January 10, 2026

In a high-security laboratory in Shenzhen, occupying nearly an entire factory floor, sits a machine that could reshape the global balance of power. This isn't a weapon in the traditional sense. It's a prototype extreme ultraviolet (EUV) lithography system—the most complex manufacturing tool humanity has ever created—and it represents China's "Manhattan Project" for semiconductor independence.

The implications extend far beyond microchips. At stake is nothing less than leadership in artificial intelligence, military supremacy, and the technological foundation of 21st-century civilization. The machine in Shenzhen, built by former engineers from Dutch giant ASML who worked under aliases and were recruited with signing bonuses approaching $700,000, signals that the era of Western monopoly over advanced chipmaking may be ending.

According to a Reuters exclusive published December 18, 2025, China has completed an operational EUV lithography prototype. While the machine is currently undergoing testing and has not yet produced working chips, the Chinese government aims to create functional semiconductors using the prototype by 2028. However, industry analysts view 2030 as a more realistic target.

To understand why this development has sent shockwaves through Washington, Brussels, and Tokyo, one must first grasp what EUV lithography represents—and why Chris Miller's landmark book Chip War: The Fight for the World's Most Critical Technology describes it as perhaps the most vivid example of the scientific and engineering breakthroughs needed to sustain Moore's Law.

Modern semiconductors are marvels of precision manufacturing. The latest chips contain transistors measuring just 3 nanometers—roughly 30 silicon atoms across. Creating patterns at this scale requires light with wavelengths far shorter than anything visible to the human eye. EUV lithography uses extreme ultraviolet light at 13.5 nanometers—about 1/10,000th the width of a human hair—to etch circuit patterns onto silicon wafers with atomic-level precision.

The technical challenge of EUV lithography borders on the absurd. As Miller documents in Chip War, ASML spent nearly two decades and required close collaboration with users like TSMC and Intel before successfully commercializing the technology. The company's current flagship system, the TWINSCAN NXE:3600D, comprises more than 100,000 precision-engineered components sourced from 800+ global suppliers across multiple continents.

Each machine costs between $150 and $ 200 million, with the newest High-NA systems approaching $370 million. They weigh 180 tons—requiring three Boeing 747 cargo planes, 40 freight containers, and 20 trucks to transport. ASML staff must remain on-site at customer facilities because the machines require constant calibration and maintenance. The assembly process alone takes months in cleanroom environments operating 24 hours a day.

This is not technology that can be stolen through espionage or reverse-engineered from blueprints. Even if China obtained complete technical documentation for an ASML system, replicating it would require rebuilding an entire ecosystem of suppliers, each with decades of specialized expertise.

The heart of EUV technology lies in generating the extreme ultraviolet light itself—a process so extreme it rivals conditions found in stellar atmospheres. The system fires high-powered CO2 lasers manufactured by Germany's Trumpf at microscopic droplets of molten tin ejected at 70 meters per second inside a vacuum chamber.

Each droplet must be hit twice. The first laser pulse flattens the tin into a pancake shape. The second pulse, arriving microseconds later, strikes the flattened droplet and vaporizes it, creating a plasma with temperatures exceeding 200,000 degrees Celsius—hotter than the surface of the sun. This plasma emits EUV light at the precise 13.5-nanometer wavelength needed for lithography.

The system repeats this process 50,000 times per second with extraordinary precision. Any deviation in timing, droplet position, or laser power disrupts the entire process. The technical challenge, as Miller notes, isn't merely achieving these conditions once in a laboratory, but doing so reliably and continuously for thousands of hours without degradation.

But generating EUV light is only the beginning. Virtually all materials, including air, absorb EUV radiation, so the entire optical path must operate in a vacuum. Traditional glass lenses are entirely opaque for EUV light, rendering conventional optics infeasible.

Instead, EUV systems use multilayer mirrors—arguably the most precise objects ever manufactured by humans. These mirrors consist of alternating layers of molybdenum and silicon, each only a few nanometers thick, with more than 100 layers per mirror. The required precision is staggering: if one of these mirrors were scaled to the size of Germany, the most significant surface imperfection would measure only 0.1 millimeters.

Only one company in the world has mastered this technology: Carl Zeiss SMT of Germany. As Miller emphasizes in Chip War, Zeiss estimates that 80% of all chips made worldwide use Zeiss optics. For EUV lithography, that figure is 100%. ASML holds a 24.9% stake in Zeiss SMT, which it acquired for €1.5 billion in 2016, thereby cementing an exclusive partnership.

