Scientists in Shenzhen have completed a working extreme ultraviolet (EUV) lithography machine that could reshape the entire semiconductor power balance. The question isn’t whether China can build advanced chips anymore. It’s when.
When US Commerce Secretary Gina Raimondo visited Beijing in late August 2023, Huawei had a surprise waiting. The Mate 60 Pro appeared on shelves powered by a 7nm Kirin 9000s chipset that wasn’t supposed to exist. Not under US sanctions. Not without extreme ultraviolet lithography. Not without ASML’s machines, which Washington had spent years keeping out of Chinese hands.
That phone wasn’t a fluke. It was a warning shot.
Now, according to Reuters, sources say China has quietly assembled a prototype EUV machine inside a high-security laboratory in Shenzhen. The machine is operational, generating the extreme ultraviolet light needed for advanced chipmaking. It hasn’t produced working chips yet, but Beijing is targeting 2028 for production. Industry observers reckon 2030 is more realistic, but that’s still well ahead of where Western intelligence agencies expected China to be.
This isn’t just industrial espionage or reverse engineering gone right. It’s a nationally coordinated programme that sources compare to the Manhattan Project. Huawei is at the centre, coordinating a network of companies and state research institutes across China. Thousands of engineers are working on this. The prototype was completed earlier this year. It fills nearly an entire factory floor. And it was built, in part, by former ASML engineers.
The machine Washington tried to kill
To understand why this matters, you need to know what an EUV machine actually does. Lithography is the process of printing microscopic circuit patterns onto silicon wafers. The smaller the wavelength of light, the finer the detail you can etch. Deep ultraviolet (DUV) machines use 193nm light. That’s fine for older chips, but as transistors shrink toward 5nm, 3nm, and below, DUV can’t keep up. You need multiple exposures, complex workarounds, and yields suffer.
EUV uses 13.5nm light. It’s a different game entirely. You can print features in a single pass that would take four or five exposures with DUV. That means faster production, fewer defects, and the ability to manufacture the kinds of chips that power AI models, data centres, and high-performance computing.
Only one company on Earth makes EUV machines: ASML, a Dutch firm that spent decades perfecting the technology. Each machine weighs over 150 tonnes, costs upwards of $150-million, and requires 20 trucks and three Boeing 747s to ship. ASML’s latest High NA EUV systems cost closer to $350-million. No single person at ASML fully understands how the entire machine works. That’s how complex they are.
And since 2019, the US has been leaning on the Netherlands to block sales of EUV machines to China. When the Trump administration couldn’t kill Huawei through sanctions alone, it went after the supply chain. If China couldn’t access EUV, the thinking went, it couldn’t make cutting-edge chips. Problem solved.
Except China didn’t stop. It doubled down.
How sanctions accelerated the timeline
US export controls weren’t supposed to work like this. The strategy was straightforward: restrict China’s access to advanced chipmaking equipment, slow down its technological progress, and maintain a generational lead in semiconductors. That lead translates into military advantage, AI dominance, and control over the global tech stack.
But sanctions have an awkward tendency to backfire. When you tell a country it can’t have something critical to its future, and that country has the state capacity and resources of China, it doesn’t just accept defeat. It mobilises.
Beijing has poured billions into domestic semiconductor development. The China Integrated Circuit Industry Investment Fund has backed everything from chip design to equipment manufacturing. Huawei, once a customer of Western suppliers, became a systems integrator. SMIC, China’s largest foundry, started pushing DUV machines to their absolute limits, using techniques like multi-patterning to achieve 7nm production without EUV.
That’s what powered the Mate 60 Pro. SMIC used ASML’s older DUV machines, legally purchased before sanctions tightened, and figured out how to make them do things they weren’t designed for. Yields were lower. It was inefficient. But it worked.
Now China’s building its own EUV machine, and the timeline keeps shrinking.
Two teams, one goal
According to Chinese media reports, two parallel teams are working on EUV light sources. One is led by Lin Nan, a former ASML scientist who left the Netherlands in 2021 to join the Shanghai Institute of Optics and Fine Mechanics. He’s now at Beihang University. The other is led by Zhao Yongpeng at Harbin Institute of Technology.
Both teams are using solid-state lasers to heat and vaporise tin droplets, generating EUV light. ASML uses CO₂ lasers supplied by its US-based subsidiary, Cymer. China’s taking a different route, partly because solid-state lasers are already widely deployed in Chinese industry, and partly to avoid patent barriers.
Lin’s team recently published a paper claiming a conversion efficiency of 3.42% in generating EUV light. That’s better than what Europe’s Advanced Research Center for Nanolithography achieved in 2019. Lin’s team reckons they can hit 6%, which would meet the commercial standard of 5.5%.
