How Export Controls Reshape the Semiconductor Map

Export controls work by targeting choke points — the small number of companies and technologies without which a modern chip cannot be designed or manufactured — rather than trying to police every transistor. The three chokepoints Washington leans on repeatedly are electronic design automation (EDA) software needed to design any advanced chip, extreme ultraviolet (EUV) and advanced deep ultraviolet (DUV) lithography tools needed to print sub-10nm features, and high-bandwidth memory and advanced packaging needed to build AI accelerators. Control any one of these and you control who can participate in the leading edge of the industry, regardless of how much raw fabrication capacity a country has.

This is why the same handful of names reappear every time a new rule is announced: they are not incidental beneficiaries of one policy, they are structural gatekeepers of the whole regime. Understanding this distinction is the difference between chasing headlines and holding a durable thesis — the specific restricted GPU model or the specific banned tool changes with each rule revision, but the companies that own the chokepoint rarely do.

EDA software is a duopoly: Synopsys (SNPS) and Cadence Design Systems (CDNS) together control the overwhelming majority of the tools used to design advanced semiconductors worldwide. Every time BIS tightens rules on what design software can be licensed to Chinese customers, it reinforces the value of the license itself — these firms get to renegotiate terms, geofence products, and sell tiered ('compliant') versions, capturing revenue either way the policy cuts.

Lithography is even narrower: ASML (ASML), based in the Netherlands but subject to U.S. re-export rules on its EUV and immersion DUV tools, is the sole global supplier of EUV systems. Applied Materials (AMAT), Lam Research (LRCX), and KLA Corporation (KLAC) round out the U.S.-listed toolmaker cohort supplying deposition, etch, and inspection equipment — every one of them has flagged China revenue headwinds from controls, but every one of them has also benefited from the offsetting subsidy wave (see below) as fabs everywhere else expand to backfill capacity that can no longer serve China directly.

On the beneficiary side of restricted demand redirection: Nvidia (NVDA) has repeatedly redesigned compliant chip variants for the Chinese market when full-spec parts were restricted, preserving a revenue channel policy tried to close. Broadcom (AVGO), which supplies custom AI accelerator silicon and networking chips, and Marvell Technology (MRVL), a similar custom-silicon and interconnect supplier, both benefit as hyperscalers diversify AI infrastructure buildouts across more suppliers and more geographies in response to a more fragmented, controlled world.

Export controls are only half of Washington's semiconductor policy. The other half is the CHIPS and Science Act, which pairs restrictions on China with direct subsidies to build capacity everywhere else. Intel (INTC) has been the largest single recipient of CHIPS Act award commitments as it builds out Arizona and Ohio fabs. Taiwan Semiconductor Manufacturing Company (TSM), the dominant foundry for leading-edge chips globally, has received subsidy commitments and tax credits for its Arizona fab buildout — a direct hedge against Taiwan Strait geopolitical risk that export-control policy explicitly incentivizes.

Micron Technology (MU), the only U.S.-headquartered maker of DRAM and NAND memory, has been both a target of Chinese retaliatory restrictions (Beijing barred some Micron sales to Chinese critical infrastructure operators in 2023) and a subsidy recipient for domestic memory fab expansion — a clean illustration of how the same company sits on both sides of the ledger simultaneously. GlobalFoundries (GFS), a specialty and legacy-node foundry, has also drawn subsidy support as policy pushes 'trusted' supply chains for less-advanced but still strategically important chips used in autos and defense.

The mechanism to track here is simple: every dollar of restricted China revenue that a chokepoint company loses is, over a multi-year horizon, partially offset by a dollar of subsidized domestic or allied-country capacity investment — often flowing to the same company or its direct suppliers.

The mirror image of the chokepoint thesis is exposure risk: companies with a high concentration of revenue tied to Chinese data-center, telecom, or fabrication customers face repeated air-pocket risk every time BIS issues a new rule, because compliance often requires an immediate halt in shipments while product lines are redesigned or licenses are renegotiated. Nvidia has taken the highest-profile hits here, disclosing charges tied to unsellable inventory when specific chip models were abruptly added to restricted categories.

Applied Materials, Lam Research, and KLA all derive a meaningful share of revenue from Chinese semiconductor customers buying legacy-node equipment (the trailing-edge capacity China is racing to build since it cannot access leading-edge tools) — a nuance worth tracking separately from EUV exposure, since legacy-node equipment sales to China have at times been less restricted than leading-edge tools, creating a partial offset within the same companies. Qualcomm (QCOM), with substantial revenue from Chinese smartphone OEMs, carries indirect exposure any time controls widen to touch handset-adjacent components.

The durable lesson: exposure is not static. A company can be a net loser under one rule revision and a net beneficiary of the next, depending on whether the specific rule closes a market (bad for revenue) or forces a competitor's product off the shelf (good for market share). Read every new BIS rule for both effects, not just the headline restriction.

Every escalation in U.S. export controls accelerates China's own subsidized push for semiconductor self-sufficiency — this is the third leg of the mechanism, and it matters to U.S.-listed investors indirectly rather than directly, since the primary domestic beneficiaries (SMIC, Naura Technology, and similar Chinese toolmakers) are not U.S.-listed and are frequently themselves Entity-Listed, making them uninvestable or untrackable through normal U.S. brokerage channels.

The investable read-through instead runs through the non-Chinese suppliers who back-fill the equipment and materials China's push displaces from the restricted channel, and through U.S. defense and government-services names that benefit from the broader strategic-competition budget cycle this rivalry sustains. This is a slower-moving, higher-uncertainty channel than the direct chokepoint trade — treat it as context for why the policy exists and persists, not as a primary place to look for tickers.

The Bureau of Industry and Security publishes export control rule changes directly in the Federal Register and on the Commerce Department's BIS website — this is the primary source, and it is public and free, well ahead of most financial media summarizing it. Company 10-K and 10-Q filings from every name above routinely include a 'China' or 'export control' risk-factor section that gets more specific and more quantified with each rule cycle — read the risk factors, not just the earnings headline, because that's where management discloses revenue concentration and inventory-charge exposure before it shows up in results.

The practical pattern-match: when a new rule targets a specific chip performance threshold or a specific equipment category, ask three questions. First, who owns the chokepoint the rule is reinforcing (EDA, lithography, packaging, memory)? Second, which named companies have disclosed China revenue that intersects the restricted category? Third, is there an offsetting CHIPS-style subsidy or reshoring announcement moving in parallel? Those three answers, cross-checked against each company's own filings, are the entire durable framework — the specific chip model or tool banned this quarter is just the trigger, not the thesis.