This piece cuts through the hype of a $1 billion valuation to ask a question the semiconductor industry has been too afraid to ask: what happens when the physics of light stops working? Vikram Sekar doesn't just report on a new startup; he dissects the fragile economic and physical limits of the current manufacturing regime, suggesting that the "evolutionary" path of extreme ultraviolet lithography may have reached a dead end. For investors and engineers alike, the argument that a radical return to X-ray technology could be the only way forward is a necessary disruption to the status quo.
The Physics Wall
Sekar begins by grounding the reader in the relentless march of miniaturization, explaining how the industry has historically shrunk transistors by shortening the wavelength of light used to pattern silicon wafers. He traces the journey from mercury vapor lamps to the complex immersion techniques of the 2000s, noting how the industry hit a "physics wall" that forced a pivot to extreme ultraviolet (EUV) lithography. The core of his argument is that the current solution, while miraculous, is becoming economically unsustainable.
"The incredible cost and complexity of the machine puts it out of economic feasibility for a lot of major chip manufacturers in the world," Sekar writes, highlighting how even mid-sized foundries like GlobalFoundries abandoned the race for 7nm nodes due to the prohibitive expense. This framing is effective because it shifts the conversation from pure technological capability to the brutal reality of capital efficiency. The industry is currently relying on a single supplier, ASML, for tools that cost a quarter of a billion dollars each, creating a bottleneck that threatens the entire supply chain.
"At an estimated cost of a billion dollars for just the lithography tool, it probably does not make economic sense to make these chips anymore."
Critics might note that dismissing high-NA EUV as "pointlessly expensive" ignores the sheer volume of demand that could eventually justify such costs, but Sekar's skepticism is well-placed given the history of Moore's Law slowing down. The argument gains weight when he contrasts the current trajectory with the past, noting that "evolution has won out over revolution" in previous decades, but that this incremental approach may no longer be viable.
The Ghost of X-Rays Past
The article then pivots to the historical context of X-ray lithography (XRL), a technology that was once the darling of IBM and Motorola in the 1980s before being abandoned in favor of ultraviolet methods. Sekar skillfully weaves in the story of how the industry chose the "safer" path of UV evolution, only to find themselves staring down a similar cost cliff today. He cites a 2016 review paper to underscore the irony: "We could argue that had we persisted in our attempts to field an XRL chip technology, lithography would be cheaper, as all the optical 'fixes' needed to achieve, say, a 32 nm product would be unnecessary."
This historical parallel is the piece's most compelling element. It reframes the current struggle not as a failure of innovation, but as a failure of long-term strategic vision. The industry chose incrementalism, and now they are paying the price in complexity and cost. Sekar points out that the original abandonment of XRL was due to intellectual property disputes and collaboration challenges, not necessarily technical impossibility. "It was 'easier' to evolve systems, to make a series of incremental (albeit expensive) changes, rather than go to more radically different approaches," he observes, suggesting that the path of least resistance has led to a dead end.
The entry of Substrate, a startup claiming to have cracked the code on X-ray sources and optics, represents a potential "revolution" that the industry is desperate for. Sekar notes that the startup has demonstrated proof-of-concept tools that could compete with ASML at a fraction of the cost, a claim that, if true, would shatter the current monopoly. However, he maintains a healthy skepticism, quoting Substrate's CEO James Proud who admits, "this is an incredibly difficult problem but an extremely important one."
"The future of American chip manufacturing was lost forever."
Sekar uses this stark phrase to describe the 2001 sale of Silicon Valley Group to ASML, a moment that cemented the US's reliance on foreign lithography equipment. This historical anchor adds emotional weight to the technical discussion, reminding the reader that the stakes are not just about profit margins, but about national industrial sovereignty. The potential for Substrate to "lead again" is framed not as a guarantee, but as a slim hope that the US can reclaim its manufacturing dominance if the technology actually works at scale.
The Economic Imperative
The final section of the commentary focuses on the potential throughput advantages of X-ray lithography, which could theoretically solve the speed and cost issues plaguing EUV. Sekar argues that while EUV requires complex multi-patterning techniques to achieve smaller nodes, XRL could achieve the same results with single patterning, drastically reducing the number of steps and masks required. "This matters because TSMC, Samsung, and hopefully Intel, are the only foundries capable of mass-producing chips at the leading edge and ASML is the sole provider of lithography tools capable of producing these advanced nodes," he writes, emphasizing the fragility of the current ecosystem.
The argument is that Substrate's technology could democratize access to leading-edge manufacturing, breaking the stranglehold of a single supplier and a handful of foundries. This would allow more companies to innovate and reduce the geopolitical risks associated with relying on a single source of truth for chip production. However, the piece also acknowledges the immense technical hurdles that remain, particularly in X-ray optics and mask creation, which are not trivial problems to solve.
"America once led chip manufacturing but ceded that position to Taiwan over decades of policy decisions. Substrate's success means that America 'will lead again' in chip manufacturing — if the technology actually pans out at scale."
This conditional hope is the piece's most honest moment. Sekar does not promise a miracle; he presents a possibility that depends on overcoming decades of technical inertia. The reader is left with a clear understanding that while the technology is promising, the road ahead is fraught with uncertainty.
Bottom Line
Sekar's analysis is a masterclass in connecting historical context with present-day economic constraints, making a compelling case that the semiconductor industry's reliance on incremental evolution has reached its limit. The strongest part of the argument is the historical parallel between the abandonment of X-ray lithography in the 80s and the current crisis of EUV costs, which suggests that the industry may need to revisit "revolutionary" approaches. The biggest vulnerability, however, is the sheer difficulty of scaling X-ray technology, a challenge that has defeated the best minds in the field for decades. Readers should watch for Substrate's ability to move from proof-of-concept to mass production, as this will determine whether this is a genuine breakthrough or another false dawn.