What Fabs Actually Fear Isn’t a New Vendor – It’s Being Second

The typical assumption about semiconductor manufacturers is that they’re conservative almost by temperament: risk-averse, slow to change, unwilling to bet a production line on anything unproven. It’s a reasonable assumption in general, and it’s also demonstrably wrong at the exact moments that matter most. The clearest counter-example in modern chipmaking history isn’t a small one. It’s EUV lithography — arguably the riskiest, most expensive, least proven vendor bet the industry has made in decades and one that the industry’s most conservative players didn’t shy away from. They funded it themselves.

The bet nobody should have made, on paper

Extreme ultraviolet lithography took ASML more than three decades and, by some accounts, over $9 billion in R&D spending to bring anywhere near commercial viability. For most of that development period, the outcome was genuinely uncertain — as late as the 2010s, a meaningful share of semiconductor experts doubted EUV could be successfully commercialized at all. This wasn’t a startup with a clever pitch deck asking for a pilot. It was a fundamentally unproven physical approach, requiring breakthroughs in optics, materials and light-source engineering that hadn’t been demonstrated to work reliably at production scale, from a single supplier the industry would become entirely dependent on if it succeeded.

By any normal read of conservative, semiconductor manufacturers should have waited on the sidelines until someone else proved it out. TSMC didn’t. After encountering an early EUV prototype through imec, TSMC chose to partner directly with ASML on its development — a sharp contrast to competitors like Canon and Nikon, whose lithography divisions stayed closed and made comparatively little effort to collaborate externally on the technology. That early bet mattered enormously to how the next fifteen years played out.

The moment conservatism flipped into a funding race

The starkest evidence that fear of falling behind, not caution, was the dominant force in the room came in 2012, when Intel, Samsung and TSMC — direct competitors, in an industry not known for cooperation — collectively acquired roughly 23% equity in ASML specifically to help fund EUV’s continued development. This wasn’t a hedge against a safe bet. It was three rivals jointly bankrolling a supplier whose core technology hadn’t yet been proven to work, because the alternative — watching a competitor secure exclusive early access if it did work — was judged to be the riskier outcome. Caution about the technology was real. It simply wasn’t as strong as the fear of being the one left behind if a rival got there first.

Once one player proved it, the race compressed fast

By 2019, TSMC was shipping high-volume chips built with EUV, and by its own account has stayed roughly a node ahead of Samsung and Intel ever since. Samsung, meanwhile, brought EUV into production for its own 7nm process and later extended it into DRAM manufacturing, each time racing to close a gap rather than waiting for further proof. Intel’s comparatively slower EUV transition is widely cited as a contributing factor in the multi-year process delays that cost it its long-held manufacturing leadership — a visible, costly demonstration of what happens to whoever ends up last in this particular race.

That dynamic hasn’t stayed in the past. The same pattern is playing out again right now with High-NA EUV, ASML’s next-generation system: SK Hynix installed its own High-NA tool in September 2025 to prepare for mass production. And Samsung, in spite of financial challenges, is solidly progressing with High-NA EUV. The specific technology has changed. The competitive mechanism hasn’t.

The actual tipping dynamic

What EUV’s adoption history shows isn’t that semiconductor manufacturers stopped being cautious. It’s that caution holds only up to a specific threshold: as long as the perceived cost of adopting something unproven outweighs the perceived cost of falling behind a rival who adopts it first, the conservative choice is to wait. Once that balance flips — once one credible peer moves and starts pulling ahead — the same organizations that spent years being cautious can turn into some of the fastest adopters in the industry, precisely because standing still has become the more expensive option.

What this means for sequencing, not for certainty

None of this means every unproven technology gets this treatment or that any specific process technology today is guaranteed to follow the same arc EUV did — most don’t and betting on inevitability would be its own kind of overclaiming. What it does mean is that the right strategic question for an early-stage supplier in this industry isn’t “how do we convince every conservative customer at once.” It’s closer to “who is the one credible first mover whose adoption would change the calculus for everyone else watching.” That’s a sequencing problem, not a persuasion problem — and it’s the problem EUV’s own history shows has a solvable, well-documented shape, whatever specific technology is running through it next.