ASML Holding

We are updating our long-term ASML forecasts as the firm intends to expand Low-NA EUV and DUV immersion capacity by 30% in 2027 and another potential 30% in 2028.

Why it matters: Expansion would bring capacity to 85/110 Low-NA EUV systems by 2027-28 and 170/220 DUV immersion systems by 2027-28, compared with today’s 65 and 130 units, respectively. The high-demand environment has become more real in the second quarter, as customers have begun committing orders with down payments, giving ASML confidence to expand. The new 2027 capacity is virtually all booked, and we estimate the additional 30% 2028 expansion is very likely to proceed as ASML has already received large orders. While our previous 2030 revenue forecast was anchored to the top end of ASML’s long-term guidance of EUR 60 billion, this number looks irrelevant now, as we estimate ASML could deliver close to EUR 70 billion in revenue and EUR 69 in EPS already in 2028, even without fully utilizing its capacity. We must wait until the 2027 capital markets day for new revenue targets, but we estimate 2030 sales could exceed EUR 80 billion with little new capacity additions. After a 28% 2026-28 sales CAGR, we model high-single-digit growth in 2029-30, as growth moderates after the ambitious expansion.

The bottom line: We raise our fair value estimate to EUR 1,800 from EUR 1,200 per share for wide-moat ASML as we incorporate the new forecasts into our model.

Bears say: To stress test our new fair value, we model a bear scenario where after the massive 2026-28 expansion, sales growth stalls in 2029-30, similar to what happened in late 2024. Under this scenario, a two-year period of capacity digestion would arrive, bringing 2029 EPS to EUR 57 versus EUR 77 in our base case. ASML’s forward PE would likely contract to the 20-25 times range, where it has historically traded during cycle troughs, resetting valuation to the EUR 1,100 to EUR 1,400 range. This means our bear case is aligned with our previous base case.

Importantly, the 30% 2027 capacity expansion and a potential further 30% in 2028 can be achieved with existing cleanroom capacity utilization. In our view, this reduces execution risk and leaves room for gross margin potential upside, which we estimate will reach 60% in 2028. FactSet revenue consensus for 2028 stands still at EUR 57 billion, which we estimate could be almost achieved in 2027, and assumes little capacity expansion in 2028. 2028 revenue consensus hence looks conservative, and we believe it will need to rise.

2028 looks to be exceptionally strong, and 2029-30 is still where the debate will now live and what could mark the next leg up or down in the stock. Given customers are already hurrying up to secure 2028 capacity, we could get 2029-30 visibility within the next 6-12 months if current customer behavior continues. For the time being, we remain more comfortable moderating growth into the high single digits by 2029-30.

We also remain prudent on the pace of High-NA EUV adoption. While ASML should have capacity for 20 High-NA EUV tools by 2027, most customers will remain busy opening new fabs and aggressively ramping up production during the next three years, potentially delaying technology adoption. TSMC, ASML’s largest customer, still trails Intel on High-NA, so we expect High-NA sales will get close to 20 units by the end of this decade, breaking above 30 only past 2030.

The US Match Act, still to be approved, remains a near-term risk, as the US could impose new export restriction controls on ASML’s DUV lithography machines. While this represents a headline risk, the 2022-25 period proves that export restrictions can sometimes have counterintuitive second-order effects. During this time period, ASML China sales more than tripled despite increased restrictions, as Chinese customers front-loaded purchases not only to satisfy local demand but in anticipation of future controls.

Although the entire semiconductor equipment sector has produced strong year-to-date returns, ASML and ASM International’s performance has lagged US peers Applied Materials and Lam Research. While Lam’s outperformance could be justified by its higher revenue exposure to memory, which has outperformed year to date, Applied remains the best-performing stock year to date while having a similar memory mix as ASML, leaving potential room for ASML's stock to catch up.

Our fair value estimate for the ADR shares is $2,050 per share.

We expect 2026 and 2027 sales of EUR 44 billion and EUR 56 billion, respectively. The medium-term outlook has improved meaningfully since the summer of 2025, and ASML is positioned to expand Low-NA EUV and DUV immersion capacity by up to 70% from 2026-28. While our previous 2030 revenue forecast was anchored to the top end of ASML’s long-term guidance of EUR 60 billion, this number now looks irrelevant, as we estimate ASML could deliver close to EUR 70 billion in revenue and EUR 69 in EPS already in 2028, even without fully utilizing its expanded capacity.

