Beyond Nvidia: how US export curbs are forcing China to redesign its AI chip industry

The fight for China’s AI chip future is no longer about cloning Nvidia. Under the weight of sustained US export controls, domestic chipmakers are being forced to choose between two very different paths: the versatile, general-purpose GPU or the hyper-specialised application-specific integrated circuit (ASIC). That choice is reshaping the entire hardware landscape. For years, the GPU was the undisputed king. Nvidia recognised early that a chip designed to render video game graphics could handle the parallel calculations needed for complex AI networks. That insight made its products indispensable. But with the most advanced Nvidia chips now off-limits to China, the domestic industry is scrambling to fill the void. The irony is that some Chinese chips are already outperforming what Nvidia is allowed to sell here. Morgan Stanley data shows that Huawei’s Ascend 950 and Cambricon’s Siyuan 690 beat Nvidia’s H20 by 50 to 150 per cent. That performance edge, however, masks a deeper challenge. The fight is no longer about finding a single Nvidia clone; it is about building a domestic ecosystem of chips that can reliably support top Chinese AI models from the likes of DeepSeek and Alibaba. For China’s highly commercialised market, which focuses on deploying AI apps to millions of users, this high speed reflects that domestic hardware and software are working hand in hand. But the ecosystem is still fragmented. Leading the domestic GPU charge is Moore Threads, founded in 2020 by Nvidia’s former China executive Zhang Jianzhong. The company is betting that China’s developers will stick with the GPU paradigm they know. That is a reasonable wager, given the vast installed base of software written for Nvidia’s CUDA platform. But the US export curbs have created an opening for a radically different approach. Instead of being a general-purpose computer chip, an ASIC is custom-designed for a single, narrow task. By maximising hardware efficiency just for AI, these chips offer domestic companies a targeted way to bypass US hardware restrictions. Three different streams are gaining traction in the domestic market: neural processing units (NPUs) built specifically to mimic brain-like AI networks, along with other specialised architectures. These chips are far less flexible than GPUs, but for the specific workloads that power China’s booming AI app market, they can be dramatically faster and more energy-efficient. The catch is that ASICs are expensive to design and risky to deploy. A GPU can pivot to new AI models; an ASIC is locked into its original purpose. That trade-off is forcing China’s chip designers to make a strategic bet. Those who back GPUs are betting on flexibility and ecosystem compatibility. Those who back ASICs are betting on raw performance and the ability to leapfrog US restrictions entirely. What a casual observer might miss is that this bifurcation is not a weakness but a deliberate hedge. China’s semiconductor industry is effectively running two parallel experiments. One path preserves compatibility with global software standards; the other pursues independent innovation at the cost of interoperability. The winner will determine not just which chips power China’s AI future, but whether that future remains tethered to Nvidia’s legacy or strikes out on its own.

Under the weight of sustained US export controls on advanced semiconductors, China’s AI chipmakers are battling to forge a self-reliant silicon ecosystem capable of breaking Nvidia’s stranglehold on the market.

US export curbs force China’s chip designers to choose between versatile GPUs and specialised ASICs, reshaping the domestic AI hardware landscape.

The development adds to a wider China semiconductors story in which companies are being judged on execution, capital access, regulatory fit and the credibility of their regional expansion plans.

For business readers, the important question is whether this becomes an isolated announcement or part of a more durable operating pattern across customers, financing channels, partners and public-market expectations.