In early 2025, inside a high-security laboratory in Shenzhen, a machine the size of several factory floors successfully generated extreme ultraviolet light for the first time. It was, by every account, a working prototype of an EUV lithography system — the same class of machine that only one company on Earth, ASML in the Netherlands, has ever commercialized. It was built by former ASML engineers working under aliases and false identification cards, as part of what insiders describe as China's semiconductor "Manhattan Project."
It has not produced a single working chip.
That single sentence contains the entire story. The prototype exists. It generates EUV light. It is undergoing testing. And the distance between "operational prototype" and "machine that can manufacture competitive chips at scale" is measured not in months but in years — possibly a decade. Here is what the prototype actually means, what it cost to build, and why the trajectory matters more than the headline.
What Reuters Actually Reported
The detailed picture comes from a December 2025 Reuters exclusive by Fanny Potkin, drawing on multiple sources with direct knowledge of the project. The key facts:
- A prototype was completed in early 2025 in Shenzhen by a team that included former ASML engineers who reverse-engineered the company's EUV systems.
- The machine is operational and generating extreme ultraviolet light, but has not yet etched functioning circuits onto silicon wafers.
- Huawei coordinates the supply chain, from chip design and fabrication equipment through manufacturing and final product integration, though the project is government-run under President Xi Jinping confidant Ding Xuexiang.
- The government's internal target is 2028 for producing working chips on the prototype, but people close to the project say 2030 is more realistic.
- Recruits were given false names and ID cards. One veteran Chinese engineer from ASML was surprised to find, upon arrival, that colleagues recognized him but were using aliases themselves.
- About 100 recent university graduates are focused on reverse-engineering components, with individual desk cameras documenting their disassembly and reassembly work.
ASML confirmed to Reuters that no EUV system has ever been sold to a customer in China. Everything in that Shenzhen lab was built from scratch, salvaged parts, and reverse-engineered knowledge.
The Prototype vs. Production Gap
ASML's own timeline is the most instructive comparison. The Dutch company built its first working EUV prototype in 2001. Its first commercially-available chips using EUV came in 2019. That is 18 years and billions of euros of R&D between "proof of concept" and "production tool."
China has some advantages ASML did not. EUV is no longer theoretical — the basic physics and engineering approaches are documented in patents, academic papers, and ASML's publicly described system architecture. China is not inventing from zero. It also has the benefit of state-directed resource mobilization at a scale that even ASML, with its 5,000-plus supplier network, cannot match for sheer intensity of effort.
But the disadvantages are equally real, and they cluster around precision manufacturing at the bleeding edge.
The Optics Problem
The single biggest bottleneck is the optical system. ASML's EUV machines use mirrors made by Germany's Carl Zeiss AG that take months to produce and require surface precision measured in picometers — trillionths of a meter. These are among the most precisely manufactured objects ever created.
China's prototype uses optics developed by the Changchun Institute of Optics, Fine Mechanics and Physics (CIOMP), which achieved a breakthrough in integrating EUV light into the optical system, enabling the machine to become operational. But the sources describe the optics as still requiring "significant refinement" — which in semiconductor engineering terms means they are not yet close to what Zeiss produces.
This is not a problem that money alone can solve. Zeiss spent decades developing its mirror manufacturing capability. The precision is cumulative — built through generations of instrument makers, process engineers, and measurement scientists working in lockstep. You can recruit former ASML engineers who understand the specifications, but building the industrial ecosystem to actually produce optics that meet those specifications is a different challenge entirely.
The Light Source Challenge
EUV lithography works by firing a CO2 laser at molten tin droplets 50,000 times per second, generating plasma at roughly 200,000 degrees Celsius. The resulting extreme ultraviolet light is then focused by mirrors onto the silicon wafer.
For production, the light source needs to deliver sufficient power reliably, with minimal contamination, for thousands of hours of operation. SemiAnalysis analyst Jeff Koch, himself a former ASML engineer, framed the benchmark clearly: China needs to demonstrate that its "light source has enough power, is reliable, and doesn't generate too much contamination."
The prototype has generated EUV light. Generating it with the stability, power output, and cleanliness required for high-volume manufacturing is another matter entirely.