Creating a single Zeiss mirror can take months. The company uses techniques such as ion-beam figuring, in which individual molecules are removed from the surface to correct microscopic imperfections. The mirrors must maintain their atomic-level precision while withstanding intense EUV radiation, extreme temperatures, and the mechanical stresses of continuous operation.

This is China's critical bottleneck.

The Shenzhen prototype represents the culmination of an extraordinary mobilization launched after the Netherlands, under intense U.S. pressure, banned ASML from selling EUV systems to China in 2019. What followed was an initiative that industry insiders compare to the original Manhattan Project—not just in scale, but in the fundamental challenge of building something many experts considered impossible under the constraints China faced.

The effort centered on aggressive talent acquisition. Beginning in 2019, Chinese recruiters targeted a specific vulnerability: Chinese-born engineers working for ASML in the Netherlands and other Western semiconductor companies. The offers were designed to be irresistible—signing bonuses of 3 to 5 million yuan ($420,000 to $700,000), generous housing purchase subsidies, and the chance to participate in what was framed as a historic national mission.

Among the key recruits was Lin Nan, ASML's former head of light-source technology. His team at the Chinese Academy of Sciences' Shanghai Institute of Optics has filed eight EUV light-source patents in just 18 months—a remarkable pace that demonstrates both technical capability and strategic focus.

The Changchun Institute of Optics, Fine Mechanics and Physics (CIOMP) achieved a breakthrough in integrating extreme-ultraviolet light into the prototype's optical system, enabling it to become operational in early 2025. The institute has been offering substantial recruitment incentives for lithography experts, with some reports indicating packages approaching those offered to Silicon Valley engineers.

According to Reuters, Huawei plays a central role in coordinating a nationwide network of companies and state research institutes involving thousands of engineers. The company's involvement extends to every step from chip design to manufacturing equipment development—a level of vertical integration unprecedented in the semiconductor industry.

This coordination reflects lessons China learned from its failed attempts at semiconductor self-sufficiency in previous decades. Rather than allowing fragmented efforts by competing institutions, the current approach concentrates resources and expertise under centralized direction, much as the original Manhattan Project unified American nuclear research.

A team of around 100 recent university graduates works in a facility where each worker's desk is filmed by an individual camera to document their efforts. Their mission: reverse-engineer components from both EUV and DUV lithography machines. Components have been sourced from older ASML systems obtained through secondary markets—a gray area that falls outside export control restrictions on new equipment.

This approach mirrors the Soviet Union's strategy during the original Cold War, when Moscow systematically reverse-engineered Western technology. But as Miller documents extensively in Chip War, the Soviet semiconductor industry remained perpetually second-rate precisely because copying kept them one generation behind. The question facing China is whether they can avoid this trap.

The completion of China's prototype marks a significant milestone, but the gap between generating EUV light and mass-producing advanced semiconductors remains vast. As Miller explains in his interview about Chip War, replicating ASML's tools is "easier than it was for ASML first to produce them. But these are the most complex and precise pieces of equipment humans have ever made."

The prototype reportedly lags behind ASML's commercial systems largely because researchers have struggled to secure ultra-precision optical systems, such as those supplied by Carl Zeiss AG. Sources indicate the Chinese system uses export-restricted components from Japan's Nikon and Canon, obtained through unclear channels, but lacks the advanced Zeiss mirrors that define ASML's capabilities.

Without Zeiss-quality optics, the prototype faces fundamental limitations:

Precision: Lower-quality mirrors reduce the resolution of patterns that can be reliably etched. This suggests that China may produce 7nm chips while competitors are manufacturing at 3nm or below.

Reliability: Manufacturing requires not only achieving nanometer precision once but also maintaining it for thousands of hours of continuous operation. Commercial EUV systems demonstrate 99%+ yield rates. Early prototypes typically show much lower yields.

Throughput: ASML's current systems can process up to 200 wafers per hour. The Chinese prototype's throughput remains unknown but is almost certainly far lower, making commercial viability questionable even if the technology works.

Contamination: EUV systems operate in vacuum chambers where even microscopic particles can ruin chips. Managing contamination at the required level demands sophisticated engineering that takes years to perfect.

Miller emphasizes in Chip War that lithography tools, while critical, represent only one component of semiconductor manufacturing. China also needs:

China lags significantly in all these areas. Applied Materials, Lam Research, and Tokyo Electron dominate the deposition and etching markets. KLA Corporation controls metrology and inspection. Even if China perfects EUV lithography, building a complete fabrication facility requires mastering dozens of other technologies.

