Weekly: China's secret EUV machine
15 min read.
Highlights
China’s secret EUV machine. Three pieces on ASML, including a bombshell investigation by Reuters that reveals that China has developed a functioning EUV machine in secret. China’s lack of advanced photolithography machines is what is holding them back; the report calls this China’s ‘Manhattan Project.’ The team that is building the machine is made up of former ASML engineers. To produce this at mass scale and reliably is a different question, one that took ASML decades and billions of dollars of investment, but still, it is a major move for China’s semiconductor self-sufficiency drive.
Elsewhere, the FT also covers China’s efforts to push the older DUV machines to their technological limit, and Bloomberg does a profile of ASML CEO Christophe Fouquet as he leads the Dutch company through the AI boom.
On the H200 decision. Several more pieces on the H200 decision by the US government. A piece in CFR by a former US government official argues that Huawei is not a rising competitor to Nvidia and is in fact evidence that export controls are working. A piece in CNAS argues that the next big question is how many chips Nvidia may sell to China; many is bad, few could be better. Lastly, an op-ed in WSJ argues for sending Nvidia’s chips to China, to keep China’s AI ecosystem on the US tech stack.
FT Person of the Year: Jensen Huang. The FT has two pieces on Jensen Huang, whom they declared as FT Person of the Year. The publication reports on Jensen’s journey as engineer to Washington’s AI whisperer to great success.
Thanks for reading.
Table of Contents
Fanny Potkin, “How China built its ‘Manhattan Project’ to rival the West in AI chips,” Reuters, 12/18/2025.
Eleanor Olcott, “China boosts AI chip output by upgrading older ASML machines,” FT, 12/19/2025.
Peter Elstrom, Sarah Jacob, and Tom Mackenzie, “How ASML’s CEO Plans to Keep Pace With Soaring AI Demand,” Bloomberg, 12/12/2025.
Chris McGuire, “China’s AI Chip Deficit: Why Huawei Can’t Catch Nvidia and U.S. Export Controls Should Remain,” CFR, 12/15/2025.
Alasdair Phillips-Robins, “Don’t Panic Yet Over AI Chip Sales to China,” CNAS, 12/12/2025.
Aaron Ginn, “The Case for Sending Nvidia’s Chips to China,” WSJ, 12/16/2025.
Joe Miller, Demetri Sevastopulo, and Michael Acton, “‘Game recognises game’: How Jensen Huang won over Donald Trump,” FT, 12/14/2025.
Richard Waters and Michael Acton, “FT Person of the Year: Jensen Huang,” FT, 12/12/2025.
The Economist, “Saudi Arabia wants to host the world’s cheapest data centres,” The Economist, 12/17/2025.
1.
Fanny Potkin, “How China built its ‘Manhattan Project’ to rival the West in AI chips,” Reuters, 12/18/2025.
In a high-security Shenzhen laboratory, Chinese scientists have built what Washington has spent years trying to prevent: a prototype of a machine capable of producing the cutting-edge semiconductor chips that power artificial intelligence, smartphones and weapons central to Western military dominance, Reuters has learned.
Completed in early 2025 and now undergoing testing, the prototype fills nearly an entire factory floor. It was built by a team of former engineers from Dutch semiconductor giant ASML who reverse-engineered the company’s extreme ultraviolet lithography machines or EUVs, according to two people with knowledge of the project.
EUV machines sit at the heart of a technological Cold War. They use beams of extreme ultraviolet light to etch circuits thousands of times thinner than a human hair onto silicon wafers, currently a capability monopolized by the West. The smaller the circuits, the more powerful the chips.
China’s machine is operational and successfully generating extreme ultraviolet light, but has not yet produced working chips, the people said.
In April, ASML CEO Christophe Fouquet said that China would need “many, many years” to develop such technology. But the existence of this prototype, reported by Reuters for the first time, suggests China may be years closer to achieving semiconductor independence than analysts anticipated.
Nevertheless, China still faces major technical challenges, particularly in replicating the precision optical systems that Western suppliers produce.
The availability of parts from older ASML machines on secondary markets has allowed China to build a domestic prototype, with the government setting a goal of producing working chips on the prototype by 2028, according to the two people.
But those close to the project say a more realistic target is 2030, which is still years earlier than the decade that analysts believed it would take China to match the West on chips.
The people described it as China’s version of the Manhattan Project, the U.S. wartime effort to develop the atomic bomb.
