
The contest between the United States and China over artificial intelligence has moved beyond commercial competition into a struggle over strategic capacity, digital infrastructure and geopolitical influence. Artificial intelligence is now treated as a resource linked to military strength, cyber capability, economic competitiveness and state power.
At the centre of this rivalry are advanced semiconductors, particularly the high-performance graphics processing units needed to train and operate powerful AI systems. Nvidia’s chips have become strategically significant because they support large language models, military-grade computing and wider AI infrastructure. Its position has also made the company central to Washington’s effort to limit China’s access to advanced computing capacity.
The United States has tightened export controls since October 2022, with further restrictions introduced in October 2023, December 2024 and March 2025. These measures were designed to restrict China’s ability to obtain advanced chips, maintain supercomputers and manufacture sophisticated semiconductors. Washington has also pushed countries including the Netherlands and Japan to restrict semiconductor exports, turning technology controls into instruments of economic statecraft.
The strategy has had mixed effects. Export controls have complicated China’s access to advanced AI hardware, but they have also accelerated Beijing’s drive for technological self-sufficiency. Chinese firms including Huawei, Baidu and Alibaba have expanded domestic AI chip production, while Chinese semiconductor companies reportedly shipped about 1.65 million AI accelerators in 2025. Huawei has become a focal point of that response, promoting its Ascend processors as replacements for restricted Nvidia products.
The unresolved dimension is whether restriction strengthens control or hastens fragmentation. As AI supercomputing demands expand, with performance doubling roughly every nine months and energy requirements rising sharply, the US-China rivalry is reshaping not only technology supply chains, but the architecture of globalisation itself.