The semiconductor industry continues racing forward. ASML has begun shipping High-NA EUV systems with a numerical aperture increased from 0.33 to 0.55, enabling 8nm resolution and supporting future 2nm nodes. Each system costs approximately $370 million and represents another generation of technology.

As Miller notes, if China produces its own EUV lithography systems within five years, it will still be substantially behind the cutting edge. By the time China achieves 2025-equivalent technology, the industry will have moved to 2030-equivalent systems. This is the fundamental challenge facing any nation attempting to catch up in semiconductors: the finish line keeps moving.

China's EUV breakthrough comes amid intensifying technological warfare. On December 2, 2024, the Biden administration announced its third significant wave of semiconductor export controls aimed at China—the most comprehensive yet.

The new restrictions targeted 140 Chinese companies, including major semiconductor equipment manufacturers such as Naura Technology Group, Piotech, ACM Research, and SiCarrier Technology. The rules banned the sale of 24 types of chipmaking equipment and three software tools essential for advanced semiconductor manufacturing. For the first time, the U.S. also restricted high-bandwidth memory (HBM) chips—the specialized memory required for training large AI models.

Perhaps most significantly, the rules expanded the Foreign Direct Product Rule, reducing the threshold of U.S. content that triggers U.S. jurisdiction to zero. This means Washington can now regulate virtually any semiconductor equipment shipped to China from anywhere in the world if it contains even minimal U.S. components—a dramatic assertion of extraterritorial authority.

Commerce Secretary Gina Raimondo framed the action explicitly in terms of AI competition, emphasizing the administration's targeted approach to impairing China's ability to indigenize advanced technology production.

The Chinese government responded within 24 hours. On December 3, the Ministry of Commerce announced a complete export ban to the United States of gallium, germanium, antimony, and "superhard materials"—critical inputs for semiconductor and defense manufacturing. China supplies roughly half of the United States' gallium and germanium, raising immediate supply chain concerns.

Four major Chinese industry associations—representing semiconductors, automobiles, telecommunications, and internet companies—issued coordinated statements urging domestic firms to view U.S. chips as "no longer safe and reliable" and to diversify away from American suppliers. The China Semiconductor Industry Association stated that U.S. restrictions were disrupting supply chains and inflating costs for American companies.

This tit-for-tat escalation reflects a fundamental shift in how both nations view semiconductor technology: not as a commercial industry best left to market forces, but as a strategic asset requiring government intervention regardless of economic cost.

The semiconductor battle is the foundational layer of a broader conflict that analysts increasingly describe as an "AI Cold War"—a phrase that emerged after China published its AI Development Plan in 2017, setting the goal of becoming the global leader in artificial intelligence by 2030.

Like the original Cold War, this competition encompasses multiple dimensions: technological capabilities, economic systems, ideological frameworks, and alliance structures. Unlike that conflict, which centered on nuclear weapons and ideology, the current rivalry focuses on "accelerator technologies": artificial intelligence, advanced semiconductors, quantum computing, and biotechnology.

The stakes are existential. Advanced AI systems have applications across every domain of national power—from autonomous weapons and cyber warfare to economic competitiveness and social control. The computing power required to train frontier AI models has been doubling roughly every six months, making access to cutting-edge semiconductors increasingly critical.

OpenAI's GPT-4 reportedly required training on approximately 25,000 Nvidia A100 GPUs over several months. Google's Gemini and Anthropic's Claude models demand similar or greater computational resources. China's leading AI labs face severe constraints in accessing such hardware due to U.S. export controls on high-performance chips.

This computational bottleneck directly impacts China's ability to develop competitive AI systems. While Chinese researchers have made advances in algorithmic efficiency and model architectures, there are limits to what can be achieved without sufficient computing power. Breakthrough AI capabilities increasingly depend on "scaling laws"—increasing compute for larger models trained on more data.

As Miller documents in Chip War, the 1991 Persian Gulf War was the world's first demonstration of how modern semiconductors impact military effectiveness. American precision-guided munitions, satellite communications, and electronic warfare systems overwhelmed Iraqi forces equipped with Soviet-era technology.

Today's military applications of AI and advanced semiconductors go far beyond precision weapons:

The side that achieves a decisive advantage in these areas could fundamentally alter the military balance. U.S. officials increasingly worry that access to semiconductors provides China with a pathway to offset American conventional military superiority.

