Tuesday, December 30, 2025

20251231 ■ Asia_Times-20251219-Made-in-China EUV machine targets AI chip output by 2028【by Jeff Pao】

 20251231 ■ Asia_Times-20251219-Made-in-China EUV machine targets AI chip output by 2028【by Jeff Pao】 

URL https://asiatimes.com/2025/12/made-in-china-euv-machine-targets-ai-chip-output-by-2028/

Asia Times

Technology

Made-in-China EUV machine targets AI chip output by 2028

State-backed push likened to Manhattan Project, highlighting ambition, hurdles and long timelines of China’s EUV lithography quest

by Jeff Pao

December 19, 2025

Chinese scientists are using alternative methods to build extreme ultraviolet (EUV) lithography machines. Credit: ASML

China has reportedly built an extreme‑ultraviolet (EUV) lithography machine inside a high‑security laboratory in Shenzhen, in what sources described as a nationally coordinated push to overcome the most tightly held chokepoint in advanced chipmaking.

The machine is described as operational and capable of generating EUV light, though it has yet to produce functional chips, Reuters reported. Sources said Beijing is targeting 2028 for chip production, though they suggested that 2030 is a more realistic milestone.

The report said Huawei is playing a central role, coordinating a web of companies and state research institutes across the country involving thousands of engineers. Sources compared the effort to China’s version of the Manhattan Project, the top‑secret US wartime program that mobilized scientists, industry and the state to develop the atomic bomb in 1942-1947.

Sources told Reuters that the prototype EUV machine was completed in early 2025 and is now undergoing testing. The report said the machine fills nearly an entire factory floor and was assembled by a team that included former engineers from Dutch semiconductor equipment maker ASML.

According to Chinese media, two teams of scientists are working in parallel to develop EUV light sources for high-end chip production.

One team is led by Lin Nan, a former ASML engineer who is currently a professor at the School of Integrated Circuit Science and Engineering at Beihang University. Zhao Yongpeng, a professor at Harbin Institute of Technology, leads the other team. Both teams use solid-state lasers to heat and vaporize tin droplets for EUV light generation, while ASML uses CO2 lasers supplied by its US-based subsidiary, Cymer.

It is unclear whether Lin is in charge of the top-secret laboratory in Shenzhen. Before this, some Chinese media had reported in March this year that Huawei was testing a customized EUV machine at a factory in Dongguan, Guangdong.

They said Harbin Institute of Technology was responsible for the light source, the Changchun Institute of Optics for optical systems, and Shanghai Micro Electronics Equipment Group (SMEE) for overall integration.

Huawei and Shenzhen state-backed firm SiCarrier were also deeply involved, coordinating more than 3,000 researchers across lithography, deposition and etching equipment development.

The Global Times, a newspaper aligned with the Chinese Communist Party, on Friday published a commentary titled “There is no need for Reuters to be anxious about China’s technological progress.”

The article argued that the Reuters report, which relied heavily on unnamed sources, reflected Western unease rather than facts, portraying lithography machines as a supposed final stronghold of Western technological dominance.

However, tt did not deny the existence of a high-security EUV research laboratory in Guangdong. The state-aligned newspaper said China has long pursued domestic alternatives to imported lithography tools, citing officially disclosed progress in deep ultraviolet (DUV) machines, and described Beijing’s technology strategy as one of self-reliance paired with continued openness to international cooperation.

The commentary said export controls had not slowed China’s technological advance, but instead spurred domestic innovation. It argued that China’s breakthroughs would benefit humanity and coexist with openness and cooperation, warning that efforts to block China could fragment global technology supply chains.

China’s ongoing research

After the United States moved in 2019 to block sales of ASML’s extreme ultraviolet lithography machines to China, Beijing began pouring substantial resources into domestic lithography development. However, parts of the effort have been weakened by inefficiency and corruption scandals in the semiconductor sector.

Lin’s own career trajectory illustrates both the talent Beijing has sought to mobilize and the opacity surrounding the EUV project.

In April 2021, Lin left the Netherlands, where he had worked as a scientist at ASML Research since October 2015, to take up a post as deputy director at the State Key Laboratory of Ultra-Intense Laser Science and Technology at the Shanghai Institute of Optics and Fine Mechanics, Chinese Academy of Sciences (CAS). In October this year, he joined Beihang University.

