US tightens AI chip export controls to deter Chinese companies from accessing compute remotely
Original reporting by TechCrunch

U.S. export controls on AI chips and data centers refer to an expanded regulatory framework designed to prevent countries of concern, notably China, from circumventing existing restrictions by remotely accessing advanced artificial intelligence computing infrastructure located outside their borders. Under an interim rule set to roll out, the Commerce Department will now require companies to obtain licenses for exporting or transferring AI chips and servers to data centers anywhere in the world if those facilities could provide compute to businesses in China or other designated nations within the D:5 group, including Russia and Iran. This move specifically targets remote access, aiming to close a loophole that allowed restricted entities to bypass direct hardware bans.
The new controls are a logical extension of existing efforts, explicitly designed to "undermine the ability of companies headquartered in countries of concern from using U.S. advanced Super Intelligence (SI) semiconductors to train advanced SI models." This expansion follows recent reports of Chinese AI firms seeking overseas computing capacity amid tightening restrictions and comes shortly after a scandal involving an alleged Chinese spy linked to an Nvidia customer accused of illegally supplying compute to Chinese businesses.
A Policy Reversal
Crucially, this new directive also rescinds the Biden administration’s planned AI Diffusion Rule, which sought broader country-level limits. The Trump administration argued its predecessor's framework imposed unnecessary restrictions on U.S. companies and allies, opting instead for targeted controls on high-power data centers exceeding 12 million total processing performance (TPP) — roughly equivalent to 750 Nvidia H100 chips. This strategic pivot emphasizes preventing access to computing power capable of supporting military and intelligence activities that could threaten U.S. national security.
The new export controls represent a significant hardening of U.S. policy, aiming to comprehensively deny restricted nations, particularly China, access to advanced AI computing capabilities. By targeting remote access to data centers and indirect supply routes, Washington is closing critical loopholes that allowed foreign entities to circumvent existing restrictions on high-performance chips. This move underscores a deepening U.S. commitment to safeguarding its technological advantage, viewing control over cutting-edge AI infrastructure as paramount to national security and a linchpin in the broader strategic competition.
Global AI Landscape Shifts
For the global technology industry, these rules introduce a substantial compliance burden, requiring extensive due diligence on customer ownership and activity across international data center operations. The explicit inclusion of new "higher-risk" countries like Malaysia and Singapore for certain chip exports further signals a broadening U.S. concern over diversion and a more aggressive stance on global supply chain integrity. While the rescission of the Biden administration's broader "AI Diffusion Rule" suggests a more focused, rather than expansive, regulatory approach under the current administration, the overall trajectory points towards a more fractured global AI ecosystem. This strategic tightening is likely to accelerate efforts by restricted nations to develop indigenous AI capabilities, fostering self-sufficiency even at the cost of efficiency, and intensifying the technological arms race. Such measures will fundamentally reshape international collaborations in advanced AI research and deployment, with long-term implications for innovation pathways, economic competition, and geopolitical stability as nations increasingly prioritize national security over globalized technological integration.
Frequently asked questions
- How do new US export controls prevent foreign adversaries from accessing advanced AI computing power?
- The US implemented new export controls to prevent restricted countries, especially those in the D:5 group like China, Russia, and Iran, from remotely accessing advanced AI infrastructure. Companies now require licenses to export or transfer AI chips and servers to data centers worldwide if they might provide computing power to businesses in these countries, undermining their ability to train advanced "Super Intelligence" models for military or intelligence activities.
- What specific actions are US companies now restricted from taking regarding AI chip access?
- US companies must obtain licenses for exporting, reexporting, or transferring AI chips and servers to any data center globally if it could provide computing power to entities in restricted D:5 group countries. This includes indirect access via brokers. Exporters and data center operators must monitor customer ownership to ensure compliance. The rules also introduce new licensing requirements for advanced semiconductors to Malaysia, Singapore, Indonesia, Thailand, and Brazil.
- Which countries face new licensing requirements for advanced AI chip and server exports?
- The primary target of these controls includes China, Russia, Iran, North Korea, and 18 other nations in the D:5 group, which face strict restrictions on accessing advanced AI computing power. Additionally, new licensing requirements now apply to advanced semiconductor exports to Malaysia, Singapore, Indonesia, Thailand, and Brazil. This is due to a "higher-than-normal risk of diversion" in these countries, although these specific license requirements do not apply to chips destined for US-based companies.