While we must wait until the 2027 capital markets day for new revenue targets, we estimate 2030 sales could exceed EUR 80 billion with little new capacity additions. After a 28% 2026-28 sales CAGR, we model high-single-digit growth in 2029-30, as growth moderates after the ambitious expansion.

Memory makers Samsung, SK Hynix, and Micron are citing a shortage of high-bandwidth memory capacity, driven by AI demand, that could last until 2028. Memory makers are deliberately expanding capacity more carefully to avoid the mistakes of past cycles, in which oversupply and price crashes hurt the market. This measured approach should support a longer runway in the medium term.

For the next decade, we model a 15% revenue CAGR. The continued long-term growth in semiconductor fabs, the migration of new chip layers to EUV lithography, and machine price increases are ASML's main growth drivers. Gross margin and EBIT margin expansion will come from the operating leverage of R&D and operating expenses, more expensive EUV machines at better gross margins, and an improvement in EUV service margins.

Service revenue also contributes to long-term revenue growth as more and more complex machines reach the field with 30-year average useful lives. Low-NA and high-NA EUV have an estimated service revenue/equipment revenue ratio of more than 150% over their lifetimes, compared with DUV’s 130%. We model EBIT margin expansion from 33% in 2025 to 45% in our terminal year.

We model slightly higher R&D and selling, general, and administrative expense intensity compared with management’s long-term targets, as ASML will have to keep investing in technology and productivity to support growth and justify the high price tag of its machines.

We assign ASML a Wide Morningstar Economic Moat Rating supported by intangible assets, cost advantages, and switching costs. ASML is the world's largest supplier of lithography machines for semiconductors with around 90% market share. It enjoys a wide technology gap versus competitors Nikon and Canon, and its large investments in research and development should continue widening ASML’s moat and act as a barrier to entry. Intangibles come from decades of internal know-how and long-term collaboration with firms like Carl Zeiss and scientific research institutes. Switching costs come from software and servicing of the machines; fabrication plants cannot afford unplanned downtime, which can cost millions of dollars.

ASML sells semiconductor lithography machines, which are used to print nanometer patterns in chips. Its two main product lines are DUV and EUV lithography machines. DUV has been ASML’s profit engine for almost two decades. It has been used since the early 2000s and is still widely used to manufacture chips today. ASML is the only company capable of producing EUV lithography machines, which use a light source to print chip patterns and are required to manufacture the most advanced chips used in smartphones, computers, and artificial intelligence training. ASML’s lithography machines can print patterns that are up to 30,000 times thinner than a human hair. These patterns form a highly intricate 3D puzzle, with dozens of interconnected layers that optimize computational performance, energy consumption, and heat dissipation in a chip. Moore’s Law, which states that the number of transistors in a chip will double every two years, is becoming more challenging to achieve. However, ASML machines keep providing new workarounds to create 3D transistor structures, shrink chip patterns, and improve semiconductors' energy efficiency.

ASML machines are a very complex combination of physics, electrical, and mechanical systems. An EUV machine takes 12-18 months to be assembled, weeks to get installed, and can occupy dozens of square meters and weigh up to 200 metric tons. These machines have so many subsystems that, according to ASML, a single engineer would not be able to understand the entire machine, but only his or her area of expertise. ASML’s lithography machines are the result of the assembly and integration of thousands of high-quality supply parts, with ASML acting as the coordinator and assembler. Machines are also the result of decades of internal know-how, R&D, and partnerships with firms like Carl Zeiss and Philips or research centers like Imec. ASML’s executive vice president of technology, Jos Benschop, said in an interview on Dutch television station VPRO in 2022: “You could put 10,000 intelligent people in the same room, give them all the drawings of our machine, and the machine wouldn’t work. Some things are simply not in the drawings.”

