Semiconductor Equipment Stocks: ASML, Lam Research, Applied Materials

Why Semiconductor Equipment Is the Highest-Conviction AI Trade

The AI investment landscape is littered with uncertainty. Will Nvidia maintain GPU dominance or will custom ASICs erode its moat? Will OpenAI or Anthropic or Google win the foundation model race? Will enterprise AI adoption meet the $200 billion in projected 2026 spending, or is this a classic capex overbuild?

Semiconductor equipment sidesteps all of these questions. Regardless of which AI chip wins, which model architecture prevails, or which hyperscaler captures the most market share, every single AI chip must be manufactured using lithography, etch, deposition, and inspection equipment from the same small group of companies. TSMC cannot produce Nvidia's B200 without ASML lithography. Samsung cannot fabricate HBM3E without Lam Research etch tools. Intel cannot ramp its foundry business without Applied Materials deposition systems.

The numbers back this up. Global wafer fab equipment (WFE) spending reached approximately $105 billion in 2024, up from $76 billion in 2022 and $56 billion in 2020. The Semiconductor Industry Association projects WFE spending will exceed $130 billion by 2027. This is not speculative demand — these are contractual orders backed by capital commitments from TSMC ($32 billion annual capex), Samsung ($45 billion cumulative through 2027), Intel ($100 billion in fab investments under the CHIPS Act), and a wave of new fab construction across the US, Japan, and Europe.

Here is what most generalist investors miss: the semiconductor equipment market is an oligopoly with staggering barriers to entry. It is not a competitive market. It is five companies with effective monopolies in their respective process steps, selling machines that take decades of R&D to develop and cannot be replicated. For investors who studied the Nvidia AI dominance thesis, equipment stocks represent the layer below — the infrastructure that makes Nvidia's dominance possible.

ASML: The Most Important Technology Company Most Investors Ignore

ASML Holding, headquartered in Veldhoven, Netherlands, is the sole manufacturer of extreme ultraviolet lithography systems. This is not market leadership. It is a monopoly in the strictest definition of the word. There is no second source. There is no alternative technology. Every advanced semiconductor produced anywhere on Earth — by TSMC, Samsung, or Intel — passes through ASML's machines.

An EUV lithography system is arguably the most complex machine ever built by humans. The light source generates 13.5nm wavelength EUV radiation by firing a high-powered CO2 laser at molten tin droplets 50,000 times per second. The resulting plasma emits EUV light that is collected by multilayer mirrors (polished to sub-angstrom precision) and focused through a pattern mask onto a silicon wafer. Each system contains over 100,000 components, weighs 180 tonnes, ships in 40 freight containers, and requires six months to install and calibrate. ASML's latest machine, the TWINSCAN EXE:5000 (High-NA EUV), costs approximately $380 million per unit.

ASML's financial profile reflects its monopoly position. Revenue reached approximately $30 billion in 2024, with gross margins of 51%. The order backlog exceeded $40 billion at year-end 2024, providing roughly 18 months of revenue visibility. ASML has guided for revenue of $44–60 billion by 2030, implying 10–15% annual growth from a base that is already enormous.

The High-NA Inflection

High-NA EUV is ASML's next growth chapter. By increasing the numerical aperture of the EUV optics from 0.33 to 0.55, High-NA enables finer feature resolution needed for 2nm and sub-2nm chip nodes. Intel has taken delivery of the first High-NA system for its 18A process node. TSMC is evaluating High-NA for its N2 and A14 nodes. Each High-NA tool costs 2.5x more than standard EUV — approximately $380 million versus $150 million — meaning the ASP uplift per tool dramatically increases ASML's revenue per wafer start.

The bear case on ASML centers on China. The Dutch government, under US pressure, restricted ASML from shipping any EUV systems to China since 2019, and expanded restrictions to include advanced DUV tools in 2024. China represented roughly 29% of ASML's 2024 revenue through unrestricted DUV orders, and this percentage will decline as restrictions tighten. We believe the market has already priced this headwind — ASML's $40B+ backlog is entirely from non-China customers — but any escalation in export controls could create near-term selling pressure.

Lam Research, Applied Materials, and KLA: The Rest of the Oligopoly

Lam Research (LRCX) — The HBM Beneficiary

Lam Research dominates etch and deposition, the two process steps that have the highest growth leverage to AI chip manufacturing. Etch is the process of selectively removing material from a wafer to create circuit patterns, while deposition adds thin films of material. These steps are performed hundreds of times per advanced chip, and the number of steps is increasing with every new node.

