Why China’s Favourite Colour Is…..Ultra Violet!

China has just crossed a threshold in the semiconductor war which matters far beyond semiconductors. At the end of July 2026, Reuters reported that Shanghai Aishengna Electronic Technology Group had begun producing domestically developed immersion deep-ultraviolet lithography machines, with initial systems expected to go to major Chinese chipmakers including SMIC, Hua Hong and CXMT. The machines remain well behind ASML’s best equipment in throughput, reliability and precision, while China still lacks a commercially competitive EUV system. The important point is therefore not that China has suddenly caught ASML. It has not. The important point is that China has begun manufacturing domestically one of the most strategically important categories of machine that Washington and its allies had hoped would remain a technological chokepoint. 

Lithography is essentially the process by which extraordinarily small patterns are projected onto silicon wafers so that transistors and other components can be created. Deep-ultraviolet lithography uses ultraviolet light to print those patterns, while immersion DUV improves resolution by placing a thin layer of water between the optical system and the wafer. The most advanced chips are now produced far more efficiently with extreme-ultraviolet lithography, where ASML retains an extraordinary technological lead, yet DUV can be pushed much further than its nominal resolution suggests by exposing layers several times through multiple patterning. That process is slower, more complicated and more expensive than using EUV, with more opportunities for defects and lower yields, yet it means that lacking EUV does not automatically prevent a manufacturer from producing surprisingly advanced chips. 

This distinction changes the way we should think about the US-China semiconductor conflict. For several years the Western strategy has rested upon a powerful idea. China could design increasingly sophisticated chips and spend enormous amounts building semiconductor factories, yet it would remain dependent upon a small number of technologies which were fantastically difficult to reproduce. Lithography was the most important example because ASML became something close to the ultimate technological chokepoint. Its machines are not simply large pieces of industrial equipment that can be copied once someone has obtained the blueprints. They embody decades of accumulated knowledge across optics, lasers, sensors, control systems, materials and precision engineering, while the wider ecosystem supporting them is almost as formidable as the machines themselves.

Control access to that equipment and, in theory, you can control how quickly another country reaches the technological frontier. That logic explains why Washington has devoted so much diplomatic effort to restricting Chinese access to advanced semiconductor manufacturing tools and why the Netherlands became such an important participant in American technology policy. Proposed US legislation in 2026 went further by targeting immersion DUV sales and servicing, reflecting the growing importance attached not simply to the most advanced EUV systems but to the older equipment China has learned to use increasingly effectively. 

The logic is perfectly understandable. Advanced semiconductors power artificial intelligence, communications networks, cyber systems, surveillance, autonomous weapons and modern military capabilities. Restricting the machinery required to manufacture them can slow a strategic competitor. Economic warfare, however, rarely stops with the first-order consequence. Export restrictions made foreign lithography harder for China to obtain, while simultaneously making domestic lithography dramatically more valuable.

Before restrictions intensified, a Chinese semiconductor company choosing between an ASML machine and a substantially weaker domestic alternative faced a straightforward commercial calculation. The ASML machine offered higher productivity, better yields, greater reliability and access to an immense engineering ecosystem. Buying the weaker Chinese machine imposed a significant economic penalty. Restrictions changed the definition of reliability itself. If the superior foreign machine might become unavailable tomorrow, if servicing could be restricted later or if spare parts might be controlled by a foreign government, reliability no longer means simply whether the machine breaks down. It also means whether someone outside China can stop you using it.

A weaker domestic machine can therefore acquire enormous strategic value simply because it is domestic. Chinese manufacturers may be prepared to accept lower throughput, greater costs and more difficult yields because technological sovereignty has itself acquired economic value. This is the central unintended consequence of export controls. They create a commercial market for technologies which might otherwise have struggled to compete with Western incumbents.

The argument becomes much more important when we think about industrial learning. Aishengna’s first machines will almost certainly be inferior to ASML’s. That fact matters less than it might appear because industrial technologies improve through use. Machines are installed, engineers discover problems, manufacturers obtain feedback, components are redesigned and each subsequent generation becomes better. Customers create the environment in which the technology learns.

China possesses an extraordinary advantage at precisely this stage because it has scale. SMIC is China’s largest contract semiconductor manufacturer, Hua Hong operates major foundries, CXMT is expanding rapidly in memory and Huawei creates immense downstream demand for semiconductors. Reuters reported that the first Chinese immersion DUV systems are expected to go to major domestic manufacturers, while reported production plans of around five machines this year and twenty in 2027 remain tiny next to ASML’s output. That immediate numerical gap is real. The strategically interesting question is what Chinese machines look like after several years of engineers installing them, fixing them and redesigning them. 

