The US-China technology contest keeps intensifying. Against that backdrop, China appears ready to shift from passive defence to building walls of its own. According to people familiar with the talks cited by the Financial Times, and echoed by Reuters, Beijing has moved into active consultation. Regulators led by the Ministry of Commerce have held intensive discussions with Alibaba, ByteDance, Huawei, and Zhipu AI (Z.ai). Together they are weighing far stricter export and transfer rules for the most advanced AI models and chip designs.
Observers read the move as a riposte to years of American chip restrictions. Yet it signals something deeper. Beijing now treats frontier AI models and infrastructure as strategic national assets deserving close protection.
Three Pillars of the Proposed Controls
The consultations reportedly span three broad fronts. Officials are weighing whether to fold several new measures into China’s export-control regime for restricted technologies.
1. Curbing the Outflow of Model Weights and Training Data
China has lately shone in the open-source arena. Models such as Kimi K3, DeepSeek, and GLM-5.2 have won wide affection among developers worldwide. New rules, however, could stop foreign users from downloading those fiercely competitive model weights directly. In addition, regulators would tightly police the transfer of core training data abroad.
Overseas users would not lose access entirely. Cloud APIs and model-as-a-service offerings would remain open. In effect, Beijing would restrict possession while still permitting use.
2. Blocking Foreign Foundries From Building Chinese Designs
The hardware front looks equally pointed. The Ministry of Commerce is exploring whether to bar overseas manufacturers such as Qualcomm and TSMC from fabricating chips based on Chinese designs. Huawei, Alibaba, and ByteDance developed many of those blueprints. Consequently, China wants more than hardware self-sufficiency. It also aims to keep its homegrown silicon IP and custom architectures out of Western hands.
3. Tightening Scrutiny of Foreign Acquisitions
Strategically valuable AI fields face a third layer of protection. Agentic AI, in particular, sits at the centre of attention. Beijing plans to raise the bar for mergers and technology transfers in this space.
One episode largely lit the fuse. Meta paid roughly two billion dollars for Manus, an agentic AI startup founded by Chinese entrepreneurs and later headquartered in Singapore. China’s National Development and Reform Commission subsequently ordered the deal unwound. That intervention marked the first time Beijing forced a completed transaction to be dismantled under its foreign investment security review rules.
From Fast Follower to Fierce Guardian
In my view, the most telling aspect is psychological. This proposed wall reflects rising technological confidence and a genuine change of posture.
For years, China’s AI ascent leaned heavily on open-source and open-weight strategies. Global developers downloaded and fine-tuned Chinese models freely. In return, Chinese labs harvested worldwide feedback quickly. That loop helped them narrow the gap with OpenAI, Anthropic, and other American leaders.
The calculus has since shifted. Chinese models now rival top American systems on frontier benchmarks. Moreover, many European developers treat them as affordable substitutes for costly US services. Giving away such crown jewels no longer squares with protecting national competitiveness.
A Two-Way Wall Around the Global AI Supply Chain
Suppose this framework eventually takes effect. The fallout would reach well beyond Chinese and American firms. Open-source communities and developer ecosystems everywhere would feel the tremors. Startups and academic labs that lean on free Chinese models to cut costs could suddenly find their access narrowed.
Chinese companies face a double-edged sword too. Participants in the consultations have already voiced unease to regulators. Overly harsh limits, they warn, could slow the domestic industry’s own pace of iteration. Worse still, China might forfeit its chance to expand influence abroad and shape the global ecosystem.
Washington works to keep advanced compute out of Chinese hands. Now Beijing prepares to answer by fencing off its finest models and chip designs. Step by step, this rivalry is pushing the world toward a hard partition of the AI ecosystem. Nothing is settled yet, however. How forcefully these rules land, and by what standards, should become the policy question that global technology and semiconductor watchers track most closely through the rest of the year.
Support Our Threat Intelligence
If you find our technology report and cybersecurity news helpful, consider supporting our work.
Buy Me a Coffee
PayPal
