When the world’s most important chipmaker loosens its wallet strings, the rest of the semiconductor industry pays attention. When it does so at the cutting edge of logic manufacturing, the consequences ripple unevenly across the supply chain. That is precisely what is happening with the latest capital-expenditure upgrade from Taiwan Semiconductor Manufacturing Company s. While equipment makers across the board will benefit, the real prize looks set to land with one firm above all others: ASML, the monopolist supplier of extreme ultraviolet (EUV) lithography machines.

At first glance, the numbers appear reassuring rather than revelatory. According to analysts at Bernstein Research, TSM’s updated guidance implies that capital expenditure in 2026 will rise by roughly 32% year on year from 2025 levels. That pace is broadly in line with expected revenue growth of around 30%, keeping capex intensity elevated at about one-third of sales. In other words, TSM is not embarking on a reckless spending spree; it is merely maintaining a long-standing habit of ploughing cash back into its factories to defend technological leadership.

Yet averages conceal composition, and composition is where ASML’s advantage lies. TSM has made clear that 70–80% of its spending will now be directed towards advanced logic, up from around 70% last year. This matters because not all tools are created equal. A dollar spent on a mature node buys a broad assortment of etchers, deposition systems and metrology tools from a wide range of suppliers. A dollar spent on leading-edge nodes, by contrast, is disproportionately channelled towards lithography—and, more specifically, towards ASML’s EUV systems.

The near-term picture is complicated by timing. Bernstein’s David Dai notes that some of the increased allocation to advanced logic is being offset by a temporary reduction in equipment purchases, as TSM accelerates the construction of new cleanrooms ahead of heavier capacity requirements in 2028 and 2029. In plain English, the buildings are going up before the machines are wheeled in. Even so, Dai expects advanced logic equipment growth of around 30% in 2026, with a sharper acceleration once those empty shells are ready to be filled later in the decade.

For ASML, the crucial variable is not merely how much TSM spends, but where it spends it. The foundry’s capacity additions are concentrated in its N3 and N2 nodes—process generations that sit at the frontier of logic manufacturing and are central to the artificial-intelligence boom. These nodes carry significantly higher lithography intensity than the wafer-fab-equipment (WFE) average. Put differently, a greater share of every incremental dollar flows to the lithography stage of production.

Dai estimates that lithography accounts for roughly 30%-plus of total equipment spending at N3, easing slightly at N2 but remaining well above the industry norm. That compares with ASML’s current lithography intensity of around 25% across its installed base. The implication is straightforward: as TSM tilts its spending mix towards N3 and N2, ASML captures a growing slice of the pie, even if total industry spending grows at a more pedestrian pace.

This asymmetry explains why Bernstein believes ASML stands to gain more than its peers. Under the revised capex framework, Dai estimates that ASML’s revenue from TSM could grow by around 27% year on year—far above consensus expectations, which imply growth closer to 10%. For suppliers of less node-sensitive tools, the uplift will be real but more modest.

There is also an upside option embedded in the numbers: EUV volumes themselves. Based on the updated outlook, Dai believes TSM alone could add 120,000–130,000 wafer starts per month of combined N3 and N2 capacity. At those nodes, roughly 2.5 EUV machines are required for every 10,000 wafers per month of capacity. By that arithmetic, TSM may need to purchase around 30 additional EUV tools—an outcome that would exceed Bernstein’s prior forecasts.

Such figures underscore the peculiar economics of ASML’s position. Each EUV system costs well over €150m, takes months to assemble and can only be sourced from one supplier. For ASML, incremental demand does not merely add volume; it reinforces pricing power, stretches order books and deepens technological dependence. For customers like TSM, there is no substitute, only scheduling negotiations.

Investors, unsurprisingly, have taken note. Bernstein has reiterated its Outperform rating on ASML and maintained a €1,300 price target, keeping the company as a top pick within European semiconductors. The optimism rests not on a cyclical rebound alone, but on structural forces: the relentless march to smaller nodes, the compute-hungry nature of AI workloads and the capital intensity of staying at the frontier.

That does not mean risks have vanished. Semiconductor cycles remain volatile, geopolitics intrudes uncomfortably into supply chains, and ASML itself must navigate export controls and capacity constraints. Moreover, TSM’s current emphasis on cleanroom construction implies that some revenue is deferred rather than pulled forward. But delay is not denial. The machines will be needed; the question is when, not if.

Should investors be buying ASML right now? Tools such as ProPicks AI promise algorithmic clarity, sifting through hundreds of metrics to identify attractive risk–reward profiles. They may or may not flag ASML as a current favourite. What is harder for any model to ignore, however, is the underlying logic of the lithography business. When the world’s leading foundry doubles down on the most advanced nodes, the company that owns the bottleneck stands to benefit disproportionately.

In that sense, TSM’s capex upgrade is less a general stimulus than a targeted transfer. The rising tide will lift many boats, but ASML’s vessel sits directly under the waterfall.