Washington’s Overlooked Leverage Over China’s Chip Ambitions

Later this year, China’s leading chipmakers may receive a domestically built advanced lithography machine, which prints tiny circuit patterns onto silicon. Its arrival marks a step toward a Chinese alternative to ASML (originally Advanced Semiconductor Materials Lithography), the Dutch company that dominates this equipment market. China relies on just 343 deep ultraviolet (DUV) lithography machines for advanced chipmaking because existing export controls prevent it from acquiring ASML’s more advanced extreme ultraviolet (EUV) machines or even new DUV ones. These DUV tools can substitute for EUV in some advanced processes but require more manufacturing steps, costs, and production time.

But building a machine does not make it production-ready. China still has to transition from “it works” to “it works every day.” That requires repeatedly, accurately, and quickly printing aligned nanometer (nm)-scale circuit patterns. Continued Western equipment servicing in Chinese fabs—or chip factories—could help China cross the gap. The next few months could be especially consequential as China tries to qualify its domestic scanners for mass production. These benefits of mass producing Chinese DUV scanners are not just hypothetical. Washington’s new tool sale restrictions might slow China’s artificial intelligence (AI) and military chipmaking, but China still makes advanced chips with regularly serviced ASML DUV machines.

Congress has proposed broader servicing restrictions, but the Netherlands has opposed extending U.S. controls over ASML. Washington already has a narrower option, though. It can enforce existing rules against servicing specific export-regulated items connected to export control violations, slowing China’s effort to make its own scanners production-ready

China’s Hardest Work Is Post-prototype

Fabs build circuits on silicon discs called wafers. Lithography machines, also called scanners, use light to print patterns onto those wafers. China’s new machine uses immersion DUV lithography, which adds water between the lens and wafer to print finer patterns. Immersion DUV uses 193 nm light compared to EUV’s much shorter 13.5 nm wavelength. Shorter wavelengths print smaller features in fewer steps, making advanced chipmaking cheaper and faster. Fab engineers can compensate for DUV’s longer wavelength by repeating multi-patterning steps on a single circuit layer. SMIC already uses DUV to make 7 nm and smaller chips. Since existing DUV scanners let SMIC continue making advanced chips despite U.S. export controls, limiting China’s homegrown DUV capabilities is critical. Still, compared to EUV, each additional DUV patterning step increases production time, costs, and defects.

EUV access remains a major U.S. and allied chipmaking advantage. But potential homegrown DUV machines still matter. If Chinese engineers make those scanners reliable enough for mass production, China’s fabs could significantly expand advanced domestic chip production. Because advanced fabs use both DUV and EUV for different chip layers, DUV would remain incredibly useful even if China eventually develops viable EUV machines.

On July 27, The Information reported that a state-backed Shanghai company had begun producing China’s first domestic immersion DUV machines. Reuters subsequently identified the firm as Shanghai Aishengna Electronic Technology Group, citing an anonymous person familiar with the matter. Founded just three years ago, the secretive firm lacks even a website but has roughly $1 billion in registered capital and two state-owned shareholders. After absorbing teams from Huawei-linked Yuliangsheng, which began testing a DUV prototype in 2025, and Shanghai Micro Electronics Equipment (SMEE), Aishengna now plans to produce five systems this year and 20 in 2027. These could help China go from producing no more than 20,000 7-nm chips per month in early 2026 to 100,000 by 2028. Its prospective 2026 customers include major Chinese chipmakers SMIC, Hua Hong Semiconductor, and CXMT.

While these production targets are eye-catching, they do not yet translate into successful chip manufacturing. ASML shipped 131 immersion DUV systems in 2025 and still captures over 99 percent of the market. Besides matching a mature ASML scanner’s ability to print fine patterns, Aishengna must align successive circuit layers on each wafer precisely and reliably. Chinese fabs still need months or longer to test homegrown DUV machines’ accuracy, reliability, and compatibility with other equipment before integrating them into mass production. Even if China’s “homegrown” DUV machines can effectively substitute for ASML’s, Aishengna’s dependence on key Japanese scanner components and local supplier delays constrain machine output.

China has historically sought overseas expertise for its chipmaking efforts. Reuters reported in December 2025 that former ASML engineers had helped build a separate EUV light-generating prototype in Shenzhen. China’s DUV efforts may be similarly accelerated. Permitting foreign technicians to continue China-based servicing and training efforts thus accelerates China’s homegrown chipmaking push. After all, how quickly China closes the remaining gap depends heavily on how much its engineers can learn.

Service Calls Teach Valuable Lessons

Working fabs equipped with advanced foreign tools simplify qualifying a homegrown machine for mass production. Chinese engineers can benchmark a prototype Aishengna scanner against a well-serviced ASML fleet. They could adapt ASML’s processes to maximize Aishengna output and compatibility with other chipmaking tools.

