WASHINGTON, The Bureau of Industry and Security issued its broadest-ever semiconductor export control rule on October 7, 2022 (87 FR 62186), blocking China from acquiring advanced computing chips above 600 TOPS performance thresholds and barring the sale of fabrication equipment capable of producing sub-16nm logic. The rule directly affects Nvidia, Intel, TSMC, Huawei, and SMIC, and has reshaped global chip supply chains in every quarter since.
The Legal Framework: EAR, BIS, and the Entity List
BIS administers the Export Administration Regulations under the Export Control Reform Act of 2018 (50 U.S.C. § 4801 et seq.), codified at 15 C.F.R. Parts 730–774. The EAR govern the export, reexport, and transfer of dual-use goods: a category that now includes virtually every advanced semiconductor above a specified performance threshold.
The Entity List (15 C.F.R. Part 744, Supplement 4) is BIS’s primary targeting instrument. Companies placed on the list require a license, presumed denied, for U.S.-origin goods. Huawei was added in May 2019 after Commerce found the company was acquiring U.S. technology in violation of export law. SMIC, China’s largest domestic foundry, was added in October 2020, effectively blocking access to EUV lithography equipment from ASML and cutting off advanced process node technology from U.S.-origin sources.
General Prohibition 10 (GP10) of the EAR extends enforcement reach: it prohibits any person from proceeding with a transaction knowing a violation has occurred. BIS invoked GP10 formally in May 2025 guidance targeting Huawei Ascend AI chips, declaring that any use, export, reexport, or servicing of the Ascend 910B, 910C, or 910D constitutes an ongoing EAR violation absent prior BIS authorization.
Foreign Direct Product Rules, codified at 15 C.F.R. § 736.2(b)(3), extend EAR jurisdiction to goods manufactured outside the United States when produced using U.S.-origin equipment or software. The Huawei and SMIC FDP Rules, expanded multiple times since 2020: are why TSMC halted shipments to Huawei in September 2020 despite being a Taiwan-based company.
October 2022: The Rule and What It Blocked
The October 7, 2022 interim final rule (87 FR 62186) established three interlocking controls that remain the foundation of U.S. chip export policy.
First, the rule restricted advanced computing chips destined for China: any chip exceeding 600 TOPS aggregate performance, or 300 TOPS with a performance density above 10 TOPS per mm², required a license. This threshold captured Nvidia’s A100 and H100 GPUs, the workhorses of AI model training, along with equivalent chips from AMD. Companies exporting below-threshold chips designed with intent to aggregate performance above the limit also required a license.
Second, semiconductor manufacturing equipment controls under 15 C.F.R. § 742.6 blocked export of tools capable of producing logic chips at or below 16nm, DRAM at or below 18nm, and NAND flash with 128 or more layers. Applied Materials, Lam Research, and KLA, the three dominant U.S. semiconductor equipment suppliers: could no longer ship advanced deposition, etch, or metrology tools to Chinese fabs without a license presumed denied for advanced node production.
Third, the “U.S. persons” rule effectively prohibited U.S. citizens, permanent residents, and corporations from supporting advanced chip production in China without a license. The provision forced several senior engineers at SMIC and YMTC, Chinese nationals holding U.S. green cards, to resign their positions within weeks of publication.
The immediate commercial impact was substantial. Nvidia disclosed in its FY2023 10-K that the October 2022 rule resulted in a $400 million charge for excess A100 and H100 inventory that could no longer ship to China. The company subsequently designed the A800 and H800 as China-compliant alternatives, chips that fell just below the threshold at the time of their release.
2023–2024: Closing the Workarounds
BIS published updated rules in October 2023 specifically to block the A800 and H800 loophole. The revised thresholds for China were reduced: chips above 4,800 TOPS total performance, or meeting revised density parameters, now required a license. The A800 (3,500 TOPS) and H800 (3,958 TOPS) were blocked. BIS simultaneously expanded geographic controls, adding licensing requirements for chips destined to approximately 40 additional countries assessed as presenting diversion risk to China.
The 2023 rule introduced “country A” controls: a licensing tier for advanced chips shipped to allied nations including the Netherlands, Japan, and South Korea in quantities that could plausibly be diverted. The practical effect was to require end-user documentation for large datacenter purchases in allied countries involving chips above the threshold.
