US Considers Permit System for Global AI Chip Sales Amid Opposition

The United States is currently exploring the implementation of a permit system for the global sales of AI chips, particularly targeting shipments from major companies such as Nvidia and AMD. Under this proposed system, any export of AI chips would necessitate approval from American authorities, allowing the US government substantial oversight in determining the global landscape for AI technology.

This potential regulatory shift has elicited significant concern among Chinese experts, who perceive it as yet another manifestation of the US attempting to assert its dominance over the semiconductor industry through extraterritorial control. Such measures are anticipated to prompt considerable resistance from the US's international allies, who may view this as an infringement on their technological autonomy.

This development arrives amid China's intensified initiatives to bolster its technological self-reliance. Analysts speculate that, as China progresses in its technological capabilities, attempts by the US to stifle its advancements are poised to be counterproductive.

According to a recent report by Bloomberg, draft regulations from US officials would impose limitations on AI chip shipments without US consent. This would effectively enable Washington to dictate not only whether other nations can develop the necessary infrastructure for training and utilizing AI models but also under what circumstances such activities can occur.

Specifically, the new regulations would require shipments of up to 1,000 of Nvidia's GB300 GPUs to undergo a relatively simplified review process, with certain exemptions possible. However, companies looking to acquire larger clusters would face stricter regulations, including pre-clearance for export licenses, disclosing operational models, and potentially allowing US officials access to sites, depending on the specific data center configurations.

The proposed changes signal a shift towards establishing a global licensing framework for AI chips, as described by Zhang Xiaorong from the Beijing-based Cutting-Edge Technology Research Institute. He articulated that the US’s strategy would effectively reclassify advanced semiconductors as scarce resources, drawing parallels to a new form of currency in the tech landscape.

Zhang emphasized that the US's strategy appears to transition from a defensive containment approach to an offensive dominance, aiming to dictate terms regarding AI development globally. He characterized this posture as an embodiment of technological colonialism, one that could elicit strong backlash from US allies including the European Union, who may view it as a threat to their own ambitions in AI technology.

In the short term, Chinese firms may face limitations in accessing high-performance chips necessary for their computational needs. However, Zhang predicts that these pressures will catalyze a long-term transition towards building a domestic ecosystem for AI, which, although challenging, is critical for achieving true self-reliance.

Significantly, a recent collaborative publication by prominent figures in China’s semiconductor field has advocated for a concerted national effort to develop a domestic alternative to ASML during the upcoming 15th Five-Year Plan (2026-2030). This initiative reflects an urgent need to prepare for the sustained challenges posed by tightening US technology restrictions.

The call for action from industry leaders, including those affiliated with SMIC and YMTC, underscores the imperative of overcoming the heightened US export controls that have limited access to essential lithography technology and other critical components. They are advocating for a move from isolated innovations towards a comprehensive national strategy that can ensure supply-chain security.