The US administration has introduced a new ban restricting exports of advanced GPUs to China. A senior executive at Tencent, one of China's largest tech companies, said the move is setting back the country's AI sector overall — while also noting that it will likely accelerate China's own chip production.
What happened?
The US has put a new export restriction into effect, further tightening the sale of advanced GPUs to China. Tencent's leadership argues that in the short term, the ban is slowing down how quickly Chinese companies can build out AI infrastructure — but at the same time, it's acting as a catalyst for the domestic chip industry.
Why does it matter?
US chip export restrictions on China are nothing new. In 2019, the 5G and semiconductor embargo against Huawei was launched on similar logic: isolate the company from the global market and hold it back. The result was the opposite of what was intended. When cut off from Google's Android services, Huawei pushed its own operating system, HarmonyOS, to the forefront; when severed from US suppliers, it began scaling up chip production in-house with domestic foundries like SMIC. Today, Huawei's Ascend series AI chips are taking up an increasingly large share of the Chinese market as an alternative to Nvidia.
A similar cycle may be playing out with GPU export bans. In an environment where even Nvidia's deliberately weakened chips built specifically for China (the H20 and similar models) are periodically restricted, Chinese cloud and AI companies — Tencent, Alibaba, ByteDance among them — are being forced to diversify their GPU supply. That's translating directly into demand and state support for domestic GPU designers like Biren, Moore Threads, and Cambricon. What the Tencent executive means by "it's setting everyone back" is that the performance gap still hasn't closed — but at the same time, this pressure is accelerating the need to grow the domestic ecosystem.
Who does this affect, and how?
In the short term, the parties most affected are the Chinese cloud providers and AI labs that depend on the latest GPU clusters to train large models. The performance gap is still significant; Nvidia's top-tier data center chips are generally considered to be one or two generations ahead of most domestic Chinese alternatives.
| Period | Restriction | China's response |
|---|---|---|
| 2019 Huawei embargo | Android and US chip supply cut off | Accelerated investment in HarmonyOS and SMIC |
| 2022-2023 first GPU bans | A100/H100 sales halted | Nvidia-specific "China models" (A800, H20) introduced |
| 2026 new GPU ban | Broader chip export restrictions | Increased demand for domestic GPU designers (Biren, Cambricon) |
In the long run, the real winner will likely be China's own semiconductor supply chain. State-backed funding, SMIC's investment in advanced node production, and domestic GPU companies' efforts to build out a software ecosystem (layers meant to serve as alternatives to CUDA) are all gaining momentum with each new ban.
What's next?
There are three things worth watching: when China's domestic GPUs will reach real production scale, whether the US will tighten restrictions even further, and how willing Nvidia will be to keep producing "weakened" chips in order to avoid losing the Chinese market entirely. China used to represent a significant share of Nvidia's revenue; the market's closing off is also shaping the company's own strategy.
In the end, this is a classic sanctions paradox: a move designed to weaken a rival in the short term ends up forcing that rival to stand on its own two feet in the long run. The Huawei case showed this is possible; whether the same scenario repeats on the GPU front depends on how fast China can scale up its domestic chip production.
Yorumlar (0)
Henüz yorum yok. İlk yorumu sen yaz.
Yorum yapmak için Sinyal'i indir
Yorumlar Sinyal hesabıyla yapılır. Mobil uygulamada giriş yap, yorum bırak.