Why China Cant Copy Asml

TITLE: Why China Can’t Copy ASML CHANNEL: Behind Asia DATE: 2026-04-16 URL: https://www.youtube.com/watch?v=pUHLbwZYVcA ---TRANSCRIPT--- To build an artificial intelligence empire, to launch a hypersonic missile, or to simply manufacture the modern smartphone resting in your pocket, cutting-edge microchips are an absolute non-negotiable prerequisite. Semiconductors are the crude oil of the 21st century. They are the foundational resource upon which the entire modern global economy rests, dictating the balance of geopolitical power and the future of technological supremacy. And to manufacture the most advanced logic chips on the planet, an apparatus that seemingly defies the laws of common sense, physics, and economics is required. It is known as an extreme ultraviolet or EUV lithography scanner.

The scale of this machinery borders on the absurd. A single unit weighs approximately 150,000 kg, which is roughly the physical equivalent of two fully loaded Airbus A320 commercial airliners. Inside its sleek, sterile casing, the machine contains around 100,000 distinct hyper specialized parts, kilometers of complex wiring, and fluid routing systems that must operate in a pristine vacuum. Depending on the specific configuration and the numerical aperture of the lens system, these machines cost anywhere from roughly $180 million for the baseline models to well over $380 million per unit for the absolute latest high numerical aperture iterations. Crucially, they are produced by exactly one entity on the entire planet — not a Japanese technology conglomerate, but ASML, a company headquartered in the quiet, unassuming suburb of Veldhoven in the Netherlands.

Over the past decade, the Chinese government has directed immense capital, most notably through massive state-backed initiatives known as the National Integrated Circuit Industry Investment Fund, or simply the “big fund” into its domestic semiconductor industry. The explicitly stated objective is to achieve total technological self-sufficiency and insulate the nation from western export controls. This leads to a very natural pressing geopolitical question. Why is it that the factory of the world, a nation with effectively unlimited state capital and millions of brilliant engineers, cannot simply purchase an ASML extreme ultraviolet scanner, disassemble it piece by piece, and reverse engineer the technology?

The answer lies in the reality that duplicating ASML’s achievements is not fundamentally a problem of financial resources, labor scale, or political willpower. It is a problem of extreme physics, hyper specialized global supply chains, and the absolute unyielding limits of human espionage.

To understand the impossibility of replicating this machine, it is necessary to first examine the technological wall the semiconductor industry crashed into prior to its invention. To achieve this continuous miniaturization, the industry relies on photolithography. In simple terms, a silicon wafer is coated with a light sensitive chemical called a photoresist. Light is then shined through a blueprint known as a mask or reticle, which projects the microscopic circuit pattern through a series of shrinking lenses and onto the silicon. The exposed areas undergo a chemical reaction, allowing the pattern to be permanently etched into the silicon.

For decades, the global semiconductor industry operated comfortably on deep ultraviolet or DUV lithography. As chipmakers relentlessly demanded the ability to draw smaller and smaller lines to fit more transistors onto a single chip, engineers continually found ingenious physics bending ways to push this 193 nanometer technology further than anyone thought mathematically possible. The most significant leap was the implementation of immersion lithography. By projecting the laser light through a microscopic layer of highly purified water situated between the machine’s final lens and the silicon wafer, engineers manipulated the refractive index of the light. Immersion allowed the industry to push deep ultraviolet machines to a numerical aperture of 1.35, allowing for smaller, denser transistors. However, as transistors approached the 10 nanometer and 7 nanometer physical thresholds, 193 nanometer light became vastly too blunt an instrument.

Due to the diffraction limit of light, using a 193 nanometer wavelength to draw a 7 nm feature is functionally equivalent to trying to paint a highly detailed microscopic portrait using a thick, clumsy house painting brush. To continue printing smaller circuits with deep ultraviolet light, manufacturers were forced to use an arduous technique called multi-patterning. The silicon wafer would be exposed to the first mask, then removed from the scanner, chemically etched, baked, washed, recoated with photoresist, and placed back into the scanner to receive the second mask, which had to be perfectly aligned with the first. While multi-patterning technically allowed for the creation of smaller features, it was an operational and economic nightmare. This predictably caused manufacturing costs to skyrocket. Every time a wafer is removed, processed, and placed back into the scanner, it must be aligned with sub-nanometer precision. The industry had reached a breaking point. It needed a scalpel instead of a sledgehammer. It needed light with a significantly shorter wavelength. It needed extreme ultraviolet.

