Chinese companies are working hard to catch up to US, European, and Japanese suppliers of semiconductor manufacturing equipment (SME), which is crucial to making AI chips. US and allied chip manufacturers are starting to use Chinese equipment and will likely use more of it as it improves. For example, Intel has reportedly tested tools from one Chinese equipment maker for its most advanced upcoming 14A process, while another Chinese company claims to have sold etching tools to TSMC, likely for TSMC’s advanced 3 nm process.
This use of Chinese equipment in advanced chip fabrication plants (“fabs”) likely speeds China’s indigenization of these machines. It helps Chinese equipment makers develop and improve their products much faster by giving them access to the fabs’ technical expertise and feedback. It could also become a significant revenue source for them and create a cybersecurity risk for US and allied fabs.
We recommend that the US and its allies adopt measures to stop the flow of knowledge and revenue from US and allied fabs to Chinese equipment makers. These should target only equipment used in advanced-node1 processes. We propose eight measures:
Stop knowledge transfer and revenue to Chinese equipment makers:
The Bureau of Industry and Security (BIS) should create a Part 744 license requirement for transfers of information and software that support Chinese SME suppliers.
The President should issue an International Emergency Economic Powers Act (IEEPA) executive order authorizing Treasury to prohibit engineering, R&D, and technical collaboration with designated Chinese equipment makers.
BIS should condition exports of controlled US equipment to foreign fabs on their agreement not to use Chinese equipment or collaborate with Chinese equipment makers.
BIS should send letters to Intel, Micron, TSMC, Samsung, and SK Hynix explaining existing duties, requesting information, and seeking an immediate pause on new Chinese equipment purchases and collaboration.
Congress should pass the Chip EQUIP Act, which would prohibit the purchase of Chinese equipment for projects funded by the CHIPS Act.
Prevent acquisitions that transfer sensitive equipment capabilities:
Treasury should make Chinese acquisitions of US equipment companies a Committee on Foreign Investment in the United States (CFIUS) enforcement priority.
The State Department should work with Commerce and Treasury to secure allied commitments to restrict purchases of Chinese equipment and Chinese acquisitions of allied equipment makers.
Require reporting on adoption of Chinese equipment and ongoing collaboration:
Commerce should direct BIS to establish mandatory confidential reporting on the use of Chinese equipment and collaboration with Chinese equipment makers, through Defense Production Act reporting authorities.
The rest of this post goes into more detail on how Chinese equipment is entering leading fabs, why the US should act, and what it should do.
Chinese equipment has been making slow inroads into leading fabs outside China
Chinese equipment exports are currently small, with only a few high-profile cases of leading fabs evaluating or buying Chinese tools, most notably Intel’s testing of ACM Research tools and TSMC’s purchases from AMEC. Most other Chinese equipment sales to non-Chinese companies go to their fabs in mainland China, such as TSMC’s or SK Hynix’s, or come from US and allied equipment firms acquired by Chinese firms. (A full list of Chinese equipment exports is in the appendix.)
Export reporting is patchy, partly because the amounts are small. But across the largest equipment companies in China, total exports in 2025 were likely a couple hundred million dollars.2 That is a tiny sliver of the $135 billion of semiconductor equipment that the industry association SEMI estimates was sold worldwide that year.
China’s equipment makers are growing rapidly in the domestic market, supplying a large build-out at leading Chinese fabs such as YMTC, SMIC, and CXMT, as well as at a long tail of smaller, less advanced producers. Companies such as AMEC, NAURA, and SiCarrier are building broad equipment portfolios and operating at scale, with extensive investment in R&D. They aim to become real challengers to incumbents like Applied Materials or Tokyo Electron.
To date, however, China’s equipment makers have lacked the capability and scale to compete head-to-head with global giants. They have been incubated in a protected domestic market, benefiting from large-scale state capital and pressure on Chinese fabs to adopt more domestic equipment. So their exports have been low.