Zhao’s team is using a laser-assisted discharge-produced plasma (LDP) approach, which adds high-voltage electricity to accelerate plasma formation. The output is around 100 watts, still well below ASML’s 600 watts, but it’s functional.
These aren’t theoretical exercises. This is working hardware.
ASML reported in its 2022 annual report that a former employee in China had misappropriated proprietary data. The company initiated an internal review and reported the incident to authorities. There’s no evidence linking Lin to that case, but the timeline is suggestive. He left ASML in 2021. The report came out in 2022. The prototype was completed in 2025.
What this means for the US-China tech war
If China can produce functional EUV machines by 2028, the entire logic of US semiconductor sanctions collapses. The goal was to create a chokepoint so tight that China couldn’t advance without Western technology. But chokepoints only work if they’re insurmountable. If China builds a workaround, the sanctions don’t just fail. They accelerate the very self-sufficiency they were meant to prevent.
This isn’t hypothetical. It’s already happening. Huawei’s Mate 60 Pro was a proof of concept. China’s domestic AI chip efforts are accelerating. Huawei’s Ascend series is designed to compete with Nvidia’s data centre GPUs. SMIC is testing Chinese-made DUV systems from companies like Yuliangsheng, which has ties to Huawei-linked equipment maker SiCarrier.
Western policymakers assumed China was years, maybe a decade, away from indigenous EUV capability. That assumption is looking shaky.
The real risk isn’t just that China catches up. It’s that China’s model of state-coordinated industrial policy proves more effective than the fragmented, market-driven approach in the West. The US can outspend China on R&D, but can it match the focus and coordination of a system where the state, industry, and academia move in lockstep?
The backlash won’t stop the machine
Washington’s response has been predictable. More sanctions. More export controls. More entities added to blacklists. The Biden administration tightened restrictions on high-bandwidth memory chips, chipmaking tools, and advanced AI hardware. The Trump administration added dozens of Chinese entities to trade blacklists. Congress is now debating bills that would codify bans on advanced chip sales to China.
But enforcement is porous. Smuggling networks operate openly in Shenzhen. Nvidia GPUs that aren’t supposed to reach China turn up in data centres anyway, routed through Malaysia, Singapore, or shell companies. Huawei reportedly tricked TSMC into manufacturing two million chiplets for its Ascend AI processors by using intermediaries.
China’s also hedged its bets. While building EUV machines, it’s still buying up every DUV system it can get. ASML’s older tools aren’t banned. China’s been stockpiling them, betting that even if EUV takes longer than expected, it can push DUV far enough to stay competitive.
The Global Times, a state-aligned newspaper, published a commentary dismissing the Reuters report as Western anxiety. The article didn’t deny the EUV lab exists. It just framed the whole thing as evidence that sanctions only strengthen China’s resolve.
That’s not propaganda. It’s strategy.
What happens when the monopoly breaks
ASML’s monopoly on EUV machines has shaped the entire semiconductor industry. Only a handful of companies in the world can afford to buy them: TSMC, Samsung, Intel, and a couple of others. That concentration of power has kept chipmaking centralised, predictable, and under the influence of Western export controls.
If China breaks that monopoly, even partially, the calculus changes. Chinese fabs won’t need to rely on ASML. They won’t need to worry about export licences or sanctions. They’ll have their own supply chain, from design tools to lithography to packaging.
That doesn’t mean China will suddenly leapfrog TSMC or Samsung. EUV machines are just one piece of the puzzle. You still need photomasks, photoresists, metrology tools, and a thousand other specialised components. But it removes the single biggest bottleneck.
And once that bottleneck is gone, China’s scale kicks in. It has more engineers, more state funding, and more political will to push semiconductor self-sufficiency than any other country. The US and Europe are still trying to figure out how to reshore chip manufacturing. China’s already doing it.
The uncomfortable truth
The Mate 60 Pro embarrassed Washington because it proved that sanctions leak. The EUV machine in Shenzhen is worse. It proves that sanctions accelerate the very thing they’re meant to prevent.
Beijing didn’t want to build its own EUV machines. It would have been easier, cheaper, and faster to just buy them from ASML. But once that option was taken off the table, China had no choice but to innovate. And now it’s on the verge of succeeding.
The 2028 target might slip to 2030. The machines might not match ASML’s performance for years. But the trajectory is clear. China’s building its own semiconductor ecosystem, and every sanction, every export control, every blacklist just makes that ecosystem more self-contained.
Washington wanted to slow China down. Instead, it gave China a roadmap for independence.