The competition has already reshaped global economic structures. The 2022 CHIPS and Science Act allocated over $52 billion to revitalize domestic U.S. semiconductor manufacturing, with significant subsidies flowing to Intel, TSMC, and Micron for new fabrication facilities in the United States. The European Union followed with its own €30 billion European Chips Act. Meanwhile, China has invested an estimated $150 billion in its semiconductor industry since 2014, with reports of an additional $70 billion package being prepared.

This massive capital mobilization represents the most significant industrial policy initiatives by Western governments in decades—a dramatic departure from the free-market orthodoxy that dominated policy since the 1980s. The fact that such interventions now command bipartisan support in Washington demonstrates how thoroughly AI and semiconductor competition have reframed strategic thinking.

As Miller observes, the next generation of High-NA EUV systems will require investments approaching $100 billion globally. The next generation is anticipated to require trillions in investment. The only way these capital quantities can be raised is through substantial government investment driven by national security concerns.

At the geographic center of this technological cold war sits Taiwan—a democratic island of 23 million people that produces approximately 70% of the world's advanced semiconductors and over 90% of the most sophisticated chips.

Taiwan Semiconductor Manufacturing Company (TSMC) is the world's most advanced chip foundry, producing processors for Apple, Nvidia, AMD, and virtually every major technology company. As Miller extensively documents in Chip War, TSMC's rise from a startup in the 1980s to global dominance represents one of the most consequential geopolitical developments of the past half-century.

TSMC's manufacturing capabilities are, in the short to medium term, irreplaceable. A single advanced fabrication facility costs $15-20 billion and requires 3-5 years to build even under optimal conditions. TSMC operates multiple fabs, each representing accumulated knowledge from decades of continuous refinement.

The company's edge comes not just from having ASML's latest equipment, but from the ecosystem of expertise surrounding it. As Miller notes, Taiwanese and Korean firms have honed specialized know-how through the construction of numerous fabs—knowledge that resides in engineers' experience rather than in documentation.

China does not recognize Taiwan's sovereignty and has never renounced the use of force to achieve "reunification." As semiconductor competition intensifies, Taiwan's strategic value has become almost incalculable.

U.S. policymakers increasingly worry that access to semiconductors could motivate Chinese military action. If China controlled TSMC—even if 80% of Taiwan's fabs were destroyed in the process—it would, overnight, add to China's semiconductor production capacity an amount equivalent to that of the entire United States.

Alternatively, China could impose a blockade or embargo, using chip access as geopolitical leverage without firing a shot. In a scenario where China controls Taiwan but leaves fabs intact, Beijing could demand exorbitant fees from democratic nations or deny access entirely to disfavored countries.

Russia's invasion of Ukraine provides an ominous preview. Despite massive Western sanctions and military support for Ukraine, Russia's economy has stabilized, and its forces continue fighting more than two years into the war. China has observed that the West's collective action, while substantial, has not broken Moscow's resolve—a lesson that may influence calculations regarding Taiwan.

Perhaps the most profound long-term consequence of the AI Cold War is the fragmentation of the global digital ecosystem—what analysts call the "splinternet" or "silicon curtain."

For decades, the semiconductor industry embodied globalization at its most sophisticated. As Miller meticulously documents, a single advanced chip might contain intellectual property from the United States, be manufactured in Taiwan using equipment from the Netherlands and Japan, incorporate materials from multiple countries, and be assembled in China for worldwide distribution.

This intricate supply chain, built on decades of collaboration and specialization, is now being deliberately severed.

The Western/Allied sphere revolves around:

This ecosystem prioritizes innovation through market competition, intellectual property protection, and alliance-based supply chain security.

The Chinese sphere is coalescing around:

This ecosystem subordinates commercial considerations to strategic imperatives, accepting higher costs and lower efficiency in exchange for technological sovereignty.

The cost of this bifurcation will be enormous. Duplication of research and development efforts, incompatible standards, restricted collaboration among researchers, and balkanized supply chains will slow innovation and increase costs globally.

Middle powers in Southeast Asia and the Middle East face intense pressure from both Washington and Beijing to adopt their respective technology stacks, often coupled with broader economic and security relationships. Indonesia, Saudi Arabia, the UAE, and numerous African nations find themselves courted as "swing states" in this technological competition.

The global semiconductor market could fragment into distinct spheres with limited interoperability—a profound reversal of the integration that has characterized the industry's growth. This represents a failure of the globalization project in one of its most successful domains.

