Publication date: 13 October 2026
Nvidia remains the dominant name in global artificial-intelligence computing, but its position in China is being reshaped by policy as much as technology. New IDC data reported by Reuters show Nvidia’s share of China’s data-centre AI-accelerator market fell to 34% in the first half of 2026.
The company shipped 678,000 accelerators in China during the period, 44% fewer than a year earlier. Huawei shipped 572,000, up from more than 300,000, giving it a 28% share. Nine of the 11 Chinese chipmakers tracked by IDC increased shipments.
Restrictions are changing the competitive field
US controls limit which advanced Nvidia products can be sold to China and require licences for certain exports. Washington says the rules are intended to restrict access to computing capabilities that could support military and security applications. A revised policy announced in January allows some products, including Nvidia’s H200, to be reviewed case by case under security and compliance conditions.
The commercial effect is visible. Nvidia has repeatedly excluded China data-centre compute revenue from its forward outlook because of geopolitical uncertainty. Its latest public filings indicate that shipments of Hopper data-centre products to China represented less than 1% of data-centre revenue in the quarter ended July 2026.
Export controls have not removed Chinese demand for AI computing. They have redirected it toward domestic suppliers and a wider mix of processors.
The market is moving beyond GPUs
Total Chinese AI-accelerator shipments reached just over two million units in the first half, according to IDC figures cited by Reuters. Shipments of conventional graphics processors fell 26%, while other accelerators designed for specialised AI tasks rose nearly 80% and overtook GPUs by volume.
Huawei, Alibaba’s T-Head, Baidu’s Kunlunxin and Cambricon accounted for 95% of this non-GPU category. That shift matters because competition is no longer determined only by the speed of one chip. Customers are evaluating software, networking, memory, power use and the ability to combine processors into reliable clusters.
A strategic gain with a technological cost
For Beijing, greater domestic market share reduces dependence on a US supplier and strengthens an indigenous technology ecosystem. For Washington, the figures suggest controls can reduce Nvidia’s direct role in China.
But market withdrawal has a second-order effect. Protected demand gives Chinese developers revenue, customer feedback and an incentive to improve complete systems. A policy designed to slow access to the most advanced chips may therefore accelerate investment in alternatives, even if those alternatives initially offer lower performance.
The new measure of competition
Nvidia’s global advantage still includes mature software, developer familiarity and high-performance systems. China’s domestic suppliers do not need to reproduce that advantage immediately across every workload. They need products that are available, compliant and good enough for priority uses.
The 34% share is therefore more than a sales statistic. It shows how export policy can reorganise an industry, changing which companies learn fastest and what customers value. The next stage of the AI-chip contest will be fought at the level of systems and ecosystems, not only transistor performance.