EUV Lithography Enters the Mainstream as New Players and Alternatives Emerge

EUV, once accessible only to TSMC, Samsung, and Intel, is now appearing on the expansion lists of Nanya Technology and Winbond. Meanwhile, a group of startups founded just two or three years ago are attempting to bypass traditional EUV lithography machines using X-rays, particle accelerators, and even helium atoms. Recently, we reported that Elon Musk appears to be considering the mass production of FELs. For details, see the article “Is Musk Set to Disrupt EUV Lithography?”

Thus, it seems that almost everyone has suddenly become aware of and able to use the EUV era. Is the barrier to EUV really disappearing?

Currently, the players that have truly achieved mass production of EUV are still highly concentrated, primarily consisting of TSMC, Samsung, Intel, SK Hynix, and Micron.

Among them, TSMC is one of the earliest and largest commercial adopters of EUV. In 2019, TSMC officially introduced EUV into its N7+ process and achieved commercial mass production. Subsequent advanced nodes such as N5 and N3 further increased the number of EUV exposure layers, and EUV has gradually evolved from an initial tool used for only a few critical layers to a core component in advanced logic manufacturing.

Samsung Electronics is advancing EUV on both its advanced logic and DRAM product lines. In addition to advanced process foundry services, Samsung has already implemented EUV in the mass production of 14nm and 12nm-class DRAM, extending EUV from advanced logic into memory chip manufacturing.

Intel's adoption of EUV came relatively late, but its progress has been rapid. Intel 4 became its first process to widely adopt EUV, with subsequent nodes, Intel 3 and 18A, further expanding its application. Meanwhile, Intel is among the first chipmakers to receive and deploy ASML’s High-NA EUV equipment, enabling early technology validation for next-generation processes such as 14A.

Among memory manufacturers, SK Hynix began incorporating EUV into 1a-nanometer DRAM mass production in 2021 and has since expanded its usage. In 2025, the company installed High-NA EUV equipment at its M16 wafer fab in Korea, bringing next-generation EUV technology forward into DRAM R&D and future mass production systems.

Micron was the last of the three major DRAM manufacturers to officially adopt EUV. It first implemented EUV at scale on the 1γ (1-gamma) DRAM node and will begin mass production of related products in Taiwan and Japan starting in 2025.

Therefore, although EUV has been developing for many years, the players in actual high-volume manufacturing remain highly concentrated among these five companies. For a considerable period in the past, whether a semiconductor manufacturer could enter this “five-member club” was, to some extent, an indicator of its capability to produce at the most advanced process nodes.

But now, this long-closed "five-person club" has begun to welcome new entrants.

The most prominent example is Japan's Rapidus.

Rapidus was founded in 2022 and, compared to giants like TSMC, Samsung, and Intel with decades of advanced process expertise, can almost be called a “new foundry.” Yet, this company, established only a few years ago, has already entered the EUV era.

In December 2024, the ASML NXE:3800E EUV lithography machine arrived at Rapidus’s IIM-1 wafer fab in Chitose, Hokkaido. Rapidus stated that this is Japan’s first EUV exposure tool intended for advanced chip mass production. In April 2025, the company completed its first EUV exposure and began using the equipment for trial production and validation of 2nm GAA processes, with the goal of achieving mass production of 2nm chips by 2027.

What sets Rapidus apart is that it almost bypassed the traditional semiconductor foundry path of gradually upgrading from mature to advanced nodes before introducing EUV, instead building its production system from the ground up around 2nm, GAA, and EUV technology.

This also highlights one thing: EUV is no longer exclusive to traditional semiconductor giants with decades of advanced process expertise. With sufficient funding, technological collaboration, and industry support, a newly established wafer fab can directly acquire EUV equipment and start at the most advanced node.

However, Rapidus also noted that EUV is still far from being truly "democratized."

