TL;DR: China on Oct. 16, 2026 proposed export bans on 14 technologies and license requirements for 26 more spanning AI, chips, solar and biotech, with public comments open until Nov. 15, 2026.

China has put forward its broadest technology-transfer plan in years, aiming future sales of advanced know-how at the intersection of artificial intelligence, semiconductors, solar manufacturing and biomedicine.

According to the Associated Press, the country's Ministry of Commerce and Ministry of Science and Technology jointly published a draft revision to the national catalogue of technologies subject to export bans or restrictions on Friday, Oct. 16, 2026. A ministry spokesperson described the step as necessary and pointed to national security among the motivations.

The draft does not take final effect immediately. According to analysis published by the Global Sanctions and Export Controls Blog of Baker McKenzie, the proposal was issued as Announcement [2026] No. 48 and remains open for public comment until Nov. 15, 2026. The same analysis reports that exporters face a temporary registration duty with the commerce ministry starting Oct. 16, 2026 for any listed item, even though registration alone is not an approval.

What the 40-item list would cover

The headline number is consistent across reporting: 40 technologies would be added.

According to Caixin Global, 14 of those would face outright prohibition, meaning no export licence could be granted, while the remaining 26 would require licences. The South China Morning Post reports the same split and says the three sectors involved are artificial intelligence, biotechnology and photovoltaic solar power.

That structure matters for U.S. buyers, developers and manufacturers because a prohibition and a restriction lead to very different outcomes. A prohibition would block transfer entirely, while a restricted listing would allow shipments only after review and permission. The Baker McKenzie analysis adds that Beijing is also discussing a new concept called automatic licensing, where certain movements of restricted technology could proceed essentially upon filing, although further rules would be needed to define how it works.

How AI models, agents and compute would be treated

Artificial intelligence dominates the draft, accounting for 23 entries, according to the Associated Press.

The most sensitive category would be barred from export. That includes core technology for large models developed with training data classified as core data under Chinese data-security rules, a grouping broadly linked to critical national-security interests and in practice generally accessible only to government or authorized bodies, according to the Baker McKenzie analysis. Also slated for prohibition are agents intended solely or mainly for highly sensitive uses such as national critical infrastructure, national defence or core government systems, as well as AI with strong cyber-offensive ability and high-performance detection and identification functions.

A second tier would need licences. That group includes large-model technology built with training data classified as important data, a less sensitive tier than core data, or that meets technical thresholds, according to the same analysis. The thresholds cited are training compute reaching 10-to-the-26th FLOPs, models with at least 2 trillion parameters, or models with globally advanced capabilities such as ranking among the top 10 in public benchmarks.

The Associated Press reports that exports of models at or above 2 trillion parameters, or with certain world-leading capabilities, could be restricted, noting that a parameter count reflects learning capacity. Chinese systems including Moonshot's Kimi K3 and Alibaba's Qwen 3.8-Max are among models described as exceeding 2 trillion parameters, though the draft itself did not name specific products.

Other restricted AI items would include high-performance agent technology not otherwise prohibited, design technology for AI chips and advanced memory chips, large-scale computing-cluster technology, AI-enabled cybersecurity tools, and information-processing functions such as speech and voiceprint recognition and Chinese-language translation, according to the Baker McKenzie analysis. Computing clusters with 10,000 or more AI accelerator cards were highlighted as a particular focus.

Chips remain central to the controls

Several chip-design technologies are included, notably those for some AI processing chips and memory chips, according to the Associated Press.

That focus reflects an ongoing contest over computing hardware. The Associated Press notes that U.S.-led limits in recent years have curbed shipments of certain advanced technologies to China, leaving the country blocked from the most cutting-edge AI chips and the advanced chipmaking machines that produce them.

The Chinese draft approaches the issue from the opposite direction by seeking oversight of outbound design knowledge rather than inbound equipment. The Baker McKenzie analysis says overseas chip-fabrication activity could qualify for facilitation, suggesting Beijing is trying to distinguish between sensitive design transfer and routine offshore production, although details remain to be clarified.

Solar and biotech add supply-chain stakes

Twelve solar-related technologies are included, according to the Associated Press, alongside five biomedicine technologies involving areas such as cell and gene therapy.

For solar, the draft targets upstream and next-generation manufacturing know-how. That includes technology tied to polysilicon, a core raw material for the solar supply chain, according to the Associated Press. Reporting by The Hindu adds specificity: perovskite and tandem solar-cell module manufacturing technologies were proposed for complete prohibition, while polysilicon engineering technologies involving production lines with annual output of at least 30,000 tonnes were listed for restricted export.