“We have experienced unauthorized misappropriation of data relating to proprietary technology by a (now) former employee in China,” ASML said in its 2022 annual report. “We promptly initiated a comprehensive internal review. Based upon our initial findings, we do not believe that the misappropriation is material to our business.”

The company added that the incident may have involved violations of export control regulations and said it had reported the matter to relevant authorities.

There is no evidence that Lin was the former employee referenced in ASML’s disclosure. Publicly available information also does not indicate that Lin relocated to Shenzhen or has any formal connection with Huawei’s reported EUV program.

However, his current research clearly builds on the expertise he has accumulated over more than five years at ASML’s research division, particularly in laser-based EUV light generation. Lin, a CCP member, was a student of Anne l’Huillier, the 2023 Nobel Prize laureate for Physics.

In December 2024, Lin and his team published an academic paper reporting a conversion efficiency (CE) of 3.42% in an experiment aimed at generating EUV light. The paper was published in March 2025 by the Chinese Laser Press, an academic journal.

The paper noted that even the Advanced Research Center for Nanolithography (ARCNL) achieved only 3.2% CE in a comparable experiment in 2019. Higher conversion efficiency means less energy is required to produce usable EUV light, making it easier to transition from laboratory experiments to practical lithography systems.

Lin’s team said, theoretically, the CE would reach 6% in the future, meeting the commercial standard of 5.5%.

According to the paper, the team used a Nd:YAG (neodymium-doped yttrium aluminium garnet) 1,064-nanometer laser with a maximum pulse energy of 600 millijoules to heat and vaporize molten tin droplets, producing EUV radiation. The laser device’s power is similar to that used for tattoo removal and fungal nail treatment.

A Hebei-based columnist surnamed Li addresses the question of why Lin’s team used a solid-state laser rather than replicating ASML’s approach of using a high-power CO₂ laser.

“Solid-state lasers are already widely deployed in industrial applications in China, which accounts for about 34% of global patent filings in the field,” he says. “Choosing this technological route allows researchers to build on existing industrial strengths while avoiding patent barriers linked to CO₂ laser systems, lowering the cost and risk of technology transfer.”

Zhao’s research team is reportedly using a solid-state laser to vaporize tin droplets, but with an added step that accelerates plasma formation by passing high-voltage electricity through an electrode disk. The approach, known as laser-assisted discharge-produced plasma (LDP), is designed to produce a stable, reliable EUV light source.

According to Chinese media, the EUV output of the Harbin system is around 100 watts, still well below ASML’s laser-produced plasma (LPP) process, which can deliver close to 600 watts of EUV power.

Some analysts noted that, due to underlying physics, Zhao’s LDP-based systems are challenging to scale to much higher power levels. This is why LDP is usually not used for chip lithography, but for applications such as photomask defect inspection and photoresist outgassing tests. ■ 

Read: China reportedly caught reverse-engineering ASML’s DUV lithography

Follow Jeff Pao on Twitter at @jeffpao3






20251231■ Asia_Times-20251230-China’s Nvidia snub reveals the price of US chip controls.

 ASIA TIMES 

20251231-Asia_Times-20251230-China’s Nvidia snub reveals the price of US chip controls.


China’s Nvidia snub reveals the price of US chip controls

US export controls designed to contain China are accelerating, not stopping, its AI ambitions and chip-making prowess


by Imran Khalid

December 30, 2025 

China has been less than enthusiastic about Donald Trump's Nvidia H-200 chip easing. Image: YouTube Screengrab

In the push and pull of contemporary US-China relations, few moves reveal more about shifting power balances than the quiet rejection of a proffered gift. On December 8, President Donald Trump announced that the United States would permit Nvidia to export its H200 artificial intelligence chips to China, replete with a 25% fee to the US Treasury.

This was billed as a pragmatic compromise: a way to keep American firms competitive in the world’s largest AI market while denying Beijing access to even more advanced models like the Blackwell series. Trump framed it as a win for national security and economic leverage, allowing sales only to “approved customers” vetted by the US Commerce Department.

Yet Beijing’s response has been cautious and calculated. Regulators convened emergency meetings with major firms, discussed potential limits on access such as requiring approvals and justifications for purchases over local alternatives and signaled a preference for domestic options, exposing the challenges in Washington’s tech diplomacy.

Far from drawing China closer, the gesture appears to have reinforced its drive toward technological independence, turning what might have been a concession into further impetus for self-reliance.

