestic chip self-sufficiency, a priority that has only intensified in response to US export controls. Neither company currently operates data centers in Hong Kong, and deploying the chips there would let mainland teams still access the computing power remotely over cross-border network links.
A Pattern Not an Isolated Move
That workaround runs into a hard physical limit. Hong Kong’s entire installed data-center base totals only about 581 megawatts across 47 facilities, and the city’s relatively poor cooling efficiency pushes actual power draw even higher than that figure suggests. Exercising even a fraction of the companies’ full 200,000-chip combined allowance would require power capacity that a meaningful share of Hong Kong’s entire existing data-center estate cannot supply.
One person familiar with the situation called it “a dilemma,” with companies unable to find anywhere in the territory to actually deploy the hardware. A planned expansion, the Northern Metropolis data center cluster awarded to Range Intelligent Computing in March, is not expected to come online until 2029, leaving a multi-year gap between the chips arriving and having anywhere to run them at scale.
The Risk of Forcing a Choice
The episode captures an unusual reversal in the US-China chip dispute: for once, it is Beijing restricting access to advanced American hardware rather than Washington. The logic is straightforward, since flooding mainland data centers with US chips could undercut the very incentive driving Chinese firms toward domestic alternatives like Huawei’s Ascend line, alternatives that still lag Nvidia in real-world performance, as illustrated by DeepSeek’s reported difficulties training its R2 model on Ascend chips instead of Nvidia hardware.
For now, the roughly 20,000 delivered chips represent a small fraction, about 2.5%, of the more than 400,000 units the companies are licensed to eventually acquire, meaning the practical effect on China’s AI training capacity remains limited. The story is less about a sudden capability boost and more about the awkward mechanics of a compromise that satisfies neither side’s original goal cleanly: Washington gets to say it approved the sales, Beijing gets to say the chips stayed off the mainland, and the companies that actually need the compute are left waiting for somewhere to plug it in.
Disclaimer: AIstify is an independent media brand owned and operated by NuvexMedia LLC, publishing news, research, and insights on artificial intelligence, emerging technologies, automation, and related industries. NuvexMedia LLC invests in and collaborates with companies across the AI, technology, software, and digital innovation sectors. These relationships do not influence AIstify’s editorial coverage, and the publication maintains full editorial independence to provide accurate, timely, and objective information. © 2026 NuvexMedia LLC. All rights reserved. This content is for informational purposes only and should not be considered legal, tax, investment, financial, or other professional advice.