“Servicing” might sound like a technician changing a part and leaving, but for semiconductor manufacturing, it is much closer to sustained co-production. Western equipment vendors maintain software, replace components, diagnose performance, increase the share of usable chips (yield), and train local fab engineers. ASML explicitly describes “on-site knowledge transfer sessions” as part of its customer support.

ASML’s Chinese operations reveal an extensive support footprint. The company writes software to improve lithography in Shenzhen, builds chip measurement and inspection modules in Beijing, and trains customer engineers to troubleshoot and maintain DUV systems in Shanghai. China accounted for almost a third of ASML’s $37.8 billion in total 2025 revenue.

Although ASML’s support is directed at customers, better-supported Chinese fabs are optimal testing grounds for domestic replacements. Teaching fab engineers to fix alignment errors and improve yield on ASML tools helps them apply proven techniques to homegrown DUV equipment. Allied and U.S. servicing, including by Dutch firm ASML, can thus help local fabs make export-controlled 7 nm Huawei Ascend chips today while teaching China how to replace those tools tomorrow.

Current Controls Leave Servicing Gaps

Existing export restrictions block specific advanced chipmaking equipment, software, and technology shipments to China, and restrict certain U.S. citizens and companies from supporting Chinese chip development. But Dutch and Japanese vendors, along with nonrestricted Americans, can still service many Chinese DUV machines.

Congress has proposed a broad response. The MATCH Act would restrict equipment servicing at covered Chinese facilities, including those operated by SMIC, Hua Hong, CXMT, YMTC, Huawei, and Aishengna. It would also seek equivalent controls from allies such as the Netherlands and Japan, with U.S. restrictions extending to covered foreign-made equipment otherwise. The bill helps address the right problem but is moving too slowly.

The MATCH Act cleared the House Foreign Affairs Committee on April 22, and its sponsors seek to include it in the National Defense Authorization Act for Fiscal Year 2027. A July 14 Senate procedural vote failed to advance the defense bill to debate but did not reject MATCH itself. Senate Democrats were still pushing for floor action on MATCH and two other chip-related bills (the Chip Security Act and AI OVERWATCH Act) as of Oct. 13. Even if the bill passes, however, the introduced Senate text allows further delays of 150 days or longer. The Dutch and Japanese governments have opposed the measure’s extraterritorial reach.

Aishengna’s first deliveries are expected this year. The next few months therefore pose a critical window: Restricting servicing during Aishengna scanner qualification would make transitioning to mass production significantly harder. While continued DUV restrictions after Aishengna qualifies its machines may still somewhat disrupt Chinese chip production, they would be much less effective as Chinese toolmakers will have already learned what they need from allied and U.S. service providers.

Commerce Already Has a Lever

The Department of Commerce does not need new legislation to enforce General Prohibition 10 (GP10), which blocks export control-violating items from being bought, financed, transported, or serviced. The Bureau of Industry and Security (BIS), which administers U.S. export controls, already prohibits servicing items subject to the Export Administration Regulations (EAR) when the provider knows that a violation has occurred, is about to occur, or is intended to occur in connection with the item. The knowledge standard includes awareness of a high probability, which can be inferred from conscious disregard or willful avoidance of facts.

While later unauthorized transfers or license breaches can trigger GP10, servicing lawfully exported equipment pre-controls is not automatically banned. For foreign-made equipment, U.S.-content and foreign-direct-product rules determine jurisdiction. BIS may maintain jurisdiction over a component but not the whole scanner.

GP10 is a scalpel. But semiconductor tools are complex, and a scalpel can still cut deeply. Blocking service to an indispensable component or software can impair an entire production line. The commercial effects could extend further if credible enforcement makes providers unwilling to service high-risk fabs.

BIS already has a GP10 playbook. After Russia’s full-scale invasion of Ukraine began in 2022, the agency published aircraft tail and serial numbers, identifying apparent violations involving EAR-subject aircraft and warning that refueling or repair could trigger GP10. In May 2025, BIS similarly identified Huawei’s Ascend 910B, 910C, and 910D chips as presumptively subject to GP10. Such public warnings help establish knowledge by putting companies on notice.

Furthermore, relevant semiconductor violations have already happened. In its 2025 Cadence settlement, BIS imposed a $95 million penalty for unauthorized exports and related conduct. The settlement included GP10-related charges of Cadence transferring to China export-restricted chip design software and technology, and attempting another hardware transfer. In February 2026, BIS also fined Applied Materials and its Korean subsidiary $252.5 million over 54 unauthorized reexports and two attempted reexports involving U.S.-origin chipmaking equipment destined for SMIC.

These cases offer BIS a starting point for a program identifying semiconductor equipment modeled on its aircraft notices. Investigators should trace equipment linked to those violations and enforce bans on associated servicing.