The commerce ministry statement summarized by The Hindu said only a limited set of photovoltaic technologies would be regulated and other photovoltaic technologies would remain freely exportable. The ministry also linked part of the solar rationale to expanding use in aerospace, defence, unmanned systems and other sensitive sectors, which it said had increased the national-security implications of exports.

Biopharma coverage is narrower but potentially significant for deal activity. The proposal lists certain cell and gene-therapy technology platforms for bans or controls, according to the South China Morning Post and The Hindu. A Shenzhen trade-compliance lawyer quoted by The National said companies involved in cross-border acquisitions, intellectual-property restructuring or international licensing should check whether transactions trigger the new registration duty, with particular relevance for biopharma after heavy merger activity in the past two years.

Exemptions for open source, papers and cloud access

Despite its breadth, the draft contains important carve-outs for routine research and conventional product delivery.

According to the South China Morning Post, restrictions will not apply to technologies already in the public domain, such as open-source systems or material published in academic papers. The Hindu similarly reports that open-source AI projects and technologies already disclosed through open publication or papers would generally be exempt.

The Baker McKenzie analysis describes an additional practical distinction for AI services. Overseas localized deployment of restricted AI technology would count as an export, but providing overseas users with access to models hosted in China through application programming interfaces, application or web interaction, software-as-a-service applications or lightweight on-device deployment would not be deemed a technology export and therefore would fall outside technology-export controls.

The same analysis says open-source release and overseas deployment where the Chinese party retains control, such as deployment in its own overseas cloud environment, could qualify for automatic licensing once that mechanism is defined. For U.S. consumers and developers, that suggests everyday use of a China-hosted chatbot or app would be treated differently from transfer of the underlying model weights or training methods.

Why Beijing says it is acting now

Chinese officials framed the move as preventive rather than punitive.

A commerce ministry spokesperson said in a statement that AI, biomedicine and photovoltaic technologies had seen expanding applications in sensitive fields, according to The Hindu. The statement said major economies had adopted varying degrees of export controls in these areas and that China was drawing on international practice to prevent risks from loss of control or inappropriate proliferation while safeguarding normal cooperation.

On AI specifically, the ministry said most AI technologies would continue to be freely exportable, according to The Hindu. It warned that tools such as autonomous malicious-code generation and mutation could be used against critical infrastructure, defence systems and core government networks or to evade defences, and said the objective was to establish safeguards, prevent misuse and keep AI under human control.

The Associated Press quoted the ministry as saying the government was implementing necessary measures with the aims of reinforcing the security bottom line, preventing misuse and abuse, and ensuring that AI remains under human control.

The timing overlaps with tentative U.S.-China engagement. The Associated Press reports the proposal came as the two countries made progress in dialogues over AI safety and governance and on establishing a notification mechanism for AI incidents that could threaten national security. The National reports that Washington and Beijing last month agreed to cut tariffs on $30 billion of goods and start dialogue on AI after a White House visit by Chinese President Xi Jinping.

Early reaction points to fragmentation

Independent analysts described the draft as expansive and politically significant.

Lian Jye Su, a chief analyst at technology research and advisory group Omdia, said the move can be seen as a response to Western countries posting sovereignty requirements on AI, robotics, telecommunications and other key technologies, according to the Associated Press, adding that the aim would be to better regulate access to Chinese technologies.

Gary Ng, a senior economist at French bank Natixis, said AI is increasingly strategic for both China and the United States and that the world is getting more fragmented in trade and investment as well as access to technologies, according to the Associated Press.

Leonhard Xu, a senior consultant at Berlin-based consultancy Sinolytics, told The National the announcement is extremely expansive and will take time to assess, while emphasizing that export control does not equal an export ban in most cases.

Wendy Cutler, a senior vice president at the Asia Society Policy Institute, told The National the fragility of the trade truce could not be overstated and warned it may only be a matter of time until the United States responds with its own restrictive measures.

For context on U.S. action, the Associated Press notes that the U.S. Commerce Department in June had asked Anthropic to bar foreign nationals from using some of its most powerful AI models, Fable 5 and Mythos 5, citing national security concerns, before those limits were later lifted.

What happens next is procedural. Comments are due by Nov. 15, 2026, after which authorities are expected to publish a final catalogue, according to The National and the Baker McKenzie analysis. Until then, companies transferring models, chip-design data, solar-process technology or covered biotech platforms connected to China will need to track whether registration, licensing or a future automatic-filing path applies.

Sources referenced