A more common pattern than exports has been sales to foreign-owned fabs in mainland China, such as SK Hynix’s fab in Wuxi or TSMC’s in Nanjing. Selling to a leading global chip manufacturer like TSMC, even in China, is a strong mark of approval and an opportunity for technical learning.
For example, supplying SK Hynix’s Wuxi fab was a breakthrough for ACM Research.3 Although ACM Research is nominally US-based, it consists almost entirely of its subsidiary, ACM Shanghai. Building on SK Hynix’s early orders, the company developed its technology and expanded its product portfolio. It now increasingly targets overseas markets.
ACM Research’s exports remain small: Almost all of its roughly $900 million in 2025 revenue came from mainland China, with just 0.4% from outside China. But one case stands out. In late 2025, Intel reportedly tested ACM Research equipment for its leading 14A node, prompting public concerns from US lawmakers. Intel denied that the machines were being used in production, though it did not deny that they had been evaluated.
Whatever happened with Intel, ACM Research is set on expanding overseas. It aims to send more than 20 machines abroad by the end of 2026. More ambitiously, it wants foreign customers to account for $1.5 billion of its $4 billion long-term revenue goal, a far larger share than the current 0.4%.
Many other sales of Chinese equipment to US and allied chip manufacturers have gone to their fabs in mainland China. AMEC has reportedly sold equipment to South Korean NAND memory fabs, likely both in mainland China and in South Korea. In August 2026, SK Hynix and Samsung reportedly considered buying AMEC equipment for their mainland fabs, though both denied testing it.
Actual exports of Chinese equipment (sales to leading global companies’ fabs outside mainland China) have been less common. KINGSEMI has claimed TSMC as a customer for its less sophisticated back-end advanced packaging tools.
For more complex front-end tools, the most notable case is AMEC. In a May 2026 interview, AMEC’s chairman and CEO, Yin Zhiyao, claimed that the company had sold about 130 machines to Taiwan4 and that its equipment was used for processes from 45 nm down to 3 nm. Given that TSMC has the only 3 nm process in Taiwan, AMEC was presumably selling to TSMC. This fits with reporting in August 2025 that TSMC moved to exclude Chinese suppliers from its 2 nm node, which suggests it had used them in earlier nodes. If so, it’s a major win for AMEC, which has no better way to improve its etching equipment than to have it running in some of TSMC’s most advanced fabs.
A final category is Chinese equipment firms acquiring smaller US, Japanese, or South Korean ones. This drives Chinese SME exports either materially, when technology and production are transferred into China while the subsidiary keeps its overseas customers, or notionally, when the overseas subsidiary remains largely untouched but the Chinese firm books more overseas revenue.
The most prominent example to date is the 2016 acquisition of Mattson Technology by a Chinese firm called Yitang, or E-Town. Besides a messy tit-for-tat legal dispute with Applied Materials over intellectual property, Yitang inherited Mattson’s installed base and its US and overseas operations. With over $200 million in overseas revenue in 2025, Yitang is, on paper, the largest Chinese equipment exporter. However, this largely reflects an existing overseas business rather than the export of Chinese-developed technology. Still, Mattson’s Chinese ownership almost certainly means its capabilities will become increasingly indigenized.
Why this is a problem
Current export volumes are small, but they will grow as Chinese equipment improves and its ecosystem matures. One key variable in how fast and far it improves is whether its makers can work with and learn from the world’s leading fabs. Denying Chinese equipment makers access to the most advanced processes would be a material setback. But it would also entail real costs: potentially strained US-China relations, reduced competition in the equipment market, and the loss of one of analysts’ best signals of how good Chinese equipment is.
The semiconductor industry is organized around technology roadmaps. Research organizations such as IMEC work with leading chip manufacturers and equipment makers to scope out what the next generation of technology requires. Fabs such as TSMC then turn to their supply chains for the right equipment and materials to make it commercially viable. More broadly, a global network of industry R&D centers, universities, and national labs helps chart the technical frontier.