In November 2024, the Biden administration published its final tightening package. New advanced packaging controls targeted outsourced assembly, test, and packaging operations that could be used to integrate U.S.-origin chiplets into systems assembled in China. The rule expanded the FDP Rule to capture additional products manufactured abroad using U.S.-origin EDA (electronic design automation) software, a provision targeting Chinese fabless companies that design chips in American tool environments and then manufacture at non-U.S. fabs.
YMTC, China’s largest NAND flash producer, was added to the Entity List in December 2022 over concerns that it had supplied advanced flash to Huawei in violation of existing controls. The Entity List now contains more than 600 Chinese entities subject to presumption-of-denial licensing policy for advanced semiconductor goods.
China’s Response: Domestic Chip Push and Workarounds
Beijing’s response to BIS export controls has proceeded on two tracks: accelerated domestic investment and aggressive procurement workarounds.
The Chinese government committed approximately $47.5 billion in the third tranche of the National Integrated Circuit Industry Investment Fund (“Big Fund III”) in May 2024, bringing total government-directed semiconductor investment since 2014 to over $100 billion. The primary beneficiaries are SMIC, CXMT (DRAM), YMTC (NAND), and HiSilicon (fabless design).
SMIC’s technical progress remains constrained by equipment controls. The foundry cannot acquire EUV lithography from ASML, the single-source supplier, because EUV systems contain U.S.-origin technology and ASML’s Dutch government export license for China was revoked in January 2023. SMIC demonstrated 7nm-class production in 2022 using multi-patterning DUV techniques, but yield rates remain well below TSMC’s equivalent processes, and scaling to 5nm without EUV is assessed by industry analysts as commercially implausible.
Huawei’s Ascend AI chips: the 910B, 910C, and 910D, are designed by HiSilicon and fabricated by SMIC. BIS stated in its May 2025 GP10 guidance that all three models were produced using U.S.-origin technology without required authorization. SemiAnalysis estimated Huawei received 2.9 million Ascend die from TSMC before TSMC’s 2020 cutoff, shipments BIS characterized as unauthorized under the Huawei FDP Rule. Any entity that uses, finances, or services an Ascend chip is now on notice that it carries an ongoing EAR violation.
Front companies and third-country procurement have been documented extensively. Commerce added 42 Chinese entities to the Entity List in March 2025 and another 23 in September 2025, the majority identified as procurement conduits routing restricted chips through Malaysia, Singapore, and the UAE.
What’s Next: Trump Administration Direction
The Trump administration’s approach has been to selectively tighten controls on China while unwinding the multilateral framework the Biden administration built through the AI Diffusion Rule.
BIS rescinded the Biden-era AI Diffusion Rule on May 13, 2025, citing concerns that the tiered country framework was overly bureaucratic and risked damaging relationships with allied partners. The rescission removed the global licensing tier system that had applied to chips destined for Tier 2 and Tier 3 countries, though China-specific controls under 15 C.F.R. § 742.6 remain in force.
Nvidia’s H20, designed to fall below the October 2023 China thresholds: was placed under license requirement in April 2025, blocking an estimated $5.5 billion in anticipated China revenue according to the company’s Q1 2025 earnings disclosure. BIS reversed course in July 2025, issuing assurances that H20 licenses would be approved under a case-by-case framework requiring end-user compliance commitments and third-party performance verification.
A January 2026 Federal Register notice (91 FR 3421) revised the license review policy for H200-class chips exported to China, establishing a presumption of approval for qualifying customers that adopt BIS-specified export compliance procedures. The rule implements a revenue-sharing mechanism announced by President Trump in December 2025: approved exporters remit 25 percent of China semiconductor revenues to the U.S. government, a construct without precedent in EAR history and currently subject to legal challenge in the U.S. Court of International Trade.
The core China controls — fabrication equipment restrictions, Entity List presumption-of-denial, the U.S. persons rule, and the FDP Rules — remain in force. BIS has indicated it is drafting a replacement framework for the rescinded AI Diffusion Rule, expected in late 2026. The direction of that framework will determine whether the multilateral control architecture built between 2022 and 2025 endures or is superseded by bilateral arrangements.
For analysis of domestic manufacturing investment designed to reduce U.S. dependence on foreign fabs, see CHIPS Act Semiconductor Investment: What the $52 Billion Means for U.S. Fab Capacity.