In brief
- The rule covers data centers and networks that can exceed 12 million in total processing performance.
- Licenses would be required for chip and server transfers that could provide remote computing power to companies in restricted countries.
- Advanced semiconductor transfers to Malaysia, Singapore, Indonesia, Thailand, and Brazil would also be subject to new licensing requirements.
- Existing data center operators would need to obtain a license within 90 days after the rule takes effect.
The US administration is preparing a new temporary export rule to restrict the provision of remote computing power to restricted countries through AI chips and data centers. According to a 16 November 2026 report by TechCrunch, the US Department of Commerce plans to require licenses for the export, re-export, or transfer of chips and servers. The rule targets cases in which this hardware is sent to data centers around the world and those facilities can provide computing power to companies in restricted countries.
Which data centers are covered?
One threshold under the new rule is exceeding 12 million in total processing performance (TPP), a measure the US uses to assess computing power. The report describes this capacity as equivalent to the maximum processing power of about 750 Nvidia H100 or H200 chips. The rule covers not only direct service provision but also indirect access through intermediaries or brokers.
Exporters and data center operators will need to track their customers’ ownership structures and activities. The aim is to prevent companies from restricted countries from using this infrastructure indirectly. The source says the US D:5 group includes China, Russia, Iran, and North Korea, as well as 18 other countries.
Transfers to five countries also require licenses
The rule also introduces new licensing requirements for the export, re-export, or transfer of advanced semiconductors to Malaysia, Singapore, Indonesia, Thailand, and Brazil. The US Department of Commerce considers these countries to pose a higher-than-normal risk of products being diverted elsewhere.
The stated licensing requirement for these five countries will not apply to chips intended for shipment to US-based companies. Data center operators already providing services covered by the rule will need to obtain a license within 90 days after it takes effect. The source does not specify when the rule will take effect.
Previous AI export framework to be repealed
According to the report, the new rule will also repeal the Biden administration’s AI Diffusion Rule, introduced in January 2025. That earlier framework aimed to impose country-based limits on exports of advanced AI chips and restrict the transfer of certain powerful AI models. The Trump administration had previously announced it would abandon this approach.
The new rule does not extend to every type of online service: according to the source, commercial recommendation systems used in e-commerce and social media are excluded. The focus is on specific computing power and infrastructure that could be used to train or run advanced AI models.
What could change for architecture and archviz workflows?
For architecture and visualization teams, the news is less about direct changes to design software compliance and more about access to remote computing resources and the licensing status of those resources. Teams that train or run AI in cloud infrastructure may need to pay closer attention to their provider’s data center capacity, how customers access it, and applicable licensing requirements. The source does not say that the rule covers every render or cloud workflow; its scope depends on the stated chip and capacity thresholds.
A visualization studio’s first practical step is to establish which data center and service provider supply its remote computing power. It may also help reduce planning risks, particularly for AI workloads, to ask the infrastructure provider about its licensing status under the rule and its approach to service continuity. The source provides no information on whether local hardware or a particular software version will be affected.
Sources
1 sourceSource texts are not republished; short quotes are marked, everything else is our own summary and commentary.
For architecture and archviz teams in Turkey, the potential impact of this development lies less in the software they use than in which infrastructure provides their remote computing power. Offices using cloud resources to train or run AI should review their provider’s licensing status and service continuity.
The rule’s scope depends on thresholds and links to specific countries; it cannot be assumed that every cloud rendering service will be affected. Before investing in new hardware, asking the provider for clear information would be a more cautious approach to licensing requirements and access risks.
Frequently asked questions
Which data centers are covered by the new US AI chip rule?
According to the source, the rule covers data centers or networks that can exceed 12 million in total processing performance. Indirect access is also included.
When will the new export rule take effect?
The source does not specify an effective date. It says existing data center operators must obtain a license within 90 days after the rule takes effect.
Transfers of chips to which countries will require additional licenses?
New licensing requirements are proposed for advanced semiconductor transfers to Malaysia, Singapore, Indonesia, Thailand, and Brazil. Chips intended for shipment to US-based companies are exempt.



