On November 16, 2026, the U.S. Department of Commerce’s Bureau of Industry and Security issued an interim final rule designed to prevent Chinese entities from remotely accessing advanced AI chips and servers through data centers in third countries. The rule adds Malaysia, Thailand, Indonesia, Singapore, and Brazil to a list requiring licenses for exports of advanced semiconductors and imposes enhanced due-diligence and reporting obligations. Technology and infrastructure teams that operate or contract for compute capacity in those countries should review current and planned deployments before the rule takes effect 90 days after Federal Register publication.
The interim final rule closes a well-known circumvention path: renting advanced U.S.-origin AI compute in Southeast Asia or Brazil rather than shipping chips directly to restricted destinations. Cloud, data-center, and AI-infrastructure operators need to map their exposure and licensing posture now.
This is not legal or export-compliance advice. Export-control rules are technical and fact-specific; consult qualified counsel and your export-compliance officer before changing shipments, contracts, or remote-access arrangements.
- What changed? Interim final rule restricts remote access to advanced AI chips via data centers in five countries and requires licenses plus robust know-your-customer controls and monthly reporting for high-volume transfers.
- When? Issued November 16, 2026. Takes effect 90 days after posting to the Federal Register. Exceptions may apply for U.S.-headquartered companies under certain reporting conditions.
- Who is affected? Chip exporters, data-center operators, cloud providers, and AI companies that place or access advanced compute in Malaysia, Thailand, Indonesia, Singapore, or Brazil, or that serve customers with ties to China, Russia, or Iran.
- What to do this week? Inventory compute capacity and customer relationships in the five countries. Confirm whether any arrangements allow remote access by restricted parties. Engage export-compliance counsel on license strategy and due-diligence documentation.
What did the Bureau of Industry and Security announce?
The Bureau of Industry and Security (BIS) issued an interim final rule aimed at ensuring that existing restrictions on advanced AI semiconductors cannot be circumvented through remote computing arrangements. Previously, entities in China or other restricted destinations could access high-end U.S.-origin chips by renting capacity in data centers located in third countries that were not themselves subject to the same licensing requirements.
The rule adds Malaysia, Thailand, Indonesia, Singapore, and Brazil to the list of destinations that require licenses for the export, re-export, or in-country transfer of advanced semiconductors. These countries were identified as presenting a higher-than-normal risk of diversion. Data centers, overseas foundries, and exporters must perform enhanced due diligence to confirm that customers and their parent companies are not on restricted lists. Where licenses are granted, monthly reporting on parties and final destinations is required.
U.S.-headquartered companies may qualify for exceptions when chips are destined for entities with U.S. headquarters or U.S. parent companies, provided reporting conditions are met. The rule is effective 90 days after it appears in the Federal Register.
How does the remote-access restriction work?
The policy treats remote use of advanced chips in the listed countries as a controlled activity when it could benefit restricted end users. Operators must implement know-your-customer procedures that look through to ultimate ownership and beneficial use. High-volume transfers trigger additional documentation and ongoing reporting. The goal is to prevent a data-center rental model from substituting for a direct shipment that would have required a license.
For technology teams, this means that contractual remote-access rights, multi-tenant cloud arrangements, and capacity reservations in the five countries now carry export-control implications even if the physical chips never leave the data center.
Why is the rule appearing now?
The measure follows heightened scrutiny of export-control enforcement after the resignation of Nvidia’s former CEO and related questions about possible diversion pathways. Administration officials have described the rule as closing a loophole that allowed restricted parties to obtain advanced compute capacity without triggering the direct-export license requirement. Parallel discussions about designating certain U.S. AI companies as “national champions” and about allied supply-chain security declarations form the broader policy backdrop.
For operators, the practical effect is tighter control over where advanced AI infrastructure can be placed and who can use it, regardless of the physical location of the servers.
What should technology and infrastructure teams do this week?
Map all current and planned advanced-compute deployments in Malaysia, Thailand, Indonesia, Singapore, and Brazil. Identify customers or partners that could be linked, directly or indirectly, to restricted destinations. Review contracts for remote-access clauses, sub-leasing rights, and audit provisions. Coordinate with export-compliance and legal teams to determine whether license applications or enhanced diligence packages are needed before the 90-day effective date.
If your organization is U.S.-headquartered, confirm whether the parent-company exception applies to specific flows and what reporting will be required. Update internal playbooks so that new capacity requests in the five countries trigger an export-control review before contracts are signed.
What should operators watch next?
Monitor the Federal Register for the official posting that starts the 90-day clock. Watch for BIS guidance, frequently asked questions, or license-exception clarifications. Parallel allied measures, such as the San Francisco Declaration on AI supply-chain security signed by the United States and nine partners, may produce additional coordination requirements. Companies that rely on Southeast Asian or Brazilian data-center capacity for AI training or inference should also track customer demand shifts and any secondary-market effects on GPU availability.
The rule reinforces that export controls now reach remote use as well as physical shipment, a development that will shape data-center site selection and customer onboarding for the next several years.
Does the rule ban all AI chips in the five countries?
No. It requires licenses for advanced semiconductors and imposes due-diligence and reporting conditions. Exceptions may exist for U.S.-headquartered end users under specified reporting rules.
When does the rule take effect?
Ninety days after it is posted to the Federal Register. The interim final rule was issued on November 16, 2026.
What is “remote access” in this context?
Use of advanced chips located in the listed countries by parties outside those countries, particularly when the use could benefit restricted end users such as Chinese entities.
Are cloud providers affected?
Yes. Operators of data centers and cloud services in the five countries, as well as customers that rent capacity there, fall within the scope of the due-diligence and licensing requirements.
How does this relate to earlier China export controls?
It closes a circumvention path that allowed restricted parties to obtain compute capacity without a direct shipment of chips into China or other restricted destinations.
Last Updated: November 17, 2026.
Related reading: Cisco Reports Record $18.5 Billion Q1 Revenue and Raises FY2027 Guidance · Federal Agencies Extend Comment Period on Third-Party Risk Management · Nvidia Names Colette Kress Interim CEO After Jensen Huang Resigns
Explore more on the Technology News hub.
Discover more from Kurums | Business Intelligence
Subscribe to get the latest posts sent to your email.