“Small yard, high fence” was how Jake Sullivan, National Security Adviser under President Joe Biden, described U.S. tech policy towards China. The approach, designed to fence off sensitive U.S. technologies from China, was most commonly associated with export controls that, among others, restricted Chinese firms’ access to Nvidia’s advanced chips, the critical inputs for developing cutting-edge artificial intelligence (AI) systems. China is now fighting back by taking a leaf out of that playbook. For years, it was the U.S. erecting barriers and driving decoupling between the world’s two tech superpowers. But two developments at the start of 2026 suggest that the Chinese government is no longer just the target of such measures, but also a driver of decoupling itself: its apprehension towards Nvidia’s high-end H200 chips being allowed into the country, and its probe into Facebook-owner Meta Platforms’ acquisition of Chinese-founded AI start-up Manus.
Last week, Beijing told domestic tech companies not to purchase Nvidia’s chips unless absolutely necessary, with new rules set to cap the total number of advanced AI chips that local firms can import. A Chinese data center employee told the South China Morning Post that they have been preparing to build server stacks in Inner Mongolia based on H200 chips to serve major clients such as Alibaba and ByteDance. However, the government had instructed the firm to reserve stack space for domestic chips, the person said. In this new phase, the Chinese government has shifted to proactively restricting imports to accelerate domestic substitution and ensure self-reliance across the entire AI stack.
The key question for industry observers is whether China’s tech industry can stand on its own two feet without significant U.S. involvement. For many years, the answer to that was a no. From the late 1990s to the early 2010s, U.S. capital and technical expertise were vital for building up China’s tech capabilities, according to Chen Li, Assistant Researcher at Beijing-based think tank Anbound. During that period, tech firms from Apple to Microsoft localized their products, opened research and development (R&D) centers, and partnered with Chinese internet firms. Major Chinese companies, from Alibaba to Baidu, were able to rapidly scale up operations after blockbuster listings on U.S. exchanges.
The first Trump Administration marked an inflection point with the first round of U.S. sanctions targeting ZTE and Huawei Technologies starting in 2016. For the first time, Washington showed that it would leverage its dominance of critical components, from chips to chipmaking equipment itself, in its tech war with China. Beijing responded by kick-starting nationwide efforts to pursue self-reliance in semiconductors and AI, spearheaded by Huawei. When the Biden administration added export controls on advanced semiconductors and the tools that make these chips in a significant escalation on October 7, 2022, the fences some had hoped would be temporary instead became firm fixtures. “After that, Chinese firms had to assume access to U.S. technology was conditional and political, while U.S. firms had to price geopolitical risk into their China strategies,” said Ahmad Shadid, Founder of Swiss-based AI research lab O Foundation.
The recent loosening of U.S. chip export controls, first on Nvidia’s H20 and then H200, marked the end of extensive multi-year lobbying efforts by the leading U.S. chip designer and its charismatic founder Jensen Huang. But Beijing’s reported ambivalence over these chips now being allowed into the Chinese market suggested a different dynamic. The H200, Nvidia’s second most powerful AI chip, would be the most powerful AI processor available to Chinese firms, should they be given a license to purchase it. However, after years of navigating U.S. export controls and sanctions, Chinese policymakers seem to be taking the view that escaping this U.S. chokehold long term is more important than technological progress in the short term, the South China Morning Post reports.
