Goldman Sachs: ASML is seriously undervalued!

```

Global AI infrastructure investment continues to exceed expectations, with the semiconductor equipment sector as a whole recording strong gains. However, as a leader in the lithography segment, ASML’s year-to-date rise of about 45% has significantly lagged its peers, and market pricing is notably deviating from the company’s fundamental resilience.

Goldman Sachs’ latest research report raises ASML’s 12-month target price to 1,600 euros (bull case at 2,000 euros), reiterating its Buy rating. Historically, the company has enjoyed about a 20% valuation premium to its peers, yet its current valuation is at the 10th percentile of its five-year range, which is clearly mismatched with accelerated AI growth and ASML's strengthened EUV monopoly position.

The report points out that the market pricing on three key drivers—AI-driven EUV revenue, equipment demand from manufacturing inefficiency, and continued EUV layer count increases—is all overly conservative. In the base case, these assumptions are higher than what the market is implying, and the risk-reward ratio tilts 2x positively, meaning upside potential significantly exceeds downside risk.

Looking ahead, three catalysts deserve attention: First, after the GAA architecture transition is completed, the process shrinkage roadmap is expected to be updated, becoming a leading signal for EUV layer growth; second, increased design orders from lagging manufacturers validate that EUV demand is spreading from a single customer to multiple customers; third, accelerated enterprise AI adoption (especially Agentic AI) may further boost capital expenditure expectations.

Valuation significantly below peers, discount not sustainable

ASML’s current valuation discount is not reasonable; its fundamental strengths have yet to be fully priced in by the market.

Year-to-date, ASML’s stock gain has significantly lagged the European tech hardware sector’s overall gain of about 60%, as well as its main semiconductor equipment peers: ASM International, BESI, Applied Materials, Lam Research, and KLA have risen about 70%, 95%, 70%, 70%, and 55% respectively.

In terms of valuation, ASML has historically enjoyed about a 20% premium over global semiconductor equipment peers, but is now trading at a slight discount, in the 10th percentile of its past five-year range. This discount is clearly mismatched with the company’s rapidly growing AI business, expanding exposure to advanced memory chips, and the continually strengthening competitive moat from a higher proportion of EUV products (ASML’s virtual monopoly in EUV lithography).

AI infrastructure investment accelerates, EUV demand structurally lifted

Concerns about the sustainability of AI demand are gradually fading, and recent data from cloud vendors and the supply chain all point toward a more optimistic outlook. Goldman Sachs recently raised its wafer fab equipment (WFE) spending forecasts, providing important quantitative support for a positive view on ASML.

First, hyperscale cloud vendors’ capital expenditure remains resilient. Google’s Q1 AI revenue was up 63% year-on-year, its order backlog doubled quarter-on-quarter to $460 billion; Amazon AWS revenue rose 28% year-on-year, the fastest growth in 15 quarters, and AI-related revenue maintained triple-digit growth.

A more important signal comes from Alphabet, which raised its 2026 capital expenditure guidance to $180-190 billion, and made it clear that 2027 capex will be significantly higher, easing market concerns about a potential 2026 peak. Based on this trend, hyperscale cloud capex is projected to grow from $720 billion in 2026 to $918 billion in 2027 and $1.02 trillion in 2028, with growth rates of 28% and 11% respectively.

Second, the semiconductor supply chain is also showing signs of tightness. TSMC pointed out in its Q1 results that the advanced AI chip supply-demand gap remains highly tight and that the company is actively expanding capacity; Samsung stated that its fulfillment rate is at a historic low, and some customers have pulled forward their 2027 demand.

Against this background, global front-end equipment spending forecasts have been raised. The global WFE market is expected to reach $141 billion and $186 billion in 2026 and 2027 respectively, with year-on-year growth rates of 28% and 32%, both above previous forecasts of $132 billion and $160 billion.

Finally, focusing on ASML's EUV business, AI's contribution share is set to keep rising. Currently, it is estimated that AI-related revenue will account for a high-20% share of EUV by 2027, while the current stock price only implies a mid-20% share; in a bull scenario, the share could rise to the 30% range. Overall, between 2025 and 2027, AI demand is expected to contribute about 40% of EUV tool incremental demand in the base case and about 50% in the bull case.

Structural inefficiency in customer market: Equipment demand from global fab dispersion underestimated

Multiple wafer fabs expanding simultaneously worldwide will drive higher aggregate wafer starts, thus boosting lithography machine demand, but this logic has yet to be fully reflected in the current stock price.

The current stock price only implies about 7% incremental demand from manufacturing inefficiency, while the target price reflects 8%, which is near the upper bound of ASML’s own guidance range (5%-8%). Although the difference seems small, given ASML's large revenue base, the absolute amount is significant.

Samsung is a key catalyst in this logic becoming reality. The company recently released several upbeat signals: strong momentum in 2nm node customer orders; evaluating a second fab in Taylor, Texas, USA; and in its Q1 2026 results, Samsung made it explicit that equipment spending would substantially increase as new capacity is deployed. Additionally, Samsung noted its current fulfillment rate is at a historic low, with some customers pulling forward 2027 demand, further proving the urgency of capacity expansion.

More broadly, beyond TSMC, if Samsung and other leading fabs can achieve scaled expansion in advanced process nodes, it will break the single-vendor dominance of the past. Multiple globally distributed advanced fabs ramping expansion means the same capacity requires more equipment investment (i.e., greater manufacturing inefficiency)—this will drive higher total wafer start demand, a structural change that the market has yet to fully price in.

Growth in EUV layers: Technical penetration path is conservative; even if new tech is delayed, EPS impact remains positive

The current stock price corresponds to a technical penetration pace that remains on the low side of ASML's own guidance. ASML's official 2030 target is 25-30 layers for logic chips and 7-10 layers for memory chips (currently about 5 for memory). In the bear, current price, and bull scenarios, logic chips correspond to 22, 23, and 25 layers respectively, and memory chips to an increase of 2, 2, and 3 layers respectively.

Even if the next-gen lithography technology (high-NA) is delayed, customers would buy more of the current low-NA equipment to make up for capacity shortfall. Current tools have higher margins, and higher sales help boost ASML’s EPS. The target price corresponds to 24 EUV layers for logic chips in 2027, just above the 23 layers implied by the current price; the earlier 25-layer assumption was reduced due to a major customer’s cautious outlook toward high-NA, but the slight increase in low-NA shipments actually positively impacts EPS.

In the coming quarters, multiple catalysts could further drive ASML’s share price: the GAA architecture transition is expected to complete by the end of 2026, with nodes roadmap updates becoming a leading indicator for EUV layer growth; lagging players winning more design orders, further validating the multi-customer market; accelerated adoption of enterprise AI (especially CPU and HBM demand from Agentic AI) may support further upward revisions of capex expectations.

Risk Warning and DisclaimerThe market involves risks, and investments should be made prudently. This article does not constitute individual investment advice, nor does it consider the special investment objectives, financial status or needs of any specific user. Users should consider whether any opinions, views or conclusions in this article suit their particular circumstances. Investing accordingly is at your own risk. ```