To comprehend why an extreme ultraviolet scanner cannot be readily copied by a competitor, one must first examine the incredibly violent precision engineered physics required just to turn the machine on. Extreme ultraviolet lithography utilizes light with a wavelength of exactly 13.5 nm. Light of this wavelength does not occur naturally on the surface of the earth. Nor can it be generated efficiently by any conventional lamp, diode, or standard gas laser. To create 13.5 nanometer light, the ASML scanner relies on a process so complex it borders on science fiction, a laser produced plasma source.

Inside a massive sealed vacuum chamber, a highly specialized droplet generator fires microscopic spheres of ultra pure molten tin. These droplets are fired downward through the dark void of the vacuum chamber at a staggering velocity of 70 to 80 m/s or roughly 250 km/h. As the microscopic droplet falls through the void, it is tracked by a suite of incredibly precise high-speed sensors. A relatively low intensity prepulse strikes the spherical tin droplet first. A perfect sphere is a highly inefficient shape to vaporize because it has low surface area relative to its volume. The kinetic energy from the prepulse flattens the spherical droplet into a microscopic disc or pancake shape, vastly increasing its surface area.

A microscopic fraction of a millisecond later, while the flattened tin is still moving at extreme velocity, the main laser pulse strikes the tin target with full devastating power. This instantaneous injection of energy vaporizes the tin entirely, turning it into a superheated, violent plasma. This resulting plasma reaches temperatures approaching 220,000° C. It is this momentary microscopic supernova that ultimately emits the isotropic 13.5 nanometer extreme ultraviolet radiation needed to print the chips.

This staggering sequence of events — firing the microscopic droplet, tracking its exact trajectory, hitting it with a prepulse to deform its shape, and obliterating it with a main pulse to create a miniature sun — must be executed with flawless, unyielding precision 50,000 times every single second. As one ASML executive famously noted, controlling light beams and physical matter with such localized spatial and temporal accuracy is akin to shining a laser pointer from the Earth and successfully hitting a specific coin resting on the surface of the moon. A critical metric in this incredibly violent process is conversion efficiency. ASML targets a commercial standard of approximately 5.5% conversion efficiency. Furthermore, incinerating 50,000 droplets of tin per second creates an immense amount of microscopic debris. To combat this, ASML had to engineer a highly complex debris mitigation system that flows hydrogen gas through the vacuum chamber at specific speeds and pressures, effectively creating a microscopic hurricane that catches the stray tin atoms and carries them away before they can land on the precious optics, all without disrupting the path of the extreme ultraviolet light.

But generating the light is only the first impossible hurdle. The second is controlling it and managing the incredibly hostile environment in which it exists. Because extreme ultraviolet light has such a short wavelength, it is highly fragile. Consequently, an extreme ultraviolet scanner’s entire optical pathway must operate in a high vacuum. And instead of using refractive glass lenses to bend the light, the machine must utilize an incredibly complex system of reflective mirrors to collect, focus, shape, and project the light onto the silicon wafer.

These are not standard mirrors. The state-of-the-art mirrors utilized in ASML’s systems measure up to a meter across and must be shaped with an accuracy that essentially defies human comprehension. They are polished using highly specialized proprietary techniques including computer-controlled fine correction and ion beam figuring to achieve a surface that is smooth down to tens of picometers. To reflect the 13.5 nm light effectively, the mirrors must utilize the principles of Bragg reflection. They must be coated with approximately 100 alternating hyper thin layers of specific materials almost always alternating between silicon and molybdenum. Each individual layer is only a few nanometers thick and the exact thickness must be maintained with atomic precision across the entire meter wide surface of the mirror to create constructive interference that bounces the light back into the vacuum. The remaining 30% is inevitably absorbed as pure heat. This necessitates incredibly advanced, perfectly calibrated liquid cooling channels integrated directly into the back of the mirrors to prevent them from expanding or warping even a fraction of a nanometer under the immense thermal load.