Rapidus is not a typical startup independently spending hundreds of millions of dollars to purchase an EUV lithography machine. It is backed by substantial, ongoing financial support from the Japanese government, a consortium of major Japanese corporations including Toyota, Sony, NTT, NEC, and SoftBank, and a partnership with IBM to introduce 2nm GAA process technology.

In other words, Rapidus was able to directly enter the EUV era by leveraging state capital, industry alliances, overseas technology transfer, and advanced equipment procurement capabilities to establish a new player in advanced semiconductor manufacturing.

However, the current situation is that secondary-tier wafer fabs are also beginning to purchase EUV equipment.

On August 5, 2026, Nanya Technology’s board announced an increase in its 2026 capital expenditure from NT$52 billion to NT$69.7 billion, representing a growth of over 30%. At the same time, the company approved a capital expenditure ceiling of NT$346.6 billion for the 5A new factory from 2026 to 2029, making it one of the largest single investment plans in Nanya Technology’s history. According to Nanya Technology’s currently disclosed plans, the first phase of the 5A new factory aims for a monthly wafer start capacity of approximately 35,900 wafers, with gradual introduction of 10-nanometer-class DRAM processes including 1B, 1C, 1D, and 1E. As process technology continues to scale down, EUV will become one of the key equipment components in the next-generation DRAM production system.

Meanwhile, on August 6, 2026, Winbond announced the initiation of the construction plan for its second 12-inch wafer fab, P2, in Kaohsiung. According to the plan, construction of the new facility will begin in January 2027, equipment installation will start in early 2029, and mass production is scheduled to commence in the fourth quarter of 2029. Regarding its process roadmap, Winbond has outlined a clear strategy: in the first phase, it will produce 14-nanometer-class DRAM using non-EUV technology, followed by the official adoption of EUV in the second phase to advance toward 12-nanometer-class DRAM.

Notably, Winbond has already begun planning ahead in response to the lengthy lead times for EUV equipment. Company management revealed that the current delivery cycle for EUV equipment is approximately three to three and a half years, meaning that to implement EUV around 2029, equipment, facility, and process planning must begin now.

The entry of Winbond is more noteworthy in terms of industry signals compared to South Asia Technology. This is because Winbond has not traditionally been a major player in the HBM or advanced general-purpose DRAM markets; its core business has primarily focused on niche DRAM, automotive storage, industrial control, and Code Storage Flash markets.

In the past, a key precondition for EUV to justify its extremely high equipment and process costs was that it primarily served high-value chips such as CPUs, GPUs, advanced SoCs, and premium DRAM. However, when a manufacturer long focused on niche storage solutions begins planning EUV for 12-nanometer-class DRAM, it signals that the economic boundary of EUV is shifting.

1) Technical requirement: EUV is expanding from logic processes to DRAM.

From TSMC, Samsung, and Intel to SK Hynix and Micron, EUV initially addressed the question of how to continue advancing the most cutting-edge chips; now, with Nanya Technology and Winbond entering the scene, EUV is gradually taking on another role—as a foundational production tool that must be considered for next-generation memory processes. As advanced memory technologies themselves approach the economic limits of traditional DUV multi-patterning, continuing to push to smaller dimensions using 193nm immersion lithography means more exposures, more masks, and more etching and deposition steps, along with more complex overlay control and longer production cycles. Once process complexity reaches a certain threshold, a costly EUV tool may actually become more economical than repeatedly stacking DUV multi-exposure steps.

2) Business Driver: The AI boom makes EUV affordable for secondary memory manufacturers.

In the past, second-tier manufacturers hesitated to purchase EUV lithography machines because niche and mature DRAM prices fluctuated sharply and profit margins were thin, making it difficult to amortize the enormous depreciation of a machine costing over $100 million. After acquiring a lithography tool, they also needed to invest in supporting infrastructure—such as cleanrooms, masks, photoresist, inspection and metrology equipment, process development, and long-term maintenance. For a second-tier memory manufacturer with clearly cyclical product pricing and consistently lower gross margins than industry leaders, this was not an easy investment decision.