Scale and Integration
ASML's most advanced EUV systems are roughly the size of a school bus and weigh 180 tons, with 100,000 components from 5,000 suppliers. The three Chinese semiconductor executives who published their March 2026 analysis in Science and Technology Daily made the scale vividly clear: ASML's EUV equipment has 100,000 components from 5,000 suppliers, with ASML "merely serving as the integrator." This is not a single-invention problem. It is an industrial ecosystem problem.
China's prototype is "many times larger" than an ASML system — because the team initially failed to replicate ASML's compact form factor and compensated by increasing the machine's physical size to improve power output. This is a telling detail. In semiconductor equipment, size is not neutral. A larger machine is harder to install, harder to maintain, harder to replicate across multiple fabs, and harder to iterate on quickly. ASML's compact form factor was not an aesthetic choice — it was the result of years of engineering optimization that reduced the machine's footprint while increasing precision and throughput.
A machine that fills an entire factory floor is an impressive engineering achievement. A machine that chipmakers can install in a fab, operate cost-effectively, and maintain at high uptime is a different product altogether. The miniaturization and integration challenge is not trivial — it is, in many ways, the core engineering challenge. Every component must not only work in isolation but work in synchronization with every other component, under conditions of extreme precision, for thousands of hours of continuous operation.
How China Got Here: The Recruitment Machine
The human pipeline behind this project deserves attention on its own. China launched an aggressive recruitment drive for semiconductor experts in 2019, offering signing bonuses of 3 million to 5 million yuan ($420,000 to $700,000) plus housing subsidies. The targets were specifically Chinese-born engineers and scientists working at Western semiconductor companies — particularly ASML.
One notable recruit was Lin Nan, ASML's former head of light source technology, whose team at the Chinese Academy of Sciences' Shanghai Institute of Optics filed eight patents on EUV light sources within 18 months. CIOMP's March 2026 recruitment call advertised "uncapped" salaries for PhD lithography researchers, plus research grants up to 4 million yuan ($560,000) and personal subsidies of 1 million yuan ($140,000).
The secrecy measures suggest China is aware of the legal and diplomatic exposure. Dutch intelligence warned in an April 2025 report that China "used extensive espionage programmes in its attempts to obtain advanced technology." ASML won an $845 million judgment in 2019 against a former Chinese engineer accused of trade secrets theft, though the defendant filed for bankruptcy and continues operating in Beijing with government support.
European privacy laws limit ASML's ability to track former employees. Non-disclosure agreements are difficult to enforce across borders. These are structural vulnerabilities in the Western semiconductor ecosystem that no single policy change can fully address.
The Parts Pipeline: Salvage and Sourcing
Building the prototype required a supply chain that exists partly in the shadows. China is salvaging components from older ASML machines and sourcing parts from ASML suppliers through secondary markets. International banks regularly auction older semiconductor equipment, and as recently as October 2025, listings on Alibaba's auction platform showed older ASML lithography equipment for sale in China.
Export-restricted components from Japan's Nikon and Canon are also being used in the prototype, according to sources. Networks of intermediary companies sometimes mask the ultimate buyer.
This salvage-and-sourcing approach is both clever and fundamentally limiting. It works for building a prototype — you need one of each critical component, and older parts can demonstrate the basic physics. But high-volume manufacturing requires reliable, consistent supply of thousands of identical components meeting precise specifications. You cannot scale a production line on salvaged parts.
The secondary market is also finite. Every older ASML machine that China acquires for teardown is one fewer machine available for future prototypes. And as export controls tighten — the Dutch government is developing policies requiring "knowledge institutions" to screen personnel for sensitive technology access — the pipeline of available parts and expertise may narrow further over time.
What the Industry Is Saying
In March 2026, three of China's most senior semiconductor executives — Zhao Jinrong (chairman of Naura Technology), Chen Nanxiang (chairman of Yangtze Memory Technologies), and Liu Weiping (chairman of Empyrean Technology) — published a joint article in Science and Technology Daily that was remarkably candid about the challenges ahead.
They acknowledged that while different Chinese institutions have made "breakthrough progress" in individual EUV subsystems — laser light sources, wafer stages, and optical systems — integrating these components into a complete system remains the key challenge for the 15th Five-Year Plan period (2026-2030).
Their proposed solution: a coordinated national effort to create "China's ASML," with unified allocation of funds and human resources across institutions. The framing itself reveals the gap. When your industry leaders are publicly calling for the creation of something that already exists in prototype form, you are hearing an implicit admission that the prototype alone is not enough.