Inside an ASML scanner, the extreme ultraviolet light bounces across roughly 11 separate mirrors as it is collected from the plasma source, shaped, passed through the reticle mask, and focused down onto the silicon wafer. Less than 2% of the original extreme ultraviolet radiation successfully reaches the semiconductor wafer. The laser must pump in tens of thousands of watts of power just to guarantee that a tiny fraction of a watt of extreme ultraviolet light successfully makes it to the silicon.

Attempting to reverse engineer a machine that operates strictly on the bleeding edge of quantum physics, fluid dynamics, and material science simply by dismantling it is a fool’s errand. The hardware tells you what was built, but it does not tell you how to build it. But even if a competing nation managed to perfectly map, chemically analyze, and completely comprehend every single one of the 100,000 components inside the scanner, they still would not be able to build the machine. The company orchestrates a sprawling, hyper specialized global supply chain comprising approximately 1,200 distinct, highly advanced partner organizations spread across the globe.

Consider the high-powered carbon dioxide laser amplifier utilized to vaporize the tin droplets. That component is entirely produced by a German firm named Trumpf located in Ditzingen. The system utilizes a high power seed module consisting of two seed lasers that generate initial pulses of just a few watts. The sheer complexity of maintaining stable discharge and uniform gain in a fast flow carbon dioxide laser amplifier operating at low cavity pressures represents an absolute pinnacle of optical engineering that took Trumpf decades to perfect. The underlying extreme ultraviolet light source technology, including the highly sensitive droplet generator that spits out the molten tin and the physics required to make the plasma conversion work, was pioneered by Cymer, a company based in San Diego, California.

The impossibly flat picometer precise Bragg reflecting mirrors are exclusively crafted by Carl Zeiss SMT in Oberkochen in Germany. The relationship between ASML and Zeiss is so deeply intertwined that the two companies operate in a tightly coupled exclusive development alliance. Zeiss literally cannot sell these specific mirrors to anyone else and ASML cannot buy them from anyone else. Meanwhile, the ultra complex wafer handler, the robotic module responsible for physically moving the silicon into the machine with microscopic precision, is developed collaboratively with the VDL group, a high-tech manufacturing conglomerate, also located in the Netherlands. The stages that hold the wafers use magnetic levitation, hovering in a vacuum while accelerating at forces exceeding 3G, yet stopping with sub-nanometer precision measured by incredibly complex interferometers.

If China intends to successfully clone an ASML scanner, they do not merely have to replicate the intellectual property of one Dutch firm. They must simultaneously and perfectly recreate the apex achievements of the German optics industry, the American laser industry, and the precision mechatronics of the Dutch ecosystem. ASML operates as the conductor of a global symphony. And due to stringent United States export controls, diplomatic embargos, and multilateral agreements, China has been thoroughly isolated from the orchestra.

There is a persistent underlying assumption in geopolitical discourse that if state sponsored hackers could simply infiltrate ASML’s servers and exfiltrate the raw blueprints, schematics, and source code, China could successfully build an extreme ultraviolet scanner. This assumption fundamentally ignores the single most powerful defensive moat in high-end manufacturing, tacit knowledge.

In the realm of advanced engineering, knowledge is bifurcated. Explicit knowledge is the data written down in a manual, digitized in a CAD file, stored on a corporate server, or printed on a blueprint. Explicit knowledge tells you the dimensions of a part, the chemical composition of a coating, or the wiring diagram of a laser amplifier. Tacit knowledge, however, is the unwritten intuitive know-how that engineers, technicians, and machinists accrue over decades of hands-on trial, error, and physical tinkering. It is the deep experiential context that tells an engineer not just what to build, but how to handle the inevitable microscopic failures that occur during the building process. When a Zeiss technician oversees the polishing of a multi-layer mirror, or an ASML integration engineer aligns the dual pulse lasers to perfectly strike a microscopic droplet of liquid tin moving at 70 m/s, the precise calibration techniques and troubleshooting protocols exist solely within the minds and hands of those veteran employees.