ASML built its first working prototype of EUV technology in 2001, and told Reuters it took nearly two decades and billions of euros in R&D spending before it produced its first commercially-available chips in 2019.
2.
Eleanor Olcott, “China boosts AI chip output by upgrading older ASML machines,” FT, 12/19/2025.
China’s semiconductor manufacturers are upgrading their advanced chipmaking equipment in ways that bypass global export controls, as the country seeks to rival the US in developing artificial intelligence.
According to people familiar with the matter, Chinese fabrication plants producing advanced smartphone and AI chips have bolstered the performance of advanced deep ultraviolet lithography (DUV) machines made by Netherlands-based ASML.
US and Dutch export controls prevent ASML from supplying its most advanced DUV machines to China, leaving many Chinese fabs to rely on older equipment — notably the Twinscan NXT:1980i system — to manufacture the seven-nanometre chips needed to develop AI systems.
According to those familiar with the techniques, Chinese fabs have obtained components on the secondary market. This includes an upgraded “stage”, a mechanical platform for the silicon wafer, as well as lenses and sensors that help ensure that chip layers are aligned with greater precision.
These improvements to ASML’s DUVs have enabled Chinese fabs to bolster their AI chip production.
3.
Peter Elstrom, Sarah Jacob, and Tom Mackenzie, “How ASML’s CEO Plans to Keep Pace With Soaring AI Demand,” Bloomberg, 12/12/2025.
When Christophe Fouquet interviewed for a job at ASML Holding NV in 2007, he had an unusual request: Could he take a position one rung below what the Dutch company was offering? He wanted to study the technical details of the new chip-making machines ASML sold to tech giants like Intel Corp. and Samsung Electronics Co. After he was hired the next year, he spent weeks poring over the product catalog until he could recite the key features by heart.
Today, much depends on Fouquet and his company’s ability to navigate technology’s frontiers. ASML is the foundation on which much of the AI boom – indeed much of the tech industry – has been built over the past few years. It manufactures the machines needed to produce the most advanced chips for Nvidia Corp., which in turn run the artificial intelligence models for OpenAI, Microsoft Corp. and pretty much every other competitor in the field. Its share at the premium end of this market is a cool 100%, higher than Nvidia’s for AI chips or OpenAI’s for chatbots.
ASML is riding two seismic trends in the tech industry. Demand for semiconductors has grown steadily as chips are integrated into automobiles, consumer electronics and other products beyond computers and smartphones. Then, with the debut of ChatGPT in 2022, tech giants have accelerated efforts to build AI data centers packed with the most advanced processors and memory chips, making ASML’s highest-end gear even more vital. The global semiconductor market is projected to rise 22% to $772 billion this year and more than 25% to $975 billion next year, according to the World Semiconductor Trade Statistics organization.
Chipmakers are also diversifying geographically as governments grow concerned about the concentration of production in Taiwan and South Korea. The US government, for example, is offering billions of dollars in incentives to Taiwan Semiconductor Manufacturing Co., Samsung and others to build fabrication facilities, or fabs, on American soil. The more fabs customers build, the more ASML machines they need.
His next big test is whether he can lead a transition from what’s known as extreme ultraviolet lithography, or EUV, technology to high numerical aperture, or High NA, EUV. ASML introduced EUV machines — and is still the only company that can make them — which helped customers move from 7-nanometer chips down to the 3-nm chips used by Nvidia and Apple. High NA EUV machines are aimed at pushing those geometries below 2-nm.
ASML will work with its customers on getting the High NA machines to the point they can operate with minimal downtime through next year, and Fouquet expects high-volume manufacturing in 2027 and 2028. Sometime next decade, ASML will introduce an even more advanced technology called Hyper NA. Research for that step has already begun.
4.
Chris McGuire, “China’s AI Chip Deficit: Why Huawei Can’t Catch Nvidia and U.S. Export Controls Should Remain,” CFR, 12/15/2025.
On December 8, the Trump administration announced plans to loosen U.S. export controls on artificial intelligence (AI) chips to China by approving the sale of Nvidia H200 chips—the most powerful AI chip ever approved for export to China. That decision was driven in part by concerns that Huawei is becoming a viable competitor to Nvidia in AI chips, making U.S. export controls less effective. However, a comparison of publicly available data on AI chip performance from both companies, coupled with estimates on AI chip production capacity finds something different: Huawei is not a rising competitor. Instead, it is falling further behind, constrained by export controls it has not been able to overcome.
