China’s proposed AI tech export bans may create short-term costs, long-term gains: analysts

China's proposed export bans on artificial intelligence (AI) technologies aim to strengthen the country's strategic grip in the sector, potentially affecting the commercial prospects of domestic model developers in the global AI race, according to analysts. The new measures follow concerns over advanced AI systems' risks, cyberattacks, and AI's growing importance for national security.

The draft plan, jointly issued by China's Ministry of Commerce and Ministry of Science and Technology, would impose export prohibitions and restrictions on 23 AI-related technologies, including models, agents, and chip designs. Foundational AI models, which drive advancements in the sector, would also face several thresholds for restrictions, such as those trained on China's important data, developed with computing power reaching 10^26 FLOPs, or ranking in the top 10 global leaderboard.

Zhang Yi, founder and chief analyst at consultancy iiMedia, stated that the new measures would introduce short-term adaptation costs for Chinese companies, while underscoring AI's growing national security importance for both Beijing and Washington. Beijing's increasing scrutiny of the industry should not prevent Chinese companies from deploying their AI services globally, according to Zhang, as the key lies in keeping core engineering technologies controlled while keeping commercial services open.

MOFCOM spokesperson clarified that some ways Chinese technology companies provide AI services overseas, such as through official APIs and mobile apps, would not be covered by the new export restrictions.

A key uncertainty is how open-weight AI technologies, a global leading area for China, would be treated under the finalised export regime. Several Chinese open-weight AI developers, such as Moonshot AI and Z.ai, are entering revenue-sharing agreements with major US hyperscalers to deploy their open-weight models on their cloud infrastructure.

However, under the current wording of the new measures, some Chinese AI models, which are leading open-weight systems, would meet technical thresholds, potentially causing delays or prevention of partnerships with global cloud providers. The month-long public consultation period for the draft plan will last until November 15.