Because China is legally and strictly barred from purchasing cutting-edge extreme ultraviolet machines due to comprehensive United States export controls which recently expanded in 2023 and 2024 to restrict even advanced immersion deep ultraviolet machines from entering the country, and because they cannot independently replicate the technology overnight, the nation’s foundries are forced into an arduous financially bruising technological workaround.

Consequently, to produce advanced logic chips for modern smartphones, 5G base stations, and artificial intelligence accelerators, China’s leading foundry, Semiconductor Manufacturing International Corporation, or SMIC, must rely heavily on older existing deep ultraviolet machines that were stockpiled before Western sanctions tightened. To fabricate 7 nanometer chips using older deep ultraviolet hardware, SMIC is forced to employ the dreaded multi-patterning technique. Instead of etching the circuit onto the silicon in one clean, precise pass as an extreme ultraviolet machine does, the wafer must be run through the DUV scanner three, four, or even more times to achieve the necessary feature density.

However, from a commercial and economic standpoint, deep ultraviolet multi-patterning at 7 nanometers is an absolute nightmare. Each additional lithographic exposure dramatically multiplies the probability of catastrophic defects. If a single layer is misaligned by a fraction of a nanometer, the transistors will not connect properly and the chip will be dead on arrival. If a microscopic speck of dust lands on the wafer during the third pass, the entire chip is ruined. Consequently, reports regarding SMIC’s yield rates, which is the percentage of chips on a completed wafer that actually function correctly and can be sold to consumers, vary wildly, but consistently paint a picture of severe operational inefficiency. Yet, even at the highest and most generous estimates, these figures fall woefully short of the industry standard required for profitable high volume commercial production, which typically demands yield rates well over 80%. Because of this massive scrap rate, SMIC reportedly charges 40 to 50% more for these advanced chips than their Taiwanese equivalent, TSMC, charges for the exact same class of silicon printed with EUV.

If the primary goal is national security, artificial intelligence development for state surveillance and military self-sufficiency, a heavily subsidized state-managed economy can afford to absorb these exorbitant yield penalties. But if the goal is to aggressively compete in the global consumer electronics or cloud compute market against Western companies leveraging ASML’s extreme ultraviolet technology at high volume, China is operating at a severe structurally baked-in disadvantage. The production volume disparity highlights this reality perfectly. In recent years, TSMC reportedly manufactured approximately 100,000 to 120,000 advanced wafers per month using highly efficient EUV lines, whereas SMIC’s advanced output, bottlenecked by the slow and tedious multi-patterning process, was estimated at nearly 3,000 to 4,000 wafers per month.

China commands immense, virtually unlimited state capital, and its domestic engineering talent is undeniably brilliant. They are actively building their own internal ecosystem, heavily funding homegrown lithography alternatives like SMEE machinery, attempting to bypass traditional lithography with advanced packaging techniques, and forging massive domestic supply chains entirely in the dark, cut off from Western expertise and oversight. The resilience and speed of their semiconductor sector in the face of immense coordinated geopolitical pressure from the United States, Japan, and the Netherlands is an undeniable reality. But limitless capital cannot purchase a time machine.

China is not merely attempting to copy a physical piece of hardware. They are attempting to artificially compress over 40 years of compounded global collaboration, iterative failure, shared financial risk, cross-border academic research, and deeply entrenched tacit human experience into a single decade. Doing so completely isolated from the rest of the scientific world. They must solve the physics of superheated plasma, the atomic chemistry of Bragg reflecting mirrors, and the mechanics of perfect vacuums without the help of the Germans, the Americans, or the Dutch. And in the brutally precise realm of advanced semiconductor manufacturing, where the margin for error is measured in individual atoms and fractions of a nanometer, time is the ultimate unforgiving moat. ASML’s monopoly is not secured merely by legal patents or geopolitical export controls. It is secured by the sheer terrifying reality-bending complexity of the machine itself.