The executives also identified bottlenecks in electronic design automation (EDA) software and basic materials — silicon wafers, electronic gases — as areas requiring national-level coordination. This is a broader point worth understanding: lithography is the most visible bottleneck in China's semiconductor push, but it is far from the only one. The supply chain for advanced chipmaking involves hundreds of specialized inputs, and China faces constraints across many of them. The focus on EUV is understandable given its symbolic and practical importance, but the actual path to semiconductor self-sufficiency runs through dozens of equally challenging materials and equipment challenges.
China's 2026 government work report, released the same week, designated semiconductors as a core pillar alongside aviation, biotechnology, and the low-altitude economy, but notably made no specific mention of lithography machines — calling instead to "improve advanced process manufacturing capabilities."
Why This Still Matters
It would be easy to dismiss the prototype as a curiosity — an impressive science project that changes nothing in the near term. That would be wrong, for three reasons.
First, it demonstrates intent backed by resources. This was a six-year classified project involving thousands of engineers, operating at a level of secrecy that included false identities for senior staff. China is not dabbling. It has committed to a sustained, well-funded effort to achieve EUV capability, and it has shown it can attract the talent needed to make technical progress.
Second, it compresses the timeline, even if it does not eliminate the gap. ASML CEO Christophe Fouquet said in April 2025 that China would need "many, many years" to develop EUV technology. The prototype suggests the timeline is shorter than he assessed — not because China is moving faster than expected on the engineering, but because it started earlier and invested more heavily than outside observers realized.
Third, it underscores the limits of export controls as a long-term strategy. Restrictions bought time — probably several years. But they also created a powerful incentive for China to develop indigenous capability, and the talent and resource mobilization that resulted may ultimately prove more consequential than the delays imposed. As we noted in our analysis of smic-record-revenue-despite-sanctions, sanctions often redirect rather than restrict. They created a captive domestic market for Chinese chipmakers, and now they have accelerated the push for domestic equipment.
For readers trying to understand the broader semiconductor landscape, our china-semiconductor-industry-guide covers the full picture of Chinese chipmaking capabilities, while us-export-controls-china-chips explains how the restrictions work and where the gaps are. The question of whether China can produce advanced chips without EUV is addressed in china-5nm-chip-reality, and the equipment ecosystem that supports Chinese fabs is covered in china-semiconductor-equipment.
The Bottom Line
China's EUV prototype is real. It represents genuine technical progress achieved at significant cost and organizational effort. It also sits at roughly the stage ASML reached in the early 2000s, and the distance from there to production-grade EUV is where the real work begins.
The most informed estimate — from people inside the project — puts working chips from a Chinese EUV system at 2030, not 2028. That is five years of sustained engineering, integration, and optimization, assuming no major setbacks. Given the complexity of the optics, light source, and systems integration challenges, setbacks are more likely than not.
The strategic implication is not that China has "cracked" EUV. It is that China has decided to crack EUV, has committed the resources to try, and has made enough progress to credibly pursue the goal. Whether it succeeds on a 5-year, 10-year, or 15-year timeline depends on variables — optics breakthroughs, industrial ecosystem development, continued talent acquisition — that are genuinely uncertain.
What is certain is that the export control regime that denied China ASML machines has also guaranteed that China will continue pouring resources into this effort with the intensity of a Manhattan Project. The prototype proves that much. And in the meantime, China's mature-node capacity — 33% of global capacity at 28nm and above, unrestricted in both manufacturing and design — continues to expand, generating the revenue and industrial experience that fund projects like the one in Shenzhen.
By China Made & Tech Team. Independent publication covering Chinese manufacturing and technology innovation for global audiences.
Methodology disclosure: This analysis draws on Reuters' December 2025 exclusive reportage by Fanny Potkin, a March 2026 Reuters report by Che Pan and Laurie Chen, patent filings from the Chinese Academy of Sciences, SemiAnalysis commentary, ASML public statements, and Dutch intelligence assessments. Where claims are attributed to anonymous sources ("people with knowledge of the project"), we have noted this. No Chinese government agency, Huawei, or ASML responded to Reuters' requests for comment on the original reporting. We distinguish throughout between confirmed facts, informed estimates, and speculation.