Export Controls and the Reshaping of Global Chip Supply Chains

A Trillion-Dollar Industry Under New Rules

Global chip demand has never been higher, and the policy environment around it has never been more volatile. The Semiconductor Industry Association (SIA), citing World Semiconductor Trade Statistics data, reported that global semiconductor sales hit a record $791.7 billion in 2025, up 25.6% from $630.5 billion in 2024, with SIA president John Neuffer projecting the industry will approach $1 trillion in 2026. Logic chips the category that includes AI accelerators from Nvidia and AMD were the single largest and fastest-growing product segment, rising 39.9% to $301.9 billion, while memory products grew 34.8% to $223.1 billion. That boom is unfolding in parallel with an unprecedented tightening, loosening, and re-tightening of U.S. export rules aimed squarely at China, and the two forces are now reshaping where chips get designed, built, and sold.


Figure 1: Global semiconductor sales, 2024–2026.

Washington's Shifting Rulebook

From Blanket Bans to Case-by-Case Review

The U.S. Department of Commerce's Bureau of Industry and Security (BIS) has revised its China chip rules repeatedly since 2022, and the pace has accelerated. A December 2024 interim rule added new controls on semiconductor manufacturing equipment, chip design software, and high-bandwidth memory, while adding 140 entities to the Entity List. Then, effective January 15, 2026, BIS shifted its review of advanced AI chips such as the Nvidia H200 and AMD MI325X from a presumption of denial to case-by-case licensing, provided exporters meet new security and know-your-customer requirements a change tied to a presidential proclamation imposing a 25% tariff on advanced computing chips built outside the U.S. and a reported arrangement in which 25% of China sales revenue would flow back to the U.S. government.

The Corporate Cost of Compliance

Nvidia's own SEC filings quantify the disruption. The company's FY2026 Form 10-K describes a 'series of shifting and expanding export control restrictions' dating back to 2022, and its first-quarter FY2026 results included a $4.5 billion charge tied to H20 chip inventory and purchase obligations after new licensing rules, with executives projecting a further $8 billion revenue hit the following quarter. CEO Jensen Huang, who once valued the China data-center opportunity at nearly $50 billion a year and said Nvidia controlled roughly 95% of China's advanced AI chip market, later confirmed China no longer features in the company's baseline forecasts at all, with CFO Colette Kress telling analysts the company does not know whether any further imports will be allowed.

The Reshoring Race

TSMC's Arizona Bet

Nowhere is the supply-chain reshuffle more visible than in Arizona. TSMC's investment in its Arizona campus has grown from an initial $12 billion in 2020 to $65 billion, then to $165 billion in March 2025, and to $265 billion by July 2026 now covering ten wafer fabs, two advanced-packaging facilities, and an R&D center, according to TSMC's own investor relations disclosures. TSMC's 2025 Annual Report notes the company is 'making tangible progress in Arizona, speeding up our capacity expansion,' and Taiwan's National Development Council disclosed that the first Arizona fab turned a $514 million profit in its first full year of mass production, undercutting the long-held assumption that offshore advanced chipmaking could not be profitable.


Figure 2: Cumulative TSMC Arizona investment, 2020–2026.

What Decoupling Would Cost

Reshoring is not free, and neither is restriction. A Boston Consulting Group study cited in an Information Technology and Innovation Foundation (ITIF) report estimated that a full technology decoupling between the U.S. semiconductor industry and China would cut U.S. semiconductor R&D spending by roughly $12 billion, or 30%, as lost China revenue starves reinvestment. An earlier BCG analysis warned that if broad restrictions persist for three to five years, U.S. chipmakers could lose 8 percentage points of global market share and 16% of revenue, with 15,000 to 40,000 skilled U.S. semiconductor jobs at risk even as the same restrictions fail to make China fully self-sufficient, since roughly 70% of Chinese semiconductor demand already has established non-U.S. alternative suppliers.

China's Parallel Push for Self-Sufficiency

China's 2015 'Made in China 2025' plan set a target of 70% domestic semiconductor self-sufficiency by 2025, backed by more than $150 billion in state subsidies through its so-called 'Big Fund.' ITIF's 2025 analysis concludes China will fall well short of that target, landing closer to 30% self-sufficiency by year-end. Yet export controls have not stopped China's chip trade from expanding: research cited by Tom's Hardware in mid-2026 found China's semiconductor exports nearly doubled to $177 billion in the first half of 2026 alone, as memory prices surged and domestic suppliers such as Huawei moved to fill gaps left by restricted Nvidia shipments.