The Chinese government has set a target to produce functional chips using the Shenzhen prototype by 2028. Industry analysts and sources close to the project, however, view 2030 as a more realistic timeline. Even this assumes China can solve challenges that took ASML and its partners decades to address.

The prototype has successfully generated 13.5nm EUV light—a significant scientific and engineering milestone. But transforming this into mass production requires solving an array of interrelated challenges:

Optical system refinement: Matching or approximating Zeiss-quality mirrors without access to their manufacturing processes and proprietary knowledge. China has reportedly developed domestic alignment interferometers capable of positioning mirrors with sub-nanometer accuracy; however, achieving the consistency and reliability of Zeiss systems remains a significant challenge.

System integration: Coordinating over 100,000 components to work in perfect harmony at nanometer-scale precision. ASML employs over 40,000 people globally and has spent two decades perfecting this integration. China is attempting to compress this timeline dramatically.

Yield and reliability: Achieving consistent chip production at commercial quality standards. ASML's current systems demonstrate 99%+ yield rates—meaning they successfully produce working chips in virtually every attempt. Early prototype systems typically show much lower yields, making them commercially unviable even if technically functional.

Software and control systems: Developing sophisticated algorithms for real-time control of every subsystem, from plasma generation to wafer positioning to pattern correction. This represents millions of lines of code refined through years of production experience.

Supply chain development: Building domestic sources for thousands of specialized components, many of which require materials and manufacturing processes currently dominated by Western and Japanese suppliers.

Customer validation: Convincing chipmakers to risk their production lines on unproven technology. Even if Chinese EUV systems become technically capable, manufacturers may hesitate, given the enormous costs of semiconductor fabrication and the risks of using equipment without an extensive track record.

If China achieves viable domestic EUV production—whether by 2028, 2030, or later—the strategic landscape will shift fundamentally.

The primary tool for constraining China's semiconductor advancement will have failed. Export controls rely on chokepoints—specific technologies that cannot be easily replicated and whose access can therefore be controlled. If the EUV chokepoint falls, the entire strategic architecture of semiconductor export controls becomes questionable.

The United States would likely shift to next-generation lithography technologies (High-NA EUV and beyond) as new chokepoints, but the principle that China can be indefinitely denied access to frontier manufacturing technology would be broken. This would represent a historic failure of technology containment policy—the most significant since the Soviet Union's successful development of nuclear weapons and space technology during the original Cold War.

ASML currently holds a 100% monopoly on EUV systems, allowing it to command premium prices and set the industry's technological pace. Chinese competition, even if initially inferior, would pressure this model.

More significantly, China prioritizes domestic chip production using indigenous equipment, reducing or eliminating purchases from Western suppliers. Given that China accounts for roughly 30% of global semiconductor manufacturing capacity, this would substantially affect growth projections for companies across the supply chain.

The semiconductor restrictions on China were explicitly designed to impede AI development. If China achieves EUV capability, it could produce the advanced chips required to train and deploy frontier AI models without relying on Western suppliers or engaging in smuggling operations.

This would accelerate China's AI development across military, economic, and surveillance applications. The scenarios U.S. officials have warned about—autonomous weapons systems, advanced cyber capabilities, pervasive facial recognition networks—would become more feasible with unfettered access to cutting-edge computing power.

Countries that have attempted to maintain technological neutrality between the U.S. and China would face reduced pressure to choose sides. If Chinese semiconductor technology approaches parity with Western systems, nations could more credibly play both sides, accepting investments and equipment from either bloc based on economic rather than strategic considerations.

Conversely, Western nations that have aligned themselves with U.S. semiconductor restrictions might question whether these costly policies achieved their intended objectives—potentially creating friction within alliances if China demonstrates that technological containment is ultimately futile.

Paradoxically, Chinese EUV capability might accelerate innovation by introducing competition into a field currently dominated by a monopoly. ASML and its supplier ecosystem have faced limited pressure to reduce costs or to accelerate development timelines.

However, this assumes continued technology transfer and collaboration—an optimistic scenario. More likely is complete bifurcation, where parallel development occurs with minimal cross-pollination, reducing overall efficiency and slowing the pace of advancement relative to a unified global industry.

Looking ahead, three broad scenarios emerge:

China is expected to achieve commercially viable EUV production by 2028-2030, thereby breaking the semiconductor chokepoint. The technological cold war enters a new phase focused on even more advanced technologies (High-NA EUV, quantum computing, advanced AI architectures).