Being part of this global ecosystem is a major advantage for a Chinese equipment firm, since its domestic customers are far from the frontier. Because of export controls on key equipment, especially advanced lithography machines, Chinese chip manufacturers such as SMIC and CXMT remain multiple generations behind global leaders like TSMC and SK Hynix. SMIC’s most advanced node, N+3, is comparable to TSMC’s 6 nm node from 2020, while TSMC has already moved to 2 nm. CXMT is struggling to produce viable high-bandwidth memory that is two generations behind SK Hynix’s current product.
If a Chinese equipment maker like AMEC sells only to domestic chip manufacturers, its R&D progress and equipment capabilities will be limited by its lack of access to cutting-edge processes. It can still improve, but it would learn much faster if it could work in a process that uses leading equipment from other suppliers and the most advanced techniques.
Learning from leading fabs is one reason Chinese equipment companies are seeking to break into global markets. The other is that the market outside China is still larger than China’s. To match global incumbents in R&D spending and scale, Chinese firms need to capture a share of that market. In the words of AMEC’s Yin Zhiyao:
When we try to enter foreign markets, we often face regional barriers, cultural barriers, and high entry thresholds for new suppliers. In recent years, the international situation has posed even greater challenges for us to break into international markets. In this context, we can’t give up our plans to expand abroad. Because no matter how big our domestic market is, the majority of the market is still overseas in the short term, and that hasn’t changed. We still need to keep pushing forward.
Breaking into global supply chains means passing evaluation and qualification at leading fabs. This process is slow and costly: It can take 6 to 24 months from first contact with a customer to a sale. Equipment makers front the cost of the tool, and the fab must allocate valuable cleanroom space and technician time to put the machine through its paces. Advanced chipmaking processes require equipment that meets throughput and reliability specs without unduly interfering with other fab equipment or causing defects that hurt yield.
This complexity forces fabs and vendors into an uncomfortably close relationship, in which they must constantly share information to tune and integrate the equipment. A fab has to give up enough information to integrate the tool, but not too much, because equipment firms also sell to its competitors. Even a failed evaluation can teach an equipment maker a great deal, giving it a stronger grasp of the required technical benchmarks and the practical problems the fab faces at that node.
Collaboration between fabs and equipment makers drives technical progress. ASML and TSMC have one of the closest partnerships: TSMC’s division director for extreme ultraviolet lithography described the two as “one team” during the technology’s development. If Chinese equipment firms could work with and learn from leading fabs, their development would meaningfully accelerate.
Other factors favor restricting Chinese SME exports:
Exports will likely become a larger source of revenue for Chinese SME firms over time. Currently, these firms are unlikely to be short of capital, given heavy state support for China’s semiconductor industry. But being confined to the domestic market would limit their long-term growth and therefore the capital they could access.
The scale of state capital supporting Chinese equipment firms could generate overcapacity. Leaders of Chinese SME firms worry that “involution”, or cutthroat competition, is creating irrational pricing pressure, as in China’s solar and battery industries. Artificially cheap but capable Chinese equipment could unfairly undercut global equipment makers that don’t receive similar state support.
Chinese equipment in advanced fabs could pose a cybersecurity risk. Cybersecurity controls can limit unauthorized access to neighboring fab systems, but they do not stop vendors from learning during authorized installation and troubleshooting. Legitimate diagnostic data could still give Chinese equipment firms insight into fabs with export-controlled US and allied equipment and into AI chip production.
Acting preemptively would cost far less than restricting later. Once Chinese tools are widely used and integral to supply chains, they become much harder to remove. Retroactive rip-outs or stranded fleets of equipment are far more harmful to industry than earlier action. Fabs already seem to weigh geopolitical risks when deciding whether to buy Chinese SME, but this cost-benefit calculation is fragile and will likely break down as Chinese equipment becomes more capable.