Nvidia and Huawei’s AI chip roadmaps from this year show that the performance gap between U.S. and Chinese AI chips is large and growing. The best U.S. AI chips are currently about five times more powerful than Huawei’s best offerings. By 2027, that gap will widen to seventeen times. Perhaps most striking: according to Huawei’s own public roadmap, the company’s next-generation chip in 2026 will actually be less powerful than its best chip today. This apparent regression could indicate that SMIC and other Chinese fabs are struggling to produce high-performing AI chips for Huawei at scale. With SMIC stuck at 7nm process technology due to U.S. and allied equipment export controls, Huawei has hit a ceiling it is struggling to break through.
Huawei is not a threat that justifies loosening controls; it is evidence that the controls are working.
5.
Alasdair Phillips-Robins, “Don’t Panic Yet Over AI Chip Sales to China,” CNAS, 12/12/2025.
Selling H200s is, indeed, a sharp break in U.S. strategy, but how consequential the move will be depends on what the administration does next. The more important decision—and one that the administration has seemingly yet to make—is how many of the chips to sell. If the Trump team allows China to buy millions, it will risk upending the AI race. But if it limits the numbers, and uses each shipment as leverage, the effects will be far less dramatic.
Taking the latter course would mean repeating the strategy the administration has adopted in AI sales to the Middle East—a big announcement followed by a slow drip of exports. If it does that, Washington may manage to boost Nvidia’s market share in China without conceding America’s overall compute advantage. Of course, selling any number of H200s will bring risks, including alienating crucial allies, and Trump seems to have made his announcement without winning anything from China in return. But as so often with Trump, the reality may be less dramatic than the immediate fireworks suggest.
6.
Aaron Ginn, “The Case for Sending Nvidia’s Chips to China,” WSJ, 12/16/2025.
Your editorial “Trump Says Chips Ahoy to Xi Jinping” (Dec. 11) asks why the president would allow Nvidia to sell advanced chips to China, as if the choice were between guarding a fortress and handing over the keys. The question instead is whether we benefit from forcing Beijing to develop an independent AI ecosystem or by keeping the global one anchored to American platforms.
Critics argue that selling chips risks helping China close the technological gap. Yet banning GPU exports accelerates Beijing’s move toward alternatives. Every blocked Nvidia sale sends a demand signal for local Chinese options, supported by subsidies and geopolitical pressures. Huawei’s rise as a scaled competitor happened despite U.S. export controls.
Nor do those restrictions resolve U.S. companies’ main challenge: power generation and data-center capacity. The best way to solve our shortage is to build more infrastructure, not restrict the market for U.S. products. Nvidia’s ability to fund next-generation designs and expand domestic capacity depends on scale and global adoption. Undermining that reach weakens the industrial base and our ability to lead.
China’s objective is clear: a sovereign AI stack immune to American pressure. Let’s not give them what they want. The most significant risk to our leadership isn’t that China buys American chips. It’s that we convince ourselves that we can preserve our dominance by saying “no,” rather than by ensuring the world’s most important technology continues to run on American platforms.
7.
Joe Miller, Demetri Sevastopulo, and Michael Acton, “‘Game recognises game’: How Jensen Huang won over Donald Trump,” FT, 12/14/2025.
“I think game recognises game,” said a person familiar with the company’s strategy, of the president’s newfound fondness for Huang. “The way Trump wants to control the federal government is effectively the way that Jensen runs Nvidia. There are no fiefdoms . . . and Jensen’s instincts kind of reign.”
The $4tn company’s success in courting the president is especially remarkable because Nvidia until recently had a threadbare lobbying operation in Washington. His early access to the president was brokered by Howard Lutnick, the commerce secretary.
But the company, which sells the advanced chips that power sophisticated AI models, was drawn deeper into politics when the White House restricted the sale of its H20 chips to China — as part of Trump’s wider trade conflict with Beijing.
Understanding that the president wanted companies to commit to expanding manufacturing in the US, Nvidia soon joined a consortium that has pledged to invest half a trillion dollars domestically over the next four years.
Huang in April flew to Mar-a-Lago to talk to Trump on the sidelines of a $1mn-per-head dinner. The administration softened its stance in the following months.
As well as meeting Trump privately at least six times this year and speaking to him directly on the phone, Huang accompanied the president to the United Arab Emirates, Saudi Arabia and the UK.