The real catalyst for Lam is High Bandwidth Memory. HBM3E, the memory type used in Nvidia's H100 and B200, is manufactured by stacking 8–12 layers of DRAM dies using through-silicon vias (TSVs). This stacking process requires 2–3x more etch steps than conventional 2D DRAM. SK Hynix, Samsung, and Micron are all racing to expand HBM capacity, and Lam supplies roughly 45–50% of the global etch equipment market. Every new HBM fab is a massive etch equipment order for Lam.

Lam generated approximately $17 billion in revenue in fiscal 2025 (ending June), with operating margins of 30%. The stock trades at roughly 24–26x forward earnings — cheaper than ASML despite comparable growth driven by the HBM and advanced packaging supercycle. We believe Lam offers the best risk-reward in the equipment space for investors specifically seeking AI leverage.

Applied Materials (AMAT) — The Diversified Giant

Applied Materials is the largest semiconductor equipment company by revenue at approximately $27 billion annually. Its product portfolio is the broadest in the industry, spanning etch, chemical vapor deposition (CVD), physical vapor deposition (PVD), chemical mechanical planarization (CMP), ion implantation, and inspection. This diversification provides exposure to every semiconductor end market — logic, memory, display, and solar — making Applied less dependent on any single technology transition.

Applied's specific AI tailwinds include: advanced packaging equipment for chiplet architectures (a market Applied pioneered with its hybrid bonding tools), gate-all-around (GAA) transistor technology for sub-3nm nodes (which requires fundamentally different deposition equipment than FinFET), and backside power delivery networks being adopted by Intel and TSMC. Each of these transitions increases the number of Applied Materials process steps per chip.

At 22–24x forward earnings, Applied Materials is the cheapest large-cap semiconductor equipment stock. The discount reflects its lower growth rate relative to ASML and Lam (Applied is expected to grow revenue 8–12% annually versus 12–15% for Lam), but the valuation floor is supported by consistent free cash flow generation, aggressive buybacks ($7 billion authorized), and a diversified customer base.

KLA Corporation (KLAC) — The Quality Inspector

KLA is the dominant player in semiconductor process control and defect inspection, with roughly 55% global market share. As chips become smaller and more complex, the probability of manufacturing defects increases exponentially. A single particle on a wafer during EUV lithography can ruin an entire die. KLA's systems detect these defects at the nanometer scale, enabling fabs to maintain yield and reduce waste.

KLA's competitive moat is underappreciated. Process control is not just about hardware — it requires decades of proprietary defect databases and machine learning algorithms trained on trillions of data points from every major fab in the world. Switching costs are enormous because customers would lose their entire defect library and process calibration history. This is why KLA sustains operating margins above 40% and has grown revenue at 15% CAGR over the past five years.

The AI connection: as advanced packaging and 3D chip stacking become standard (HBM, chiplets, system-on-wafer), the number of inspection points per package multiplies. KLA estimates that advanced packaging inspection is a $3 billion+ incremental opportunity by 2028, growing from a near-zero base in 2022.

The Structural Tailwinds: Why This Cycle Is Different

Semiconductor equipment has historically been a notoriously cyclical business. WFE spending collapsed 30% in 2009, 20% in 2019, and 15% in 2023. Equipment stocks regularly traded at 12–15x earnings through the cycle. Today, they trade at 22–32x. Are investors simply paying too much, or has something structurally changed?

We believe the latter, for three reasons.

First, the number of equipment customers has shrunk to a handful of companies that are all investing simultaneously. At the leading edge (sub-5nm), there are exactly three customers: TSMC, Samsung, and Intel. All three are executing multi-year, multi-hundred-billion-dollar fab buildout programs driven by AI demand, CHIPS Act subsidies, and supply chain resilience requirements. This concentration reduces the demand volatility that historically caused boom-bust cycles — three sophisticated customers with long-term investment plans create more predictable order patterns than hundreds of smaller buyers.

Second, the CHIPS Act and its equivalents in Europe, Japan, South Korea, and India are injecting an estimated $380 billion in government subsidies into semiconductor manufacturing over the 2023–2032 period. This is unprecedented. Every dollar of subsidy translates directly to equipment orders. TSMC's Arizona fab, Intel's Ohio and Arizona fabs, Samsung's Texas fab, and TSMC's Kumamoto fab in Japan — all CHIPS Act or government-subsidized projects, all requiring billions in equipment purchases.