However, prohibiting Chinese equipment in US and allied fabs carries risks:
Given the current uneasy US-China truce, China might seize on the restrictions as grounds for retaliation. This could escalate into something like the trade war blow-up in 2025. China could also retaliate against allied fabs for complying. Restricting Chinese SME exports is almost certainly not the most important step for the US government to take in US-China AI competition; tightening export controls would have far greater impact on the margin. With a limited budget for actions that harm the relationship with China, closing loopholes on cloud rental of chips or tightening controls on semiconductor manufacturing equipment should be a higher priority.
More capable Chinese equipment firms would increase competition in the equipment market. New players with talented teams and innovative ideas could shake up a partly oligopolistic market, in which fabs can choose from only one or two vendors for some equipment categories. More competition means cheaper, better equipment for fabs, which benefits downstream users of AI servers and consumer chips.
Chinese vendors currently offer no capability in advanced front-end equipment that US or allied vendors lack, so the cost of not using Chinese SMEs is low. But a Chinese equipment firm will quite likely produce a tool with unique capabilities in the next two to three years. Being unable to buy such a tool could put US and allied fabs at a disadvantage compared with Chinese ones. This risk is real, but its impact is deeply uncertain.
The quality of Chinese equipment is hard to assess. Because Chinese fabs favor domestic suppliers, their purchases do not give a clear signal of how competitive a Chinese tool is. Purchases by leading global fabs, by contrast, provide a strong signal of Chinese equipment’s capabilities. This information helps the US government assess the effectiveness of existing export controls on equipment and, more generally, forecast how China’s AI chip ecosystem will develop. Restricting Chinese equipment exports would cut off this signal.
The US should restrict knowledge transfer and revenue
No single US policy can address the risks from Chinese equipment in leading fabs, but complementary policies can. Together, they should:
Restrict knowledge transfer from US and allied fabs to Chinese equipment companies and/or limit their revenue
Go beyond existing restrictions
Reach fabs in allied countries, not just US fabs
Be fast and cheap to implement, ideally relying on existing authorities
Be clear and targeted
Preserve Western manufacturing, innovation, and competition as far as possible
Where these goals conflict, we’ve tried to make reasonable compromises. These policies should affect only equipment used in advanced-node fabs and advanced packaging facilities used to make high-bandwidth memory, and should grandfather in previously imported equipment so that existing production is not disrupted.5 This would mean that mature-node fabs, which are essentially irrelevant to AI chipmaking, can still use Chinese equipment.
Though some of these measures apply only to US companies, or only to fabs in the US regardless of who runs them, we think they will still affect fabs in allied countries. Chip manufacturers often try to keep equipment and processes as similar as possible across fabs. For example, TSMC’s Arizona fab closely mirrors its fabs in Taiwan. So the Arizona fab can benefit from process improvements made in Taiwan, and vice versa, which would not be possible if the fabs used different equipment. If fabs in the US are prohibited from using Chinese SME, chip manufacturers will likely avoid that equipment in all their fabs running the same processes, regardless of location. (This spillover doesn’t reach companies with no advanced fabs in the US, like SK Hynix.)
To stop leading fabs from helping develop Chinese SME, we propose eight measures:
Recommendation 1: BIS should create a Part 744 license requirement for transfers of information and software that support Chinese SME suppliers. Part 744 of the Export Administration Regulations (EAR) contains BIS’s end-use and end-user controls. These controls are triggered by who receives an item or what it is for, so they can cover any items “subject to the EAR”6, including those not on the Commerce Control List. BIS could require a license under Part 744 for transferring to Chinese equipment makers and their employees any information and software related to the evaluation, qualification, or development of Chinese SME at advanced-node fabs.