In our view, Nikon and Canon are not competitors for ASML, as their lithography technology is well behind. The most advanced machines Nikon and Canon can manufacture are immersion lithography machines, which are workhorse machines that are still widely used today (also by ASML, called DUV). Even if Nikon or Canon did decide to replicate ASML’s EUV machines, this would require an immense amount of time and money. They would need to make the technology viable on their own, develop similar industry relationships, and would need decades to improve their internal processes. Aside from its technology, ASML has very long-term relationships with its customers, which gives them first-hand knowledge of fabs' pain points and an advantage when deciding where to innovate.

A decade ago, TSMC, Samsung, and Intel invested EUR 4 billion to fund the development of EUV and redesigned their fabs to make them ready for this technology. Even if a new and viable lithography technology that could compete with ASML’s EUV were developed, fabs would have to be entirely redesigned again for this purpose, something we consider very unlikely. Fabrication plants are also highly standardized to facilitate knowledge sharing and reduce manufacturing defects, so fabs would need to change their entire fleet of lithography machines if they changed suppliers, an endeavor that would cost billions. From an R&D perspective, Canon invests EUR 2.2 billion (JPY 300 billion) per year across four divisions, with lithography representing less than 10% of group revenue, compared with ASML’s R&D budget of more than EUR 4 billion, which is purely focused on lithography. Nikon invests around EUR 450 million per year (JPY 70 billion) in R&D across five divisions. This huge gap provides ASML with a cost advantage that generates high barriers to entry and seems virtually impossible to close at this point.

Lithography machines must be extremely precise, but also be able to function at high uptime and high manufacturing yield. Fabs have enormous fixed costs, so they need to function 24/7 to generate the necessary economies of scale. In addition to this, lithography machines are the bottleneck of the plant; this means any downtime can cost millions of dollars to a foundry. Even if a competitor did manage to create machines as precise as those of ASML, it would still need to deliver the same uptime and manufacturing yields. ASML’s DUV and EUV machines have uptime north of 95%. For example, Chinese lithography company Shanghai Micro Electronics Equipment claims to have reached the 5-nanometer process node. Even if this is true, it is highly debatable whether these machines can function at an acceptable uptime and provide a proper manufacturing yield. The semiconductor value chain is highly globalized, and we believe China lacks the geopolitical connections, engineering capabilities, and know-how to replicate ASML’s lithography machines at the same level of technology and reliability.

ASML has a recurring-revenue business model, with machines that were shipped in the 1980s and 1990s still generating service revenue today. The installed-base management business focuses on servicing, upgrading, and software for machines. Once a lithography machine is sold, the customer is locked in for 20-30 years; we estimate retention rates are very high. As lithography machines get more complex, they become more crucial in the manufacturing process, and ASML’s switching costs strengthen. A modern fab costs around $20 billion-$24 billion to build, of which $4 billion-$6 billion goes to lithography (20%-25%). Third-party service providers lack the technical skills to service these machines, but even if they could, we doubt customers like TSMC or Samsung would risk using an unverified service provider. During its lifespan, an ASML DUV machine—the previous generation to EUV—generates 130% in service revenue compared with equipment revenue. Management estimates low-NA EUV machines, used for cutting-edge chips, will generate around 150%, given their higher complexity, and we estimate this could be above 150% for high-NA EUV. ASML’s offering also extends beyond traditional servicing as it offers software and hardware upgrades for its customers. A customer that owns a machine can step up productivity with some investment and a few weeks of downtime. Although upgrades are still expensive, they are still cheaper than purchasing a new machine, so we believe customers find value in this proposition.

ASML’s machines can last 30 years, providing recurring service revenue. The switching costs and intangible assets required to displace these machines are enormous, with no competitor coming close to ASML’s technological leadership.

AI is driving strong semiconductor demand, which results in more fabs and more demand for tools across logic and memory markets.

Output gains support price increases whenever new machine prototypes are introduced. Product platforms can share many parts and subsystems, facilitating standardization, shorter lead times, and procurement savings.

ASML sells a low-volume, high-price product. Lithography machines represent a high proportion of customer costs, so if the company fails to innovate, customers will look for alternatives.

If controls on exports to China worsen, ASML’s near-term growth might suffer. The firm has no effective control over this risk as it largely depends on US government decisions.

The cyclical nature of the semiconductor industry adds to ASML’s risk profile. Also, ASML depends on a limited number of suppliers for certain components, so any disruption could create bottlenecks and delays.