Especially during the past several storage downturn cycles, DRAM prices fell rapidly and capacity utilization declined, prompting manufacturers to prioritize cutting capital expenditures over purchasing the most expensive semiconductor equipment.

AI has transformed this investment logic. Samsung, SK Hynix, and Micron have shifted a significant portion of their general-purpose DRAM production capacity to HBM manufacturing, causing severe supply constraints in the traditional and niche DRAM markets and driving prices sharply higher. To secure long-term supply for edge AI, automotive MCU/SoC, industrial control, and AI server auxiliary memory, customers have proactively signed 3- to 5-year long-term agreements (LTAs) with secondary suppliers. Higher product gross margins combined with stable, long-term capacity utilization have given niche players like Nanya Technology and Winbond strong cash flow and predictable capital expenditure planning, finally enabling them to overcome the financial barrier to adopting EUV technology.

This is actually a very important change: EUV hasn't suddenly become cheaper—it's that memory chips have become more valuable.

In the past, a secondary memory manufacturer might have asked, “Do we qualify to purchase EUV?” Now, they need to consider more urgently: “If we don’t buy EUV, can we still maintain cost competitiveness in five years?”

3) Tool attribute change: Low-NA EUV is becoming a relatively mature standard equipment.

Another easily overlooked reason is that EUV tools themselves have evolved. The first-generation 0.33 NA EUV systems, such as the NXE:3400C, 3600D, and 3800E, have undergone nearly a decade of mass production, achieving commercial maturity with source power, mask pellicle lifespan, and line availability all reaching 90%–95%.

The sales trend of ASML in recent years clearly shows this. 2019 was the pivotal year when EUV truly entered high-volume manufacturing, with ASML selling 26 EUV systems that year; this increased to 31 in 2020 and further rose to 42 in 2021. In 2022, revenue was recognized for 40 systems, and in 2023, it reached 53. Although 2024 and 2025 figures of 44 and 48 systems, respectively, are influenced by customer capital expenditure cycles, equipment acceptance timelines, and the rollout pace of High-NA EUV, they are not a simple linear year-over-year increase, EUV has now firmly entered a phase of annual deliveries at a scale of dozens of systems.

Moreover, ASML has clearly stated that by 2026, the company aims to achieve an annual production capacity of at least 60 Low-NA EUV systems, with plans to further increase this to at least 80 units by 2027. At the same time, the company continues to enhance its production line move rate to meet the growing EUV demand from advanced logic and DRAM customers.

As leading logic foundries (TSMC, Intel) move toward 0.55 High-NA EUV, Low-NA EUV has effectively become the standard infrastructure for secondary-tier applications. For DRAM manufacturers and secondary wafer fabs, today’s Low-NA EUV is akin to immersion ArFi (193nm) lithography machines a decade ago—technical risks have been mitigated by early adopters, making it a reliable, off-the-shelf production tool.

Therefore, technologically, DRAM is increasingly requiring EUV; economically, the AI boom has enabled more memory manufacturers to absorb the cost of EUV; and equipment-wise, Low-NA EUV has moved beyond validation of feasibility into the phase of scaling production capacity. The convergence of these three factors is now driving a genuine expansion of EUV’s customer base.

Technologies challenging ASML are rapidly increasing, such as xLight, Inversion, Substrate, Lace, Multibeam, and Canon, but most of them only target a single weakness in EUV systems: some replace the light source, others eliminate the mask, some switch to X-rays, and others use atomic beams.

Based on their impact on the lithography system's depth, they can be roughly divided into four major categories.

(Compiled and tabulated by Semiconductor Industry Observer)

(1) Light Source Reconstruction School

This group is attempting to change the way 13.5-nanometer EUV light is generated, addressing issues such as insufficient power, excessive energy consumption, and complex maintenance in current laser-produced plasma sources.

