June 12, 2026 US confirmed
Status as of June 2026: ten Chinese firms cleared, 75,000 units per customer and zero deliveries in Beijing's limbo
Half a year after the turn, the new regime's operative balance is a documented paradox: the Commerce Department cleared approximately ten Chinese firms —including Alibaba, Tencent and ByteDance— to purchase the Nvidia H200 with a limit of 75,000 units per customer, but no deliveries have been made, with the deal in legal limbo amid the tech rivalry and Beijing's new supply-chain regulations, which discourage its own companies from buying the U.S. chip. Nvidia announced at GTC 2026 that it received purchase orders from Chinese customers and is restarting manufacturing for that market, and its CEO, Jensen Huang, joined President Trump on his May 2026 state visit to China anticipating a breakthrough. The semester's lesson for the sector: in 2026 export controls, Washington's license is no longer the final barrier — the destination country's counter-regulation can freeze what the White House authorizes.
January 14, 2026 US confirmed
25% Section 232 tariff on semiconductors: the fiscal toll accompanying the H200 opening
On January 14, 2026 —the eve of the BIS rule— the White House announced an immediate 25% tariff on a narrow product group: semiconductors with the same performance thresholds as the chips whose export to China was being authorized, following a Section 232 investigation in which Commerce determined imported semiconductors threatened national security. In announcing it, the president expressly tied it to the authorization of H200 exports to China 'subject to 25% of the revenue'. Imports for domestic programs —U.S. data centers, national R&D— are exempt, completing the turn's architecture: Chinese access to compute is administered, conditioned and monetized, all at once.
January 15, 2026 US confirmed
BIS final rule: from presumption of denial to case-by-case for the H200/MI325X, with a 50% compute cap
The BIS final rule, effective January 15, 2026, revised the license review policy for commercially available advanced semiconductors to China and Macau: from presumption of denial —which effectively barred export— to case-by-case review under strict conditions. Eligible 'AI commodities' include the Nvidia H200, the AMD MI325X and their functional or lesser equivalents, provided the exporter certifies: sufficient U.S. supply, that production for China will not divert foundry capacity for similar or more advanced products from U.S. customers, independent third-party testing in the U.S. verifying specifications, rigorous KYC procedures and remote-access safeguards, and that neither recipients nor those given remote access are prohibited end users (Entity List, SDN, military). The quantitative lock: aggregate TPP shipped to China/Macau may not exceed 50% of TPP shipped for U.S. end use — measurable in units or compute capacity, to prevent disproportionate allocation of computing power.
December 8, 2025 US confirmed
Trump announces the H200 may be sold to approved Chinese customers: the regime's political turn
On December 8, 2025, President Trump announced that the United States will allow the Nvidia H200 and similar products to be shipped to approved customers in China, framing it as strengthening national security. The announcement marked the political turn the BIS rule would formalize five weeks later, and inverted the regime's direction in force since 2022: from blocking Chinese access to advanced compute to commercially administering it under conditions. Top-tier chips —the H100 and A100 and their frontier equivalents— remain restricted, drawing a two-level perimeter: the most advanced barred, the commercial tier just below it, negotiable.
November 10, 2025 US confirmed
BIS suspends the Affiliates Rule for a year: the 50% ownership rule that multiplied the Entity List by thousands
On November 10, 2025, BIS suspended for one year —until November 10, 2026— its Affiliates Rule, the interim final rule 'Expansion of End-User Controls to Cover Affiliates of Certain Listed Entities': the mechanism extending export restrictions to any foreign entity at least 50% owned, directly or indirectly, individually or in the aggregate, by Entity List, military end-user list or SDN entities — several thousand entities, mostly in China and Russia, but some within the borders of U.S. allies. The suspension instantly relieved the sector's biggest screening headache (tracing global corporate-ownership chains), but with an expiry date: if the rule reactivates in November 2026, the perimeter multiplies again overnight, and compliance departments plan against that fork.
January 2, 2025 US confirmed
The OISP takes effect: control of U.S. outbound investment in Chinese AI, quantum and chips
In January 2025, the U.S. Treasury's Outbound Investment Security Program (OISP) took effect, created by Executive Order 14105, restricting certain U.S. persons from investing in Chinese companies engaged in critical technologies: artificial-intelligence systems, quantum computing and semiconductors. It is export control's mirror —governing not what leaves in chips but what leaves in capital and know-how— and, per sector tracking in early 2026, it is one of the areas where the current administration has not deviated from the previous one, holding the investment perimeter even while loosening the chip one.