In October, Huang contributed to the president’s ballroom project.
The Nvidia CEO simultaneously began courting lawmakers. Huang made the case that blocking US technology from Chinese AI developers would not stop their advances but would encourage China’s own chipmakers to catch up.
The company’s advocacy on Capitol Hill was led by Tim Teter, an intellectual property lawyer who as the company’s top legal executive has become one of Huang’s most trusted advisers.
Unlike many of its competitors, Nvidia has made its case directly, largely eschewing established lobbyists and industry associations. It rapidly built out an in-house team, hiring a Republican lobbyist who had worked for Ivanka Trump.
8.
Richard Waters and Michael Acton, “FT Person of the Year: Jensen Huang,” FT, 12/12/2025.
On his birthday in February, Nvidia chief executive Jensen Huang and his wife were celebrating at home with cake when he noticed several missed calls from an unknown number.
The tech boss at the heart of the artificial intelligence boom was about to ignore them when his phone rang again. “Hello Jensen: this is President Trump,” the voice said.
As Huang’s dogs started to bark, drowning out the conversation, he initially thought it was a prank: “I said, ‘Really, is this President Trump?’” he tells the FT in an interview. The impromptu call turned into a 45-minute chat.
For a tech executive who for most of his career had been little known outside Silicon Valley, where Nvidia’s chips were not even considered part of the technology mainstream, the call marked the start of an unusual relationship between Trump and the self-made executive, who is the son of Taiwanese immigrants.
Huang’s move into the corridors of power bore fruit this week with the announcement that Nvidia will once again be allowed to sell advanced AI chips in China, something that has been opposed by many in the national security establishment.
Chips were long the unloved stepchildren of the digital world, largely overlooked in favour of the devices they powered and the services that ran on them. But this year, they have emerged as the impetus behind the AI mania that is sweeping through the business and financial worlds.
The FT is naming Huang as its person of the year because of the role he has played in this transformation. Huang has been at the centre of one of the biggest investment programmes ever conducted by the private sector — one that has both propped up the US economy and sustained a stock market boom. And he has been a driving force in the adoption of a technology that has the capacity to reshape entire industries.
Nvidia is now the most valuable company in the world and, at one stage over the course of the year, it became the first to have a market capitalisation of more than $5tn (On Thursday evening, it was valued at $4.4tn.) Huang himself is set to end 2025 with a net worth of more than $160bn, putting him in the ranks of the world’s 10 richest people. Even if it turns out that current valuations are inflated and the share price were to fall by a half, Nvidia would still be worth three times more than at the end of 2021.
9.
The Economist, “Saudi Arabia wants to host the world’s cheapest data centres,” The Economist, 12/17/2025.
Two hours south of Jeddah, on Saudi Arabia’s Red Sea coast, the Al Shuaiba solar farm blankets 50 square kilometres of desert. The first phase of the project, started in 2024, produces 600 megawatts of electricity at just 3.9 Saudi halalas (just over a cent) per kilowatt-hour, nearly a twentieth of the cost of generation at Britain’s planned Hinkley Point C nuclear power plant. Saudi Arabia’s plan for all this cheap electricity is to power enormous data centres for artificial intelligence (AI).
The cost of inference, the process of querying and getting answers from an AI system, is made up of two things—the fixed cost of computer hardware, and the ongoing cost of the electricity to run it. Cutting corners on hardware is a false economy, since the newest and most expensive chips are usually more efficient at running the best algorithms. Offering cheaper AI systems, therefore, comes down to using cheaper electricity. On that, Saudi Arabia reckons it has the edge.
This strategy became a national priority in May and is backed by the state’s defacto ruler, Muhammad bin Salman, known as MBS. A new company, Humain, has centralised the efforts under the leadership of Tareq Amin, boss of Aramco Digital, the tech arm of the state-owned energy company. “We hit the ground sprinting, not just walking,” Mr Amin says.
Humain’s mission is wrapped up in Saudi Arabia’s wider “Vision 2030” strategy, a goal for pivoting the country away from its dependence on extracting fossil fuels. Executing the overall vision within the constraint available is “the number one risk”, says Mr Amin. “We have no choice. We have to do this, there is no plan B.” Born in Jordan, Mr Amin has taken on big challenges before, having worked on infrastructure projects for Reliance Jio, an Indian telecoms company, and Rakuten, a Japanese conglomerate.
-