This is the point at which comparisons with other Chinese industries become tempting. China has repeatedly entered sophisticated industrial sectors with products which initially looked weaker, cheaper or less refined than those of incumbent Western or Asian manufacturers. Domestic scale then generated engineering experience, manufacturing capacity expanded, prices fell and quality improved. Solar panels followed something like this trajectory. Electric vehicles and batteries followed another version of it. Semiconductor lithography is considerably harder than any of those industries, so the analogy should never become a prediction. The useful lesson is simply that an inferior first-generation machine should not be judged only as a finished product. It can also be judged as the beginning of an industrial learning curve.

The first Chinese DUV machine does not have to defeat ASML. It has to make the second, third and fifth generations possible.

Financial markets appeared to understand this immediately. ASML lost roughly €60 billion of market value within two days of reports about China’s domestic DUV programme, even though analysts simultaneously emphasised that the immediate commercial threat looked limited and that ASML’s technological position remained formidable. Investors were not really pricing the few Chinese machines expected in 2026. They were repricing the possibility that one assumption embedded in ASML’s long-term strategic value might gradually become less secure: Chinese dependence upon Western lithography. 

The consequences for Europe are uncomfortable because ASML is one of Europe’s rare genuinely indispensable technology companies. Europe frequently worries that it lacks equivalents to America’s enormous technology platforms. ASML is something more strategically valuable than another social network or software company because it occupies a crucial point in the industrial system from which virtually the entire modern digital economy ultimately flows.

That technological position gives the Netherlands geopolitical leverage. Washington needs Dutch cooperation because American export controls cannot independently prevent a European company selling particular classes of lithography equipment. The Dutch government therefore becomes extraordinarily important to American technology strategy despite the relatively small size of the Netherlands.

Chinese progress gradually erodes that leverage before it erodes ASML’s technological lead.

The distinction is fundamental because technological monopolies do not lose geopolitical value only when competitors become equally good. Coercive power begins weakening when the target acquires a sufficiently usable alternative. Imagine a country depends completely upon a single foreign supplier for an essential component. The supplier can threaten to cut access and therefore possesses enormous leverage. Imagine instead that the dependent country can manufacture a domestic substitute costing twice as much while delivering only 70 per cent of the performance. Commercially the foreign product remains dramatically superior. Geopolitically the situation has already changed because the threatened country can survive.

Technological superiority and technological sufficiency are not the same thing.

Washington has generally sought to keep China generations behind the frontier. Beijing may increasingly conclude that it does not need to reach the frontier immediately if it can eliminate existential dependence upon foreign suppliers. America wants China to remain technologically behind. China wants to ensure that being behind does not mean being vulnerable.

This matters because the global semiconductor economy consists of far more than the most advanced Nvidia AI accelerators. Cars require chips, factories require chips, telecommunications networks require chips, missiles require chips, power systems require chips, consumer electronics require chips and almost every piece of modern military equipment contains semiconductors. Mature and intermediate nodes remain economically and strategically indispensable. A country capable of independently producing the equipment necessary for those chips becomes much harder to isolate even while remaining significantly behind in the most advanced processes.

Domestic lithography therefore increases the possibility of something Western policymakers already worry about: enormous Chinese semiconductor overcapacity. China has powerful reasons to expand chip manufacturing because semiconductors are simultaneously an economic industry, a technological foundation and a national-security asset. Once more of the equipment inside Chinese fabs can also be produced domestically, the incentive to build additional capacity grows stronger.

The pattern could eventually resemble other Chinese sectors in which state support encourages investment, capacity rises faster than domestic demand, fierce competition pushes prices down and Chinese manufacturers begin looking abroad. Western governments have already worried about Chinese dominance in mature-node semiconductors because those chips are embedded everywhere. Domestic lithography makes that prospect more credible because it reduces the capacity of export controls to constrain future expansion.

This produces a remarkably awkward contradiction for Western industrial policy. The United States and Europe want semiconductor supply chains to become more resilient, while simultaneously wanting China to become less important within them. Achieving both objectives requires enormous duplication. America subsidises fabs. Europe subsidises fabs. Japan subsidises semiconductor capacity. India wants its own manufacturing ecosystem. China spends even more heavily building independence.

Once every major government decides semiconductors are national-security infrastructure rather than merely commercial products, the logic of globalisation begins to reverse. Efficiency gives way to resilience. Factories are duplicated, government subsidies increase, capital expenditure rises and someone ultimately pays for the redundancy.

China’s DUV progress accelerates this process because it suggests that technological decoupling will not simply consist of the West denying China certain tools. China is attempting to remove the dependence that gives denial its power.

That raises the possibility of two increasingly distinct semiconductor ecosystems. One would remain centred around the extraordinary capabilities found across the United States, the Netherlands, Japan, Taiwan and South Korea, while another would become increasingly centred upon China. Complete separation is unlikely because the semiconductor supply chain is too complex and commercial incentives are too powerful, yet even partial duplication changes the global economy.