Recommendation 2: The President should issue an IEEPA executive order authorizing Treasury to prohibit engineering, R&D, and technical collaboration with designated Chinese equipment makers. After declaring a national emergency, the President can use IEEPA to prevent persons subject to US jurisdiction from engaging in transactions in which China has an interest. Treasury usually administers the resulting program. Executive Order 14105, issued in August 2023, and its implementing Treasury rule already forbid US investment in Chinese companies making SME for advanced-node fabrication. A new order could extend that ban to nonfinancial support. This approach covers only US persons, but these include employees of fabs like TSMC Arizona.
Recommendation 3: BIS should condition exports of controlled US equipment to foreign fabs on their agreement not to use Chinese equipment or collaborate with Chinese equipment makers. BIS already requires licenses for exports of much US-made SME to China and a few other countries. It could extend this license requirement globally and add a license exception for the newly covered countries, similar to a recommendation in a 2025 report by the House Select Committee on the Chinese Communist Party.7 This exception would be conditional on the importing fab not using Chinese equipment or collaborating with Chinese equipment makers. Using Chinese equipment in US and allied fabs may require giving technicians from Chinese companies access to the fab and connecting the Chinese equipment to the same networks as US equipment. Both could create risks of intellectual property theft. Assuming already-installed equipment is grandfathered in, these conditions would be easy to meet today and wouldn’t significantly disadvantage US equipment makers.
Recommendation 4: BIS should send letters to Intel, Micron, TSMC, Samsung, and SK Hynix explaining existing duties8, requesting information, and seeking an immediate pause on new Chinese equipment purchases and collaboration. These letters would not be binding. But the companies do not have strong vested interests in Chinese equipment, so they might comply regardless, especially since BIS can follow up with rules if they do not. Companies have made such voluntary commitments before. For example, in October 2022, ten days after BIS added YMTC to its Unverified List and amid congressional pressure, Apple froze plans to use YMTC as a memory supplier. Sending letters is fast and cheap and requires no new rule.
Recommendation 5: Congress should pass the Chip EQUIP Act, which would prohibit the purchase of Chinese equipment for projects funded by the CHIPS Act. If the bill became law, recipients of CHIPS Act funding could not buy, install, or use equipment made by Chinese companies or their subsidiaries (including US ones like Mattson) at subsidized projects for ten years, unless granted a waiver.9 While this wouldn’t cover purchases of Chinese equipment in Taiwan or South Korea, it would likely discourage them because chip manufacturers try to use the same equipment across their fabs.
Recommendation 6: Treasury should make Chinese acquisitions of US equipment companies a CFIUS enforcement priority. CFIUS, an interagency committee chaired by the Treasury Secretary under the Defense Production Act, reviews foreign acquisitions of, and some investments in, US businesses. A Chinese acquisition of or investment in a US equipment maker would already require filing with CFIUS, but the committee should not approve such deals. In hindsight, allowing a Chinese company to acquire Mattson was a mistake that should not be repeated.
Recommendation 7: The State Department should work with Commerce and Treasury to secure allied commitments to restrict purchases of Chinese equipment and Chinese acquisitions of allied equipment makers. Just as Chinese equipment makers could speed indigenization by acquiring or investing in US equipment makers, they could do the same with allied companies. The State Department should encourage allies to take similar steps domestically, particularly Japan, the Netherlands, Germany, South Korea, Taiwan, Switzerland, and Austria.10
Recommendation 8: Commerce should direct BIS to establish mandatory confidential reporting on the use of Chinese equipment and collaboration with Chinese equipment makers, through Defense Production Act reporting authorities. Currently, nothing requires US fabs to report Chinese equipment that’s installed, under evaluation, or being co-developed. For example, Intel’s testing of ACM Research equipment became public only through a news report. BIS could use Section 705 of the Defense Production Act to survey US companies on their use or planned use of Chinese equipment in advanced-node fabs, similar to the 2024 survey on Chinese chip use by US companies under the same authority. A new survey would likely cover all US companies and fabs but may not reach fabs in Taiwan and South Korea. Section 705 expires on December 11, 2026, unless Congress passes a reauthorization bill.
