The End of EUV? A New X-Ray Chipmaking Technology Could Redefine the Semiconductor Industry

For years, advanced chips manufacturing has revolved around two dominant forces. TSMC produces nearly 90% of the world’s most advanced logic chips. ASML controls the other critical half of the equation, as the only company capable of building the lithography machines required to print them. That equilibrium held—until now.

A US-based startup claims it can disrupt both pillars at once. Its new lithography system promises sub-nanometer patterning in a single exposure, at roughly half the cost of today’s tools. The real disruption lies deeper. This company does not intend to sell machines. Instead, it plans to build entirely new semiconductor factories around this technology.

The semiconductor industry is known for caution. Progress typically arrives through incremental refinements measured over decades. This approach is different. If the technology performs as claimed, it does not merely challenge ASML or TSMC. It threatens the entire economic and technical model of advanced chip manufacturing.

Every major breakthrough in semiconductor scaling eventually hits the same barrier: lithography. Chip fabrication involves hundreds of tools and thousands of process steps, but lithography dominates them all. It defines how small features can be drawn, how dense transistors can be packed, and how powerful chips can become.

At the cutting edge, lithography relies on extreme ultraviolet systems. 

Photo - ASML: EUV Lithography Machine

EUV machines are the most complex manufacturing tools ever created. Their role is deceptively simple: project transistor patterns onto silicon wafers. In practice, they determine the future of computing.

Modern transistors measure only a few nanometers across. At that scale, structures are tens of thousands of times thinner than a human hair. Creating them requires light to do the impossible—define features smaller than its own wavelength.

Early lithography relied on deep ultraviolet light at 193 nanometers. Physics eventually set a hard limit. Features smaller than the wavelength could no longer be printed reliably. The industry responded by moving to extreme ultraviolet light at 13.5 nanometers. That single shift, combined with decades of refinement, unlocked today’s most advanced chips.

EUV comes at a steep price. The light is absorbed by air, glass, and lenses, forcing the entire system to operate in near-vacuum conditions. EUV photons are generated by firing high-power lasers at molten tin droplets inside sealed chambers. The resulting light is guided by mirrors polished to near-atomic precision. After decades of research, only ASML can build such machines. Each tool costs around $400 million.

Despite the cost, the economics still function—barely. A single EUV tool can process wafers worth over $600 million annually. The challenge is not acquisition, but execution. At leading-edge nodes, only TSMC, Samsung, and Intel can operate these systems at scale with acceptable yields.

As transistor dimensions continue to shrink, lithography becomes even more challenging. Single exposures can no longer resolve the smallest features. The industry compensates through multi-patterning, splitting designs across multiple passes. This workaround increases complexity, raises defect risk, and drives costs sharply upward.

Photo - ASML: High-NA EUV Machine

To extend scaling, ASML introduced High-NA EUV, using more aggressive optics to push beyond 2 nanometers toward the angstrom era. These tools are already deployed at TSMC, Samsung, and Intel. Each one approaches $500 million in cost. The next step, Hyper-NA EUV, pushes the same concept even further—larger machines, tighter tolerances, higher prices.

At some point, the economy collapses. Leading-edge fabs are projected to cost $50 billion each by the end of the decade. Wafer costs could exceed $100,000. Advanced manufacturing becomes accessible only to a handful of players with massive capital reserves. Scaling continues, but costs rise faster than benefits.

This raises a fundamental question. What if the problem is not how far EUV can be pushed—but the decision to keep pushing EUV at all?

Substrate, US based startup proposes a different path. Instead of extending EUV, it abandons it entirely in favour of X-ray lithography. The concept has existed for decades, but practical implementation was blocked by fundamental challenges. X-rays have much shorter wavelengths than EUV, making them ideal for defining ultra-small features. They are also notoriously difficult to generate, focus, and control.

Historically, X-ray lithography required synchrotrons—particle accelerators hundreds of meters long. Such systems were confined to research facilities, not factories. What changed was not the physics, but the supporting technology. Over time, compact accelerators, improved optics, and more controllable sources emerged. Substrate integrated these advances into a single manufacturing platform.

Instead of increasing process complexity through multi-patterning, the approach focuses on single-shot exposure. EUV operates like drawing a pattern line by line. X-ray lithography aims to imprint the entire pattern in one pass. With wavelengths measured in angstroms—up to 1,000 times shorter than EUV—the theoretical resolution ceiling is dramatically higher.

The difficulty lies in control. X-rays pass through most materials instead of bending, making conventional optics ineffective. For decades, this confined X-ray lithography to academic research. Substrate claims it has crossed enough technical thresholds to transform it into a viable manufacturing tool.

Inside the system, electrons are accelerated to near-light speed using radio-frequency cavities. Magnetic structures force these electrons to oscillate, producing intense X-ray radiation—similar to a synchrotron, but compressed into a factory-scale system.

Substrate has released early results. The company reports successful printing of 12-nanometer features, suitable for sub-2-nanometer transistor architectures. More importantly, it claims single-patterning across all layers. Resolution is reported to be comparable to ASML’s most advanced High-NA EUV tools.

Consistency is where the claims become most striking. Substrate reports feature uniformity across entire wafers with variations as small as 0.25 nanometers—fractions of an atom. If verified, this enables denser logic, fewer defects, and dramatically lower costs. The proposed tool cost is around $50 million, compared to $500 million for High-NA EUV.

However, lithography alone is not enough. X-rays interact differently with matter. Existing photoresists, masks, and materials fail under these conditions. New chemistries, optics, and damage-mitigation strategies are required. Throughput remains another unknown. Demonstrations in controlled environments do not equate to full-scale manufacturing.

Mass production demands relentless reliability. EUV required more than a decade to transition from lab to fab. X-ray lithography faces a similar journey. For this reason, Substrate does not plan to sell tools. Instead, it intends to build its own fabrication facility in the United States, operate the full process internally, and offer foundry services directly—placing it in competition with TSMC and Samsung.

This is not merely a new machine. It is an attempt to reinvent the factory itself.

The challenge is scale. TSMC’s advantage lies not only in equipment, but in decades of yield learning, process control, and volume. Operating roughly 30 fabs and producing around 1.6 million wafers per month, TSMC benefits from a feedback loop few can match.

Substrate’s plan will take years. Success could mark the beginning of a new era. Failure would reinforce how unforgiving semiconductor manufacturing truly is.

The stakes extend beyond technology. Advanced chips underpin economic power, AI development, and national security. Cost reductions could unlock innovation in the same way reusable rockets transformed spaceflight. When SpaceX challenged the assumption that rockets must be disposable, launch costs fell by an order of magnitude—and entire industries emerged.

If Substrate can halve the cost of advanced manufacturing, the impact would be similarly profound. Cheaper tape-outs mean faster iteration, more experimentation, and broader access to cutting-edge compute. Innovation accelerates.

Substrate is not alone. Companies such as xLight and Inversion are also developing particle-accelerator-based light sources. Research efforts continue across Europe, Japan, and China. Most of these initiatives aim to extend EUV, not replace it. Substrate’s ambition is broader: eliminate EUV entirely and rebuild the manufacturing stack around a fundamentally different approach.

If successful, the effects would compound across computing, AI, and every technology built on top of them. Throughout history, progress has followed advances in computation. That pattern shows no sign of changing.

---

Follow Storyantra for deep dives into breakthrough technologies, global innovations, and the ideas shaping the future.