The representative company is the U.S. startup xLight. Currently, ASML’s EUV systems primarily generate plasma by striking tin droplets with lasers, releasing 13.5-nanometer EUV light. xLight aims to replace the existing laser-plasma light sources with free-electron lasers to provide higher-power, more stable light for multiple EUV scanners.

Defect distribution on the same wafer for traditional EUV and xLight EUV solutions (source: xLight)

xLight mentions on its website that, within a manufacturing system based on TSMC’s 3nm wafer unit cost of approximately $19,500, traditional EUV equipment accounts for 40% of the total chip manufacturing cost; xLight’s proposed FEL (free-electron laser) EUV solution can reduce EUV-related costs by 50%, lowering their share of total cost to 20%, thereby directly reducing the overall wafer manufacturing cost by 20%.

Halve the EUV cost (credit: xLight)

(2) Short-wave Transition School

Today, ASML's EUV lithography machines, whether the 0.33 NA NXE or the 0.55 NA High-NA EXE, still use 13.5-nanometer EUV light. The difference lies in the fact that High-NA increases the numerical aperture from 0.33 to 0.55, further improving single-exposure resolution to approximately 8 nanometers, and continues to support 2-nanometer and subsequent logic processes.

But another group of companies and research institutions began to consider a more radical question: Even if 13.5nm EUV could be continually extended through High-NA, it would still face challenges such as resolution, stochastic defects, photoresist, complex optical systems, and equipment costs. Instead of focusing solely on increasing the numerical aperture, could we directly further shorten the exposure wavelength?

Thus, BEUV (Beyond EUV) and soft X-ray lithography have re-entered the spotlight.

This group includes Inversion Semiconductor, Substrate, and a series of research projects centered on 6.7-nanometer BEUV. Although they employ different technologies, they share a common underlying principle: no longer treating 13.5 nanometers as an immutable constant, but instead attempting to transition to shorter wavelengths and rebuild the next-generation lithography system.

One of the most closely watched target wavelength ranges is 6.5–6.7 nm. According to the fundamental relationship of lithography, exposure resolution depends simultaneously on wavelength λ, numerical aperture (NA), and process factor k1. Under similar other conditions, reducing the wavelength from 13.5 nm to approximately 6.7 nm could theoretically significantly enhance resolution.

Moreover, 6.7 nanometers is not a newly emerged concept. Over a decade ago, research teams in Europe, China, Russia, and other regions had already begun studying next-generation lithography at the 6.7-nanometer scale, including plasma light sources, rare earth materials, and corresponding multilayer reflective optical systems. For instance, material systems such as La/B and La/B₄C have long been regarded as key candidates for 6.7-nanometer mirrors, with theoretical reflectivity reaching approximately 70%, and experimental systems have already achieved high reflectivity.

Russia's exploration offers another perspective.

Today, ASML's mainstream EUV light sources use LPP, or laser-produced plasma technology: high-power CO₂ lasers rapidly bombard molten tin droplets, converting them into high-temperature plasma from which 13.5-nanometer EUV radiation is extracted. This technology has undergone more than a decade of engineering validation, but tin debris, optical component contamination, energy conversion efficiency, and complex cleaning and protection systems have remained among the most challenging engineering issues for EUV light sources.

Some Russian research institutions have therefore long explored xenon and other gas-based plasma light sources, aiming to bypass the tin droplet system. Unlike metallic tin, gas-generated plasma can be evacuated through a vacuum system, theoretically reducing contamination of core optical components such as collection mirrors. For more information, click to read the article previously published by Semiconductor Industry Watch: “Russia Announces Breakthrough in Lithography Equipment.”

Inversion Semiconductor goes further.

The company is developing a compact particle accelerator based on laser wakefield acceleration (LWFA), aiming to drastically reduce the size of traditional particle accelerators. Its goal is to use high-power lasers to generate extremely strong electric fields in plasma, accelerating electrons to high energies over very short distances, and subsequently producing high-power, tunable, short-wavelength radiation.