Inventory Chipmaking Tools

Investigators should map critical U.S.-origin and other EAR-covered chipmaking tools, components, and software by serial number, export history, license status, service record, and end user. They can begin with U.S. supplier records, licensing files, and prior enforcement cases. BIS’s record production authority provides a basis for obtaining relevant records, including mandatory subpoenas to persons in the United States.

Each documented violation should prompt BIS to scrutinize related tools, transfers, and service contracts at the same fab. Following those connections would reveal the intermediaries, suppliers, and customers behind GP10 violations. The inventory would also improve Washington’s understanding of semiconductor supply chains, simultaneously helping further proposed Chip Security Act and AI OVERWATCH Act priorities. After all, BIS would uncover which production lines depend on foreign chokepoints, where diversion occurs, and where further controls would have maximum impact.

Washington should ask the Dutch and Japanese to conduct parallel inventories of suspected export control-violating equipment. Supporting a targeted criminal investigation avoids forcing either government to prohibit all equipment servicing or accept expanded U.S. MATCH Act jurisdiction.

Even if those governments decline, BIS already has the authority to pursue providers servicing GP10-subject items. Coverage may be incomplete without foreign records and enforcement assistance, but BIS can still expose violations and put service providers on notice.

BIS could, for example, investigate if EAR-covered light sources in ASML scanners or EAR-covered components in Aishengna’s machines were shipped without licenses or illegally diverted. Documenting such violations and providing providers notice could deter ASML from servicing at-risk fabs and Japanese suppliers from supplying Aishengna’s equipment. These allied measures would significantly inhibit China’s domestic chipmaking and DUV development.

Put the Industry on Notice

BIS should send GP10 warnings to chipmaking equipment manufacturers, component suppliers, and service providers worldwide. It should identify covered equipment and associated fabs, explain the underlying violations, and specify which servicing activities are prohibited. A separate section should flag facilities potentially involved in unauthorized transfers, license breaches, or prohibited military end uses.

This approach builds on BIS’s existing compliance guidance. The guidance identifies uncertain license histories as red flags for service providers. Without further information, BIS suggests firms refrain from the transaction or first consult the agency. Targeted notices to executives and compliance teams would make those risks difficult to ignore.

Warnings also require credible consequences. BIS should consider maximum financial and export denial penalties for egregious GP10-related servicing violations. Its penalty guidelines already consider prior notice, willful or reckless conduct, and deterring other companies.

These efforts would change the risk-reward calculation for allied and U.S. suppliers, such as ASML. Potential investigations coupled with financial and export sanctions make an initially profitable service contract less attractive. A supplier may decide to suspend at-risk fab relationships rather than screen every individual service call. A few well-supported GP10 cases could therefore create a de facto servicing ban without BIS even needing to identify every affected machine.

A Bridge to Broader Servicing Restrictions

GP10 enforcement thus complements new rulemaking. With enough documented violations and direct warnings, it can deter firms from servicing Chinese fabs even before broader de jure restrictions. Commerce can pursue new rules under its existing authority, and the MATCH Act could provide statutory coverage.

The first targets should be items necessary to chip production whose servicing China cannot easily replace, such as light sources of U.S. origin within ASML’s DUV machines. BIS should measure the program’s effects through service contract changes, supplier withdrawals, and replacement availability. Such restrictions would not freeze Chinese chip production. Fabs could rely on local engineers, stockpiled or cannibalized parts, and third-party repair. But sustaining mass production would become harder and more expensive without support from the most experienced foreign vendors.

Restrictions might even hasten domestic substitution. They could also disrupt allied and U.S. supplier revenue alongside lawful civilian production. Thus, BIS should issue precise notices and quickly correct mistaken identifications to limit unnecessary disruption. But substitution amid outages, uncertain benchmarks, and engineering disruption would be very costly. A teardown can reveal hardware but cannot easily reproduce the whole system. Export controls cannot erase China’s ambition. They can increase its cost, risk, and development time.

Removing the Scaffolding

Aishengna’s first fab-integrated scanner may grab headlines, but its transition to mass production deserves closer attention. Continued foreign servicing makes that process easier. Once production-ready, domestic scanners would help China accelerate its AI and military chipmaking.

BIS need not wait for new legislation. It can map equipment and violations, circulate warnings across the chipmaking industry, and penalize providers that knowingly continue prohibited servicing. Each investigation strengthens the next by uncovering supply chain relationships and clarifying firm exposure. Leveraging existing targeted authorities also avoids the diplomatic fallout of forcing new U.S. rules on the Netherlands and Japan.

Washington cannot stop China from trying to replace foreign equipment. It can, however, use documented GP10 violations to stop allied and U.S. firms from illegally subsidizing that transition. BIS could target that support by mapping equipment and warning suppliers while China is still qualifying its scanners for mass production, buying time for Congress to consider broader restrictions. Foreign servicing scaffolds China’s homegrown scanners. Washington should not wait to remove that scaffolding.