NewsMacroUS BIS Issues Interim Final Rule Capping Overseas AI Chip Access at 12 Million TPP

US BIS Issues Interim Final Rule Capping Overseas AI Chip Access at 12 Million TPP

Author: Coinotag·

Key Takeaways

  • •The US Commerce Department's Bureau of Industry and Security issued an interim final rule extending advanced-chip export controls to large-scale remote computing capacity supplied by overseas data centers.
  • •The rule caps remotely delivered advanced AI computing power at 12 million total processing performance (TPP), a metric combining chip processing throughput with interconnect bandwidth.
  • •License requirements now apply to five countries, including Singapore, while the identities of the four other covered destinations have not been disclosed.
  • •The measure is intended to close a regulatory gap in which users in restricted jurisdictions could rent advanced AI capacity from data centers in permissive countries instead of purchasing restricted hardware.
  • •Because it is an interim final rule, the requirements take effect immediately but remain open to public comment, leaving possible changes to the TPP cap or destination list unresolved.
US BIS Issues Interim Final Rule Capping Overseas AI Chip Access at 12 Million TPP

November 16, 2026 — The US Commerce Department's Bureau of Industry and Security (BIS) has issued an interim final rule that extends advanced-chip export controls to large-scale remote computing capacity supplied by overseas data centers, and adds license requirements for five countries, including Singapore.

Under the new framework, access to advanced AI computing power delivered remotely from data centers located outside the United States is capped at 12 million total processing performance (TPP). TPP is a metric that combines a chip's processing throughput with its interconnect bandwidth, and it is the benchmark BIS uses to classify high-end AI chips under its export control regime. By applying that existing classification benchmark to remotely delivered compute, the rule measures data-center capacity on the same scale used to assess high-end AI chips themselves.

BIS, the Commerce Department agency responsible for administering US export controls on dual-use and advanced technologies — goods with both civilian and military applications — has progressively tightened restrictions on advanced AI semiconductors since October 2022, when Washington first imposed licensing requirements on exports of high-end AI accelerators to destinations considered of concern, including China. Successive rule updates in 2023 and 2024 broadened the scope of those controls and lowered the performance thresholds at which chips become subject to licensing.

Policymakers have increasingly viewed remote computing services as a potential channel for obtaining advanced AI compute without directly purchasing restricted hardware, since a user in a restricted jurisdiction can rent capacity from a data center in a permissive one. The extension of export controls to large-scale remote computing capacity is aimed at addressing that gap, bringing data-center-supplied compute within the same regulatory perimeter as physical chip shipments.

As an interim final rule, the measure takes effect upon publication, which means the new licensing requirements apply now, while the text remains open to public comment before any final modification. How the comment period shapes the final rule — including whether the 12 million TPP cap or the destination list is adjusted — is the immediate variable to watch. The four countries joining Singapore on the licensing list have not been named, and their identification would establish the rule's full geographic scope.

This content was first published on COINOTAG.