The STARLIGHT source proposed by Inversion currently aims to cover a wavelength range of approximately 20 nanometers to 6 nanometers, enabling it to generate the 13.5-nanometer light required for today’s EUV lithography while also extending into the soft X-ray range. The ultimate goal is not merely to replace ASML’s tin-droplet source, but to build an entirely new generation of lithography platforms centered around this compact particle accelerator.

(Source: Inversion Semiconductor)

Substrate has chosen a different accelerator approach. It also generates X-rays using high-energy electrons, but its architecture differs from Inversion’s laser wakefield acceleration. Substrate accelerates electrons to near-light speeds using radiofrequency cavities, then passes them through alternating magnetic fields to produce extremely bright X-rays, ultimately exposing wafers through a new optical and high-speed mechanical system.

Substrate reports that random vias with a 30-nanometer pitch exhibit excellent pattern quality and critical dimension uniformity (credit: Substrate)

However, Substrate’s ambitions extend beyond merely building a new lithography machine. It aims to establish a more vertically integrated foundry model centered around this X-ray lithography technology, integrating particle accelerators, light sources, exposure equipment, and advanced wafer manufacturing into a single system.

Inversion is more like building the next-generation light source/lighting platform, while Substrate aims to use this new lithography technology to fundamentally reshape the entire foundry model.

From this perspective, ASML has currently chosen to “maintain 13.5 nm and continue improving resolution through higher NA”; meanwhile, researchers in Inversion, Substrate, and BEUV have opened another technological variable—if NA can be altered, why not change the wavelength λ itself?

However, shortening the wavelength does not mean lithography suddenly becomes simple. Once moving from 13.5 nanometers into the soft X-ray region at 11 nanometers, 6.7 nanometers, or even shorter wavelengths, nearly all systems established today for EUV—light sources, mirrors, masks, photoresists, metrology and inspection, and contamination control—would need to be redesigned.

(3) Nanoimprint School

Canon's nanoimprint lithography (NIL) is one of the most industrialized non-traditional lithography approaches currently available.

Traditional EUV lithography requires generating a high-power light source and then using multiple sets of high-precision mirrors to reduce and project the mask pattern onto the wafer. NIL, by contrast, directly presses a template with nanoscale patterns onto the wafer’s photoresist, similar to stamping a circuit design. This approach eliminates the complex projection lenses and the need for a high-power EUV light source, theoretically significantly reducing equipment cost, energy consumption, and process steps.

Canon has launched the FPA-1200NZ2C commercial equipment, initially targeting 3D NAND as its first mass-production application, with plans to later expand to DRAM and logic chips.

However, the challenges with NIL are equally significant. Since the template must come close to or even contact the wafer, particle contamination, template defects, demolding damage, and multilayer alignment issues can all directly impact yield. Therefore, Canon has bypassed ASML’s core high-power light sources and projection optical systems, but has not avoided the fundamental lithography challenges of mask fabrication, defect control, and nanometer-scale alignment.

(4) No-Light Exposure Group

This camp goes even further: not only eliminating the EUV light source and mirrors, but also abandoning photons entirely for exposure, instead using atoms or electrons to directly form patterns on the wafer.

In this area, Lace Lithography and Multibeam are attempting to directly achieve nanoscale patterning using particles such as atoms or electrons.

Among them, Lace Lithography, a startup founded in 2023, employs metastable helium atom beam lithography. The de Broglie wavelength of helium atoms is significantly shorter than the 13.5 nm EUV light, theoretically not subject to the same constraints imposed by traditional optical diffraction limits. Compared to charged ions, neutral helium atoms also reduce charging effects on the wafer surface. Lace’s fundamental approach has shifted from “photon lithography” to “matter-wave lithography”—it aims to change not just the light source, but the very medium of exposure.