Lithography matters disproportionately because it is one of the hardest links in the chain to reproduce. Progress there sends a signal to the entire Chinese semiconductor equipment industry that localisation is possible. Companies producing etching, deposition, cleaning and other fabrication equipment already have stronger domestic positions than China does in lithography. The competitive effect is beginning to extend beyond Chinese customers. Reuters reported on August 5 that Samsung and SK Hynix have been evaluating equipment from Chinese company AMEC as a hedge against the possibility of tighter American restrictions affecting their Chinese factories. Deutsche Bank estimates cited by Reuters suggest Chinese equipment companies could take a very substantial share of China’s wafer-fabrication equipment market outside lithography and metrology. 

That development hints at a potentially extraordinary reversal. Export restrictions were partly designed to prevent Chinese semiconductor equipment companies becoming globally competitive. Those same restrictions may help provide Chinese companies with a captive domestic market large enough to finance the engineering process required to become globally competitive.

This does not mean export controls have failed. Such a conclusion would be far too simplistic. China would almost certainly have progressed faster in some areas if its companies enjoyed unrestricted access to the world’s best machines. Denying EUV imposes genuine technological and economic costs. Multiple patterning with DUV requires more process steps, additional masks and greater opportunities for defects, while poorer yields increase costs. The frontier therefore remains much more difficult to reach.

The more interesting question is whether export controls have different effects across different time horizons. In the short term they slow China. In the medium term they strengthen the commercial case for Chinese substitutes. Over the long term they can potentially erode the dependence which made the controls powerful in the first place.

Sanctions elsewhere have repeatedly produced variations of this dynamic. Restrict a country’s access to an imported technology and the immediate result is pain. Sustain the restriction long enough and the targeted country acquires enormous incentives to produce the technology domestically, obtain it from another supplier or redesign its economy around the restriction. Sometimes those attempts fail. Sometimes the substitute remains permanently inferior. Sometimes the protected domestic industry becomes remarkably successful. Policymakers rarely know with certainty which of those outcomes they are creating.

China’s scale makes successful substitution more plausible than it would be for most countries. A smaller economy attempting to reproduce every component of the semiconductor supply chain would rapidly discover that its domestic market could not justify the extraordinary investment required. China can potentially spread huge research and development costs across an immense industrial base. State financing can extend the time horizon still further because the Chinese government does not require every lithography programme to satisfy the same return-on-capital expectations as a Western listed company.

Strategic independence can itself become the return.

This creates an unusual economic competition because the Chinese and Western systems may increasingly be optimising different things. ASML wants to manufacture equipment that maximises technological performance while producing commercial returns. China wants equipment which eventually makes the country impossible to blockade technologically. Those objectives overlap, although they are not identical.

The military ramifications emerge naturally from the same argument. Modern war is extraordinarily semiconductor-intensive. Radar, electronic warfare, missiles, drones, satellites, communications systems, artificial intelligence and autonomous platforms all depend upon chips. No serious military power wants to discover during a major conflict that an essential component of its weapons supply chain can be stopped by a foreign government.

Lithography is consequently becoming strategic infrastructure in much the same way that oil refineries, steel mills and shipyards were strategic infrastructure during earlier industrial eras. A domestically controlled semiconductor manufacturing ecosystem increases China’s capacity to sustain military production during any prolonged confrontation.

Taiwan then enters the discussion almost automatically. Any conflict involving Taiwan would strike directly at the most important concentration of leading-edge semiconductor manufacturing on Earth because TSMC remains central to the global chip frontier. China’s attempt to construct an independent semiconductor ecosystem gradually reduces one dimension of Beijing’s vulnerability to a disruption centred around Taiwan.

No one should mistake several Chinese DUV machines for the replication of TSMC. The technological difference remains enormous. The direction nevertheless matters because every dependency China removes changes the economic calculation surrounding a potential confrontation.

Economic interdependence can discourage conflict when governments believe they have too much to lose. Decoupling can reduce those losses. Washington wants less dependence upon Chinese supply chains because dependence creates vulnerability. Beijing wants exactly the same thing in reverse.

Success on both sides therefore creates another uncomfortable possibility. America and China could each become safer from economic coercion while simultaneously becoming freer to confront one another.

This begins to resemble the security dilemma familiar from military affairs. One state develops a capability because dependence upon another appears dangerous. The other sees the new capability as threatening and introduces additional restrictions. Those restrictions validate the original decision to pursue greater self-sufficiency. Each side can then point to the other’s behaviour as proof that its own policy was necessary.

