Can ASML Increase Chipmaking Capacity Without Building More Machines?

ASML has two capacity problems. The first is familiar: customers want more lithography machines than the company expects to supply. In its third-quarter results, ASML said its existing plans would not fully meet expected demand in 2027 or 2028, despite planned expansion in Low-NA EUV and DUV immersion output. Customers are already discussing equipment needs extending into 2029 and 2030. [1][3]

The second problem is less visible. Chipmakers need more output from machines already inside their fabs. ASML’s Installed Base Management revenue reached €3.289 billion in the third quarter, almost €0.4 billion above guidance. Management attributed the beat primarily to productivity upgrades for EUV and DUV systems. Those field options allow customers to buy capacity without waiting for an entirely new scanner. [1][3]

What 10 extra wafers per hour actually buys

One disclosed example is PEP1-E, an upgrade package for the NXE:3800E. ASML says it lifts specified productivity from 220 to 230 wafers per hour at similar overlay performance. That is a 4.5% increase. It sounds modest until it is applied to a fleet of machines that can cost hundreds of millions of euros each and sit at the critical path of advanced chip production. [4]

A simple model shows both the attraction and the limits. At 85% productive time, a 10 WPH gain creates 74,460 additional exposure slots a year. ASML’s 2025 figures indicate that advanced logic used around 19 to 21 EUV exposures per wafer; using 20 produces 3,723 gross wafer starts. If only 75% of the scanner gain reaches finished wafers because of yield and other factory bottlenecks, the effective addition is 2,792 wafers a year. [5]

The fleet effect is meaningful. Twenty-three 10 WPH upgrades provide the same gross throughput as one 230 WPH scanner. Upgrading 100 tools adds 1,000 WPH, equivalent to 4.35 scanners. For a customer facing multi-year equipment lead times, that is real bridge capacity.

The downtime problem

An upgrade has to be installed on a machine that may already be producing valuable wafers. In the base case, a two-week outage on a 220 WPH tool removes 62,832 exposure slots. The additional 10 WPH then takes 10.1 months to replace what was lost. If installation can be bundled with planned maintenance, the economic penalty falls. If it creates a fresh shutdown during peak demand, the first year of gains can largely disappear.

Figure 1. A small throughput gain can take months to repay installation downtime.
*Source: Rokin model using ASML’s disclosed 220-to-230 WPH upgrade.*

Time is worth more than hardware

That changes how the investment should be valued. The wrong comparison is simply the upgrade price against the pro-rata cost of a new scanner. The better comparison includes time. A new machine adds independent capacity, but it also requires production at ASML, delivery, installation, calibration, qualification and suitable cleanroom space at the customer.

Using an illustrative €210 million new-tool price, 20% ancillary fab capex and a 24-month lead time, the pro-rata all-in value of 10 WPH is about €11 million. The present value of receiving two years of advanced-logic output earlier is about €48.5 million under the base assumptions. Two weeks of downtime cost about €23.6 million of foregone contribution. The resulting decision price ceiling is €35.9 million.

That figure is not an estimate of what ASML charges. The company does not publish the price of the upgrade, and wafer contribution margins are customer-specific. It is a break-even threshold showing why a field option can be worth materially more than its simple hardware equivalence when capacity is scarce.

Figure 2. Timing creates most of the base-case upgrade value.

*Source: Rokin model. Prices and contribution assumptions are illustrative.*

The same upgrade can be cheap or exceptional

The result is highly sensitive to the factory around the scanner. If only half of the throughput gain reaches finished wafers, downtime is four weeks and contribution is €5,000 per wafer, the decision ceiling falls to €5.9 million. In a high-value case with full pass-through, one week of downtime and €15,000 of contribution per wafer, it reaches €93.2 million. The same technical upgrade can therefore be marginal in one fab and exceptional in another.

Why upgrades do not replace new machines

This is why upgrades will not allow the semiconductor industry to stop building machines. A field option does not create another independent process tool. It does not add redundancy, another cleanroom position or the flexibility to run a different recipe. It cannot serve a customer that has no eligible installed system. And if deposition, etch, metrology or packaging is the true bottleneck, higher scanner throughput may only create more work-in-progress.

