Chinese Regulators Approve DeepSeek and MiniMax for Overseas AI Deployment
nashnova research
DeepSeek and MiniMax have cleared Chinese regulatory approval for overseas AI deployment, even as Beijing drafts new export controls on select AI technologies — but the proposed rules exempt API and SaaS access, leaving most AI tech free to export.
What exactly got approved?
DeepSeek and MiniMax have completed the required registration process and are cleared to deploy AI models overseas, according to a post by CCTV's "Yuyuan Tantian" account on October 22.
Both companies have finalized arrangements for open-source release and overseas licensed use.
This means → Beijing has formally signed off on Chinese AI companies going global — not a policy signal, but a completed regulatory procedure.
What do the draft export controls cover — and what do they leave out?
China opened public consultation last week on revised technology export catalogs. The draft covers 23 information-processing categories, including specific large language models, AI agents, and computing clusters.
Technology already in the public domain via open-source release or academic publication falls outside the proposed controls.
In plain terms = the draft draws a boundary line: it fences off a narrow set of frontier tech while leaving the rest wide open — most AI technology remains freely exportable.
Why are API and SaaS access exempted?
The draft creates explicit exemptions for services built on restricted underlying models: access provided through APIs — application programming interfaces, the channels that let external software call on AI capabilities — web apps, or SaaS (software-as-a-service, subscription-based online access rather than outright purchase) is not subject to export restrictions.
This means → even if the base model is controlled, Chinese AI companies can still sell products to global users through the service layer. The commercialization pathway is deliberately preserved.
This reflects Beijing's logic: lock down "core technology transfer" without blocking "technology monetization."
How does this compare to the U.S. approach?
The policy note contrasts Beijing's framework with Washington's: the U.S. uses targeted export bans and technology restrictions to protect competitive advantage — an approach centered on blocking.
China's draft takes a different tack — designate a narrow set of controlled items while exempting API/SaaS delivery, an approach closer to "regulate without blocking."
Put simply = the U.S. strategy says "you can't use it"; China's strategy says "I keep the core tech, but you can buy the service."
What to watch next?
Drafting began in the first half of 2026, with input from research institutions and industry participants. The rules are currently in public consultation.
The key question: whether the API/SaaS exemption scope survives once the rules are finalized.
This means → the exemption clause is the most commercially valuable part of the draft — if the final version narrows it, the overseas revenue model for Chinese AI companies will be fundamentally reshaped.
市场有风险，内容仅供研究参考，不构成投资建议。