Given how much capital and engineering talent China has directed towards its semiconductor equipment companies, we should expect more capable and competitive Chinese equipment to emerge. A key goal for those firms will be to tap the scale of global markets and the advanced technology of leading chip manufacturers such as TSMC, Intel, and Samsung. The US and its allies can act now to cut off those potential relationships, along with the learning and acceleration they would bring to China’s efforts to displace US and allied equipment makers.
Appendix: Known cases of exports of Chinese SME
BIS defines “advanced-node” chips as (a) logic chips produced at 16/14 nm nodes or below, (b) NAND memory chips with 128 layers or more, and (c) certain DRAM chips. When we refer to advanced-node production, we also include the advanced packaging step that’s essential to making modern AI chips with high-bandwidth memory.
Published financials give fiscal year 2025 export figures for NAURA (~$70 million), AMEC (~$45 million), KINGSEMI (~$20 million), SkyVerse (~$8 million), and Hwatsing (~$6 million). The other largest firms are SMEE, as well as SiCarrier and other Huawei-linked equipment makers. All are unlisted, so we cannot ascertain their exports. But given their strong domestic focus, these are likely very small or zero, as with Piotech, which reports no overseas revenue for 2025. Yitang also made around $210 million overseas in 2025, but given its acquisition of Mattson, this does not reflect the export of Chinese-developed technology; instead, it shows ownership of a preexisting revenue stream.
SK Hynix was the first firm to evaluate and then purchase ACM Research’s 300 mm SAPS megasonic cleaning tool and is the first named customer in the company’s corporate history. SemiAnalysis describes ACM Research as having “first struck gold” with these sales to SK Hynix Wuxi.
Strictly, he claimed more than 800 reaction stations, the modules in an etching machine. AMEC’s etching tools usually have six, so this implies roughly 130 machines or more.
The Chinese equipment makers could still maintain and repair these tools, though they could not upgrade or move them, or access data about the fab’s process beyond what is needed for repairs and maintenance. Nor could they conduct joint research with the fab.
Items “subject to the EAR” are items in the US, items originating in the US, items made outside the US but containing more than a de minimis amount of controlled US-origin content, and items produced outside the US but using certain US equipment (covered by an applicable foreign direct product rule).
That report’s recommendation is somewhat different (emphasis ours): “BIS should act now to restrict the use of PRC SME outside of the PRC. First, BIS should require, as a condition for export of US SME, that PRC SME cannot be used in the same fab anywhere in the world. [...] Second, BIS should initiate an investigation of imports of PRC SME into the United States, via its Information and Communications Technology and Services (ICTS) authorities. Third, BIS should focus its Section 232 investigation on PRC SME while exempting allied SME.”
In particular, that they need a license to share technology subject to the EAR with companies on the Entity List (such as ACM Research’s Shanghai office), and their obligations related to CHIPS Act awards.
A potential disadvantage of the Chip EQUIP Act is that it isn’t limited to equipment for advanced-node fabs. We think the broader scope is acceptable since these are subsidized projects, but Congress could also amend the bill to target advanced-node fabs only.
These countries matter for different reasons:
Japan has Tokyo Electron, Screen, Advantest, Kokusai, and Lasertec, along with hundreds of smaller toolmakers and component suppliers such as HORIBA and Fujikin.
The Netherlands has the supplier network around ASML, including VDL ETG and Prodrive Technologies, as well as equipment makers like ASM International and Besi.
Germany has Zeiss, Trumpf, SUSS MicroTec, PVA TePla, and Pfeiffer Vacuum, among others.
South Korea has Samsung and SK Hynix (both potential purchasers of Chinese equipment) and equipment makers such as SEMES, Wonik IPS, PSK, Jusung Engineering, and Hanmi Semiconductor.
Taiwan has TSMC, the packaging firm ASE, and equipment makers and component suppliers like Gudeng and Grand Process Technology.
Switzerland has VAT, Comet, and INFICON.
Austria has EV Group and IMS Nanofabrication.