Third, manufacturing complexity is increasing faster than Moore's Law would suggest. The transition from FinFET to gate-all-around transistors, the adoption of backside power delivery, the move to chiplet architectures with advanced packaging, and the scaling of HBM all increase the number of equipment-intensive process steps per chip. ASML estimates that revenue per wafer start has increased approximately 2x over the past five years, and will increase another 50% by 2030. The equipment content per dollar of chip revenue is structurally rising.

The Bear Case: Cyclicality, China, and Concentration

We hold equipment stocks as high-conviction positions, but intellectual honesty requires addressing the material risks.

Cyclicality is not dead — it is dampened. Even in the best scenario, semiconductor equipment spending will have up and down years. The 2023 downturn, where WFE spending declined approximately 15%, demonstrated that AI demand alone cannot prevent cyclical corrections when memory investment and legacy node spending pull back simultaneously. Equipment stocks fell 25–40% from their 2023 peaks before recovering. Investors should expect similar drawdowns even within a structurally positive cycle.

China export restrictions remain an overhang. The US has progressively tightened semiconductor equipment exports to China, and further escalation is possible. If restrictions expand to all DUV lithography tools (not just advanced models) or mature-node etch and deposition equipment, the revenue impact could be significant — potentially 10–15% of total revenue for Lam, Applied, and KLA. The political trajectory favors escalation, not relaxation.

Customer concentration is a double-edged sword. Three customers account for 70%+ of leading-edge equipment demand. If any one of them delays or cancels fab projects — as Intel did with its Ohio fab timeline in 2024 — the impact on equipment orders is immediate and material. Intel's foundry struggles, in particular, represent a risk to the equipment demand outlook: Intel was expected to be the third major EUV buyer, and any reduction in Intel's fab investment directly reduces ASML, Lam, and Applied's addressable market.

For a deeper analysis of how custom chip development affects equipment demand, see our coverage of custom AI chips versus Nvidia GPU investment.

Portfolio Construction: Building a Semiconductor Equipment Basket

We recommend a basket approach rather than single-stock concentration for semiconductor equipment exposure. The rationale is simple: these companies have correlated demand drivers but differentiated risk profiles, and the basket approach captures the sector tailwind while reducing single-stock event risk.

Our recommended allocation for a semiconductor equipment basket: ASML 35% (monopoly position justifies the largest weight despite the premium valuation), Lam Research 25% (best AI-specific leverage through HBM), Applied Materials 20% (diversification and valuation support), KLA 15% (quality compounder with margin expansion runway), Tokyo Electron 5% (Japan fab buildout exposure, but more complex for US investors to access).

For investors who prefer a single vehicle, the VanEck Semiconductor ETF (SMH) has approximately 12% combined weight in ASML, Lam, Applied, and KLA, though its 25%+ Nvidia weight means it is not a pure equipment play. The iShares Semiconductor ETF (SOXX) has a more balanced equipment weighting but includes many non-equipment semiconductor companies. There is no pure-play semiconductor equipment ETF — which is arguably an opportunity, given the sector's quality and growth characteristics.

The 2030 Outlook: $150 Billion in Annual Equipment Spending

ASML has guided for $44–60 billion in revenue by 2030, up from $30 billion in 2024. If ASML achieves the midpoint of $52 billion, and its revenue represents roughly 35% of total WFE (consistent with historical share), that implies total WFE spending of approximately $150 billion by 2030 — a 40%+ increase from 2024 levels.

This trajectory would drive the following revenue estimates: ASML at $50B+, Applied Materials at $35B+, Lam Research at $22B+, KLA at $14B+. At current valuations, these revenue levels would imply 60–100% upside for the equipment basket over the 2026–2030 period, assuming margins hold or expand (which we expect, given operating leverage and pricing power).

The contrarian risk is that AI capex proves to be a bubble and the hyperscalers dramatically cut back chip orders. We assign this a low probability (15–20%) because AI inference demand is still in its infancy, and the CHIPS Act subsidies provide a multi-year floor for fab construction regardless of near-term demand fluctuations. But if the bear case materializes, equipment stocks could see 30–40% drawdowns from current levels.

For investors monitoring the AI infrastructure buildout more broadly, our analysis of the AI capex boom and where smart money is investing provides the broader framework that connects chip demand to equipment orders to fab construction timelines.

Frequently Asked Questions