Multibeam takes a completely different particle approach—multibeam electron beam direct writing. Electron beam lithography is not a new technology. In fact, it has long been one of the highest-resolution patterning tools in the semiconductor industry and a critical method for advanced photomask manufacturing, research and development, and low-volume chip production.

Its biggest—and most critical—problem is just one: it’s too slow. Multibeam aims to address this bottleneck. Instead of making a single electron beam faster, it increases the number of electron beam writing channels simultaneously. By deploying multiple independent electron beam columns, multiple areas can be exposed at once, transforming the traditional “single-stroke drawing” into “parallel multi-stroke writing.” More importantly, electron beams can write directly based on digital layouts, eliminating the need to pre-fabricate an entire set of photomasks as traditional EUV does. This brings another advantage: maskless manufacturing.

Multibeam has now launched the second-generation MBX-300 multi-beam electron beam lithography platform, designed for 300-mm wafers and applications including advanced packaging, silicon photonics, quantum devices, and compound semiconductors. The system can directly write from chip layout data without requiring traditional photomasks and supports seamless transition from prototyping to mass production. Its officially stated typical throughput is approximately 1–2 wafers per hour per writing chamber, with peak rates reaching up to 25 wafers per hour in specific applications such as Secure Chip ID. In 2026, National Tsing Hua University in Taiwan has ordered the next-generation MBX system for semiconductor research and joint development with chip companies.

MBX platform (credit: Multibeam)

Although both Lace and Multibeam use particles to achieve patterning, the two approaches address fundamentally different problems: Lace aims to break through the resolution limit by utilizing the extremely short de Broglie wavelength of neutral helium atoms to further reduce feature sizes; Multibeam seeks to overcome the efficiency limit of electron beams by employing parallel multi-beam, direct digital writing, and maskless manufacturing to scale electron beam technology—previously suited only for R&D and low-volume production—to larger-scale manufacturing.

The greatest challenge they collectively face is how to increase throughput to a level acceptable by advanced semiconductor fabs while maintaining nanometer-scale resolution. This reveals the ultimate challenge confronting all EUV competitors: the real barrier to lithography has never been resolution alone, but rather achieving “fine, precise, high-speed patterning” that can be sustained stably 24/7.

Overall, the challengers to ASML’s EUV can more accurately be described as dismantling EUV. The progressive relationship among these four approaches is: replace the light source → change the wavelength → eliminate projection → eliminate photons.

Returning to the original question: Has the barrier to EUV lithography machines truly disappeared?

The answer is clearly no.

An EUV lithography machine still costs hundreds of millions of dollars, and requires advanced facilities, masks, photoresist, inspection and metrology equipment, along with long-term process expertise and yield ramp-up. Both the financial barriers and the engineering challenges of turning equipment into actual mass production capabilities have by no means disappeared.

But another equally significant change is that the barrier to entry for EUV is indeed lowering—from a premium luxury item for a few giants to a production necessity for niche players. The adoption of EUV lithography machines by broader segments reveals the new rules of the game after Moore’s Law entered its second half. The “demystification” of EUV is accelerating.

But even greater changes may be occurring outside the EUV system. A technological encirclement beyond the EUV era is taking shape. Innovative alternatives challenging ASML’s roadmap are emerging everywhere. Although the vast majority of these new approaches remain confined to laboratories and pilot production, yet to cross the formidable barrier of high-volume manufacturing (HVM)—which demands extremely high yields, stable equipment utilization, and a vast ecosystem—their very existence has already cracked open a small gap in the expensive traditional lithography landscape.

Ending absolute monopoly does not mean that breakthroughs at a single point are easy to achieve. It is foreseeable that the future lithography market may no longer be dominated solely by ASML. A new, diversified competitive ecosystem is emerging—one in which standard Low-NA EUV fills the mid-range, cutting-edge High-NA EUV anchors the top tier, and emerging technologies carve out niches in specialized segments.

This article is from the WeChat public account "Semiconductor Industry Watch" (ID: icbank), authored by Du Qin (DQ).