China’s DUV progress may therefore produce tighter Western controls rather than weaker ones. American policymakers could reasonably conclude that continuing to supply Chinese manufacturers with advanced Western DUV systems simply gives China a bridge until domestic equipment becomes good enough. Restrictions on new sales, servicing, software, spare parts and technical assistance consequently become more attractive. Beijing then receives still more evidence that foreign technology cannot be considered strategically dependable.

America’s allies occupy an increasingly difficult position inside this dynamic. Washington possesses strong national-security reasons for restricting Chinese access to semiconductor technology. ASML bears some of the lost commercial opportunity. Japanese equipment manufacturers bear another portion. South Korean and Taiwanese companies operating fabrication plants in China confront their own complications.

The arrangement remains politically sustainable while export controls retain clear strategic effectiveness. The calculation becomes more uncomfortable if restrictions simultaneously accelerate the emergence of Chinese competitors.

The Netherlands then confronts a particularly difficult question. How much Chinese revenue should ASML sacrifice today to preserve a technological advantage which Chinese companies are attempting to eliminate anyway? There is no purely commercial answer because ASML’s capabilities have obvious strategic significance. There is no purely strategic answer either because commercial revenues support the research and development ecosystem which keeps ASML ahead.

Technology monopolies generate geopolitical leverage partly because they generate extraordinary economic returns. Exploit the geopolitical leverage too aggressively by excluding a very large market and it is at least possible that the commercial foundations of the monopoly become weaker while the excluded market finances a competitor.

China does not need to defeat ASML globally for this strategy to have profound consequences. For the moment it may be sufficient to replace ASML gradually within China. Chinese equipment companies would inherit a gigantic domestic market, scale would finance further improvement and improving equipment could eventually become exportable.

The next stage of the semiconductor competition could therefore move into emerging markets.

Imagine a manufacturer several years from now choosing between two machines. The Western machine remains technically superior, costs more and comes with the possibility that future access could be restricted by Western governments. The Chinese machine performs adequately, costs less and is sold without equivalent restrictions.

A commercially rational purchaser in a non-aligned country might begin calculating technological sovereignty alongside technical performance.

Versions of this dynamic already exist in telecommunications, electric vehicles and infrastructure. Technology becomes geopolitical when access is conditional.

China might eventually market itself not merely as the cheaper supplier but as the supplier which will not participate in Western technology embargoes. There is considerable irony here because Beijing has itself demonstrated a willingness to restrict strategically valuable exports when doing so serves Chinese interests. Technological sovereignty does not abolish economic coercion. It redistributes the ability to practise it.

This may be the largest consequence of China’s DUV advance.

For approximately three decades, globalisation encouraged countries to specialise. One economy designed a product, another fabricated it, another manufactured critical equipment and another supplied materials. Efficiency emerged from interdependence.

Great-power rivalry is reversing the underlying question. Governments increasingly ask not where something can be manufactured most cheaply, but whether they can afford not to manufacture it themselves.

China’s lithography programme belongs to that new world. American semiconductor subsidies belong to it too. European industrial policy does as well.

The resulting system will almost certainly be more expensive than the highly integrated global supply chains it replaces. It may also be more resilient against sanctions and political disruption. Greater resilience carries its own danger because interdependence was one of the mechanisms through which confrontation imposed costs on everyone.

China’s new DUV machines therefore matter much less because of what they can manufacture today than because of what their existence tells us about the changing nature of the semiconductor war.

The West’s strongest strategy has been built around technological chokepoints. China’s response is increasingly built around eliminating them.

ASML still leads by an enormous margin. EUV remains outside China’s commercially proven capabilities, while Aishengna’s early systems may struggle with throughput, reliability and yield for years. 

None of that makes the development strategically trivial because the decisive threshold is not necessarily technological equality.

The decisive threshold may be the moment when dependence becomes optional.

China appears to have taken another step towards that point.

Export controls can deny a rival technology, delay its progress and substantially increase the cost of catching up. They cannot guarantee permanent dependence. The more painful dependence becomes, the greater the incentive to escape it.

China has not reproduced ASML and may remain years behind in EUV. The more interesting question is whether China ultimately needs technological parity with ASML in order to defeat the strategy built around its dependence upon ASML.

If Chinese factories can manufacture sufficient quantities of sufficiently advanced chips using increasingly Chinese machinery, while each generation of that machinery improves, Washington may eventually discover that it preserved the West’s technological lead while gradually losing the West’s technological leverage.

Those are two very different things.

The economic consequences would be profound because they imply duplicated semiconductor ecosystems, enormous state investment, increased competition in mature chips and growing pressure on Western equipment companies. The geopolitical consequences could be larger because a China which can sustain its technological and military infrastructure without access to Western machinery becomes progressively harder to coerce.

The semiconductor war was supposed to determine who controls the technologies of the future. China’s leap forward in DUV lithography suggests that it may increasingly turn on a rather older question. What happens when the country you are trying to contain learns how to deconstruct the walls?