Global technology increasingly fragments into incompatible ecosystems. Supply chains reorganize around geopolitical rather than economic considerations. Taiwan's strategic importance may diminish if China achieves semiconductor self-sufficiency, although its advanced manufacturing capabilities would remain valuable.

The balance of AI development shifts as Chinese labs gain access to unlimited computing power. Military applications of AI accelerate on both sides, creating new strategic instabilities.

China produces functional EUV systems, but they remain significantly inferior to ASML's technology in reliability, throughput, and yield. Chinese chipmakers use domestic EUV for some applications but continue relying on smuggled or stockpiled Western equipment for the highest-end production.

The U.S. maintains a meaningful technological lead while China achieves partial self-sufficiency. Competition continues across multiple technological frontiers with neither side gaining a decisive advantage. This becomes the "new normal"—a persistent technological gap that narrows but never closes.

Economic costs mount on both sides from duplicated infrastructure and restricted trade. Global innovation has slowed due to reduced collaboration and fragmented markets.

The U.S. and allies have successfully extended their technological lead by accelerating the development of next-generation lithography (High-NA EUV and post-EUV technologies), tightening controls on critical materials and components, and expanding domestic manufacturing capacity.

China's progress in EUV is insufficient to close the gap as the frontier shifts to even more complex technologies. However, this scenario requires sustained political will and massive continued investment in the face of competing priorities—a historically rare combination.

Western nations successfully build resilient supply chains that reduce dependence on any single geographic chokepoint, including Taiwan. This reduces strategic risks but entails enormous economic costs.

The completion of China's EUV prototype in Shenzhen represents a watershed moment in the technological competition that will define the 21st century. Whether China ultimately succeeds in achieving commercial EUV production by 2028, 2030, or at all, the implications are already profound.

The era of seamless global technology collaboration has ended. In its place emerges a world of rival technological ecosystems, competing standards, duplicated R&D efforts, and supply chains organized around geopolitical rather than economic considerations. The "silicon curtain" has descended.

As Chris Miller documents exhaustively in Chip War, semiconductor technology has defined international politics, economic structure, and military power for decades. The machine in Shenzhen—still imperfect, still years from commercial viability, yet operational and improving—represents the determination of a rising power to control its technological destiny, regardless of cost.

For the United States and its allies, the challenge is existential. The strategic architecture built on technological chokepoints and export controls must be fundamentally reconsidered. The question is no longer whether China can be indefinitely denied access to advanced technologies, but how quickly it will achieve them and what the balance of power will look like when it does.

For China, the path forward remains daunting. EUV lithography represents the most complex engineering challenge humanity has undertaken. Success requires not just copying Western technology but building an entire ecosystem of suppliers, developing institutional knowledge accumulated over decades, and solving problems that even the current monopoly provider spent 20 years addressing.

The Zeiss mirror problem epitomizes this challenge. Creating optical systems with atomic-level precision requires not only an understanding of the physics but also mastery of manufacturing processes refined over decades of experience. Even with complete technical documentation, replicating Zeiss's capabilities would take years.

Yet if the past decade has demonstrated anything, it is that China is willing to commit virtually unlimited resources to technological independence in areas it deems strategic. The Shenzhen prototype, built by engineers recruited with $700,000 signing bonuses and working under aliases in a facility designed to avoid Western scrutiny, embodies this commitment.

The timeline remains uncertain. Government targets point to 2028, but industry experts widely view 2030 as more realistic for commercial chip production. What is certain is that this technology—or the attempt to develop it—has already fundamentally reshaped global geopolitics.

As we approach 2030, the world watches to see whether the fundamental physics of EUV lithography—plasma temperatures hotter than the sun, mirrors polished to atomic precision, and light wavelengths measured in nanometers—can be mastered by a nation determined to break free of technological constraints.

The answer will help determine not just who leads in semiconductors or artificial intelligence, but the shape of global power itself in the century ahead. The machine in Shenzhen, massive and imperfect as it may be, represents the opening of a new chapter in this competition—one in which the assumptions that governed Western technological supremacy for the past 75 years are being tested as never before.

The AI Cold War has entered its most critical phase. And at its center sits a machine that most people have never heard of, producing light most cannot see, to manufacture chips most will never touch—yet whose impact will shape the future of human civilization.

Whether as breakthrough or boondoggle, China's EUV program has ensured that the semiconductor cold war will define international relations for decades to come.