What this means for ASML and chipmakers

ASML nevertheless benefits from the scarcity. Installed Base Management sales rose from €6.494 billion in 2024 to €8.193 billion in 2025. Adding the first three quarters of 2026 to the company’s fourth-quarter guidance produces €12.239 billion, almost half again as much as in 2025. The category includes service as well as field options, so it is not a pure upgrade measure. But the third-quarter commentary shows that productivity upgrades are becoming an important way to monetize the installed base. [1][6][7][8]
For chipmakers, the optimal strategy is not upgrade or buy. It is upgrade first where the process is genuinely constrained, install during scheduled maintenance wherever possible, and continue preparing for additional machines. For ASML, the task is also dual: expand new-system production and develop field options that turn existing tools into higher-capacity assets.

The answer is therefore conditional. ASML can help customers add lithography capacity without delivering another scanner, and during a tool shortage that bridge capacity can be worth much more than its simple hardware share. But the upgrade creates economic output only when lithography is genuinely constrained and installation does not sacrifice too much existing production. Upgrades buy time; new machines still provide the independent capacity the industry ultimately needs.

Research note. This is a scenario-based capacity and investment analysis built from ASML public disclosures and explicit assumptions. ASML discloses the 220-to-230 WPH throughput change but not customer-specific upgrade pricing, installation downtime, wafer contribution margins or fab bottlenecks. The €5.9 million to €93.2 million scenario range and €35.9 million base decision ceiling are Rokin model outputs, not ASML prices or guidance.

Explore the research

The technical report develops the throughput conversion, downtime recovery, lead-time value, fleet-scale analysis and scenario sensitivities. The accompanying ASML Capacity Upgrade Model contains the formulas, assumptions and editable cases behind the €35.9 million base decision ceiling and the broader sensitivity range.

Sources and research

[1] Q3 2026 results and capacity outlook. ASML, “ASML reports €11.7 billion total net sales and €5.0 billion net income in Q3 2026” (14 October 2026): https://investor.asml.com/news-releases/news-release-details/asml-reports-eu117-billion-total-net-sales-and-eu50-billion-net-income-q3-2026. Supports Q3 sales, Installed Base Management revenue, gross margin, guidance and the supply-demand outlook.

[2] Q3 system mix and quarterly detail. ASML, Q3 2026 Investor Presentation (14 October 2026): https://investor.asml.com/static-files/44d4bee1-b2f9-41b7-b3f8-1bf28f993fe6.

[3] Upgrade demand and same-footprint capacity. ASML, Q3 2026 Prepared Remarks (14 October 2026): https://investor.asml.com/static-files/40c56a6b-1460-4f69-9340-6a869560d24b. Supports the approximately €0.4 billion Installed Base Management beat, productivity-upgrade demand and capacity commentary.

[4] PEP1-E throughput benchmark. ASML, Q1 2026 Investor Presentation (15 April 2026): https://investor.asml.com/static-files/0d685898-3b07-4658-8037-d60d422e66d2. Supports the disclosed NXE:3800E productivity increase from 220 to 230 wafers per hour at similar overlay performance.

[5] EUV exposure-count benchmark. ASML, Investor Day, End Markets and Lithography Spending (14 November 2024): https://investor.asml.com/static-files/cedf7981-ffc9-498e-a672-207a4bc3e975. Supports the 2025 average 0.33NA-equivalent EUV exposure counts of roughly 19-21 for advanced logic and 5 for DRAM.

[6] Q1 2026 Installed Base Management sales. ASML, Q1 2026 Financial Results (15 April 2026): https://investor.asml.com/news-releases/news-release-details/q1-2026-financial-results.

[7] Q2 2026 Installed Base Management sales. ASML, Q2 2026 Financial Results (15 July 2026): https://www.investor.asml.com/news-releases/news-release-details/q2-2026-financial-results.

[8] Historical Installed Base Management sales. ASML, Q4 / FY2025 Financial Results (28 January 2026): https://investor.asml.com/news-releases/news-release-details/q4-2025-financial-results.

[R1] Rokin calculations. Increasing_Semiconductor_Capacity_from_the_Installed_Lithography_Base_Technical_Report.pdf and Rokin_ASML_Capacity_Upgrade_Model.xlsx. The 74,460 exposure slots, 3,723 gross wafer starts, 2,792 effective wafers, 62,832 lost exposure slots, 10.1-month recovery period, €35.9 million base decision ceiling and €5.9-€93.2 million sensitivity range are Rokin scenario outputs, not ASML disclosures.