US Tightens China’s AI Chip Access Before Trump-Xi Summit

The U.S. Commerce Department has introduced new controls on access to advanced American AI chips through data centers outside China, just ahead of President Donald Trump’s meeting with Chinese President Xi Jinping in Shenzhen. The rules extend scrutiny beyond physical chip shipments to remote computing services. For Nvidia (NASDAQ: NVDA) shareholders, the key issue is whether licensing delays and customer restrictions reach sales channels outside mainland China. The Commerce Department announced the interim final rule on November 16, 2026.

What the new US AI chip export controls actually do

The new policy targets an important distinction in semiconductor trade. A Chinese AI developer may be unable to import a restricted accelerator directly but could potentially rent computing capacity at a foreign data center that owns the hardware. The interim final rule adds a license requirement when covered advanced chips are destined for a data center that will provide significant computing power to companies headquartered in China or another restricted destination, including through linked data centers or intermediaries.

The threshold is more than 12 million aggregate Total Processing Performance (TPP), a U.S. export-control measure of computing capability. It is not a blanket prohibition on every overseas cloud service. Regulators will review qualifying license applications case by case, and the rule explicitly excludes qualifying commercial recommendation systems used by retailers and social media platforms.

Five countries face additional licensing requirements

The policy also introduces destination-based licenses for covered advanced semiconductors shipped to Singapore, Malaysia, Thailand, Indonesia and Brazil. The government identifies these markets as carrying elevated diversion risk. This provision is distinct from the restriction on computing supplied remotely to companies in China.

A limited exception applies to shipments destined for companies headquartered in the United States, or whose ultimate parent is U.S.-headquartered, when the specified reporting conditions are met. Existing overseas data centers covered by the remote-access provision must obtain authorization within 90 days of the rule’s effective date to continue qualifying services. The draft specifies that the effective date is tied to publication in the Federal Register, so the deadline should not be counted from the press announcement. The published regulatory text sets out those conditions.

Why closing the remote-computing route matters

Export restrictions have traditionally focused on where sensitive semiconductors are shipped and who receives them. But remote access allows a customer in one country to use hardware installed somewhere else. The latest measure makes data-center customers, ownership structures and connected computing clusters more important to compliance decisions.

ALSO  Applied Materials Revenue Rises 53% as AI Chip Demand Lifts 2027 Outlook

Washington is also formally withdrawing the broader Biden-era AI Diffusion Rule, which the Trump administration had previously said it would not enforce. The replacement is more targeted: it combines tighter scrutiny of certain foreign users with explicit licensing requirements for five destinations rather than applying the abandoned worldwide framework. BIS says the objective is to prevent workarounds through third-country data centers.

What Nvidia and AMD shareholders should watch

Nvidia’s fiscal second-quarter 2027 results illustrate the stakes. The chipmaker reported $96.2 billion in total revenue and $89.0 billion in Data Center revenue for the quarter ended July 26, 2026. Its forecast for fiscal third-quarter revenue was approximately $108 billion, plus or minus 2%, and it expressly assumed no Data Center compute revenue from China. Those figures appear in Nvidia’s August 26 earnings release.

That guidance limits the immediate relevance of directly forecast China sales, but it does not resolve how overseas operators, distributors or third-country end customers may be affected. Approval timelines, compliance costs and canceled or redirected projects could matter even where a supplier has no planned direct shipment to China. Earlier Nvidia chip-diversion allegations explain why tracing the final user of computing capacity is already a shareholder concern.

Advanced Micro Devices (NASDAQ: AMD) is also exposed to advanced-computing export policy, although the rule alone provides no reliable estimate of lost AMD orders. Investors comparing the two suppliers should focus on the difference between their existing AI businesses, as outlined in the AMD-versus-Nvidia revenue comparison, rather than assume that restrictions automatically shift demand from one to the other.

Why the Trump-Xi meeting adds uncertainty

Trump is due to meet Xi on Thursday, November 19, in Shenzhen, not Beijing, during the APEC gathering, according to Reuters. The announcement puts technology access on the agenda immediately before their talks, alongside broader trade and AI-safety discussions.

Investors should distinguish a diplomatic statement from a change in enforceable rules. Any future concessions, exemptions or enforcement revisions would need specific details before analysts could adjust expected chip sales or customer capital spending.

Three signals for the next earnings and policy updates

  • Licensing outcomes: Whether affected data-center operators receive approvals and how long processing takes.
  • Company disclosures: Whether Nvidia, AMD or their customers quantify orders delayed, redirected or lost because of the controls.
  • Cloud spending: Whether infrastructure operators maintain deployment and cash-flow plans as customer-screening obligations expand. CoreWeave’s recent earnings analysis provides a useful view of the financing risks around AI data-center construction.
ALSO  Anthropic IPO: Everything Investors Need to Know Before ANT Starts Trading

Bottom line: The new policy raises compliance and execution risk in a part of the global AI market that previously relied on remote access to U.S. chips. For shareholders, the financial impact depends on actual licenses, customer demand and supplier guidance, not simply the announcement or the upcoming summit.

Data as of 1:40 p.m. ET, November 16, 2026. For informational purposes only; not investment advice.