The backdrop to this episode is a trade war that has evolved from blunt tariffs into a sophisticated battle over supply chains and innovation. Earlier this year, Trump imposed a 20% “fentanyl tariff” on Chinese goods to pressure Beijing on opioid precursors, later halved to 10% in October after talks in Busan yielded promises of stricter export controls on those chemicals.

China, in turn, paused new restrictions on rare earth minerals – vital for producing everything from electric vehicles to fighter jets – for a year, while issuing general export licenses for materials like gallium and germanium.

These steps, alongside resumed purchases of US-grown soybeans, suggested a fragile detente. But the chip decision reveals deeper fault lines. US export controls, tightened since 2022, have long aimed to starve China’s AI ambitions by limiting access to high-performance semiconductors.

Nvidia, once reliant on China for up to a quarter of its revenue, adapted by designing downgraded versions like the H20. Even those faced bans in September, forcing firms such as Tencent and ByteDance to pivot to domestic options from Chinese tech giants Huawei and Alibaba.

Trump’s H200 approval was meant to thread the needle: unlock sales worth billions for Nvidia, fund US priorities through the surcharge and potentially extract further concessions on fentanyl or trade imbalances.

Nvidia’s chief executive, Jensen Huang, met with Trump in early December, expressing cautious optimism about resuming shipments. Yet reports indicate Beijing is weighing limited access to the H200, potentially through vetted uses and prioritizing local providers, while private demand from Chinese firms remains robust, highlighting how export controls continue to carry backfire risks.

Rather than crippling Chinese innovation, the controls have fostered ingenuity. Companies are optimizing algorithms to squeeze more performance from restricted hardware, reducing the need for cutting-edge imports. This is not mere defiance; it is a calculated strategy to build an ecosystem resilient to external pressure.

Consider DeepSeek, the Hangzhou-based AI startup that has emerged as a quiet disruptor. Founded in 2023 by hedge fund veteran Liang Wenfeng, DeepSeek has released models claiming strong performance relative to global leaders, trained efficiently on restricted hardware at low cost through innovations like mixture-of-experts architectures.

Backed by state-linked funds, the firm exemplifies how restrictions have spurred efficient, adaptive AI development that appeals worldwide, including in the Global South.

This pivot extends beyond startups. Huawei’s latest chips now power AI training at scales rivaling Nvidia’s older generations, bolstered by SMIC’s advances in mass production. Beijing’s Politburo, in its December 8 economic meeting, reiterated calls for “core technology breakthroughs,” channeling billions into semiconductors.

The result is a domestic market where Nvidia’s share has plummeted and firms such as Alibaba report seamless transitions to homegrown alternatives. Analysts note the irony: Trump’s “art of the deal” dangled chips amid ongoing debates over China’s acceptance.

For the US, the implications are stark. Export controls were sold as a firewall against Chinese military AI and economic espionage, yet they risk isolating American firms instead.

The US Congress, already fractious, saw bipartisan senators introduce a “Safe Chips” bill on December 4 to block such easing for 30 months, citing security risks. But blocking sales outright could accelerate Huawei’s global rise, as seen in its growing footprint in Europe and Africa.

Trump’s approach – dangling tech carrots amid tariff sticks – assumes Beijing values access over autonomy. History suggests otherwise. China’s 2010 rare earth embargo on Japan spurred Tokyo’s diversification, much as today’s chip curbs are doing for semiconductors.

Beijing, meanwhile, is treading carefully. Rejecting the H200 outright could invite Trump’s retaliation, like renewed fentanyl tariffs or entity list expansions. Instead, it has opted for graduated restrictions, buying time to scale domestic production.

This mirrors President Xi Jinping’s doctrine of “dual circulation”: fortifying internal markets while engaging globally on more equal terms. By framing US tech as a security risk and echoing Washington’s own rhetoric, China justifies its barriers without alienating partners.

The broader geopolitical lesson is one of unintended momentum. US controls have bought time, slowing China’s AI ascent by perhaps two years, per certain think tank estimates. But they have also seeded a rival innovation engine, leaner and more adaptive.

As the H200 situation remains fluid, with potential for limited sales amid China’s self-reliance push, the US faces a choice: double down on isolation, ceding moral high ground in the global tech race, or recalibrate toward alliances that include Europe and partners such as Australia in setting joint standards.

Trump’s H200 gambit, while perhaps well intentioned, highlights the peril of treating technology as a unilateral lever. In the end, the real masterstroke belongs to Beijing, which has turned constraint into capability, one optimized algorithm at a time.