This piece breaks down Applied Materials, the world's largest chip-making equipment maker. The author argues it may be more undervalued than ASML because it has several 'mini-monopolies'—like CMP (chemical mechanical planarization, polishing wafers to atomic precision) where it holds 90% share. Each is small but hard to replace. Key holdings: Applied Materials itself is favored due to high switching costs for clients like TSMC; ASML is compared as a simpler but more expensive story; Chinese equipment makers (e.g., Naura) are flagged as a long-term risk, having gained share in lower-end areas.
Applied Materials is the world's largest manufacturer of chipmaking equipment, with 2022 revenue of $26 billion, R&D spending of $3 billion, and 17,300 patents. This episode is hosted by Zack Fuss of Irenic Capital, featuring SemiAnalysis Chief Analyst Dylan Patel, who delves into the evolution of t
Dylan Patel (Chief Analyst at SemiAnalysis) and Zack Fuss of Irenic Capital conduct a deep dive into Applied Materials, the world's largest semiconductor equipment manufacturer. The core thesis: By focusing on materials engineering ("sanding atoms") rather than lithography, the company has established dominance in critical processes such as deposition and etching, with deep collaboration with manufacturers creating extremely high switching costs. The most significant takeaway from the episode: Dylan Patel believes Applied Materials may be more undervalued than ASML — because it holds six or seven "mini-monopolies," each accounting for only 3-5% of wafer fab equipment spending, yet collectively forming a stable and hard-to-replace market share.
Dylan Patel argues that the core trend in the semiconductor industry over the past 50 years has been "increasingly refined division of labor," and Applied Materials is a beneficiary of this trend.
Historical Context: In the 1950s-1970s, companies like IBM and Intel were vertically integrated—they designed chips, manufactured equipment, produced chips, and assembled computers themselves. Patel notes: "Intel was one of the first to break apart—it said, 'We only sell chips, not systems.'" Subsequently, TSMC led an even more radical separation: focusing solely on foundry services without design, giving rise to fabless companies such as NVIDIA and Apple. Later, EDA (Electronic Design Automation) software firms like Cadence and Synopsys became independent, enabling more players to design chips.
Mechanism Breakdown: The essence of this division of labor is Adam Smith's principle from The Wealth of Nations—players at each stage can become more focused and innovative. Patel emphasizes: "Small players are always more agile and innovative than large conglomerates." Applied Materials seized the opportunity at the equipment manufacturing layer, becoming the world's largest equipment supplier.
Data Chain: Currently, 7-8 of the Fortune 10 companies design their own chips; globally, 8-9 equipment firms have annual sales exceeding $10 billion, but the largest ones (Applied Materials, ASML, Lam Research, KLA) generate annual revenues of over $20 billion.
Dylan Patel argues that Applied Materials' business model is more complex and harder for the market to fully understand than ASML's — it holds six or seven "small monopolies," each individually modest but collectively forming a deep moat.
Mechanism Breakdown: ASML's core is lithography, which accounts for roughly 20% of wafer fab equipment spending and is the sole supplier for the most advanced process nodes — a clear and simple story. In contrast, Applied Materials' product portfolio spans deposition, etching, CMP (chemical mechanical planarization), ion implantation, annealing, PVD, and other categories, each representing only 3-5% of fab spending. Yet Applied Materials holds a 70-90%+ market share in most of these categories.
Specific Case — CMP (Chemical Mechanical Planarization): Patel uses a vivid analogy: "CMP is like polishing a wafer — not to nanometer precision, but to sub-nanometer precision, even to atomic thickness." Applied Materials commands roughly a 90% share in this category and is nearly the sole supplier for the most advanced nodes (5nm and below). However, CMP accounts for less than 3% of total wafer fab equipment spending.
Switching Cost Analysis: Why doesn't TSMC replace Applied Materials? Patel offers a sharp explanation: TSMC may have only about 20 expert engineers dedicated to CMP. Their options are: (A) spend time researching how to replace Applied Materials' equipment with that of a Japanese second-tier supplier, or (B) spend time optimizing existing equipment to improve yield and performance. "Do you want to make Apple happy — because you produced 5% more chips — or do you want to spend time swapping equipment?" The answer is self-evident. Moreover, second-tier suppliers have not participated in the joint R&D of past process generations (14nm → 10nm → 7nm → 5nm), so TSMC would need to "hand-hold" them through all the accumulated technical know-how — a near impossibility.
Data Chain: Applied Materials' gross margin is approximately 45% (slightly above its long-term average), with service revenue accounting for 25% of total revenue and carrying higher margins. R&D spending is about $3 billion per year, representing 13-14% of revenue. Operating margins range between 20-30% (reaching 30% in 2021-2022). Revenue has grown at an average annual rate of over 10% over the past decade, and about 15% over the past five years. EPS growth is close to 20%, partly driven by the buyback of roughly 30% of outstanding shares over the past eight years.
Dylan Patel argues that Applied Materials has successfully reduced the inherent cyclicality of the semiconductor equipment industry through its services business and counter-cyclical buybacks.
Mechanism Breakdown: The semiconductor equipment industry is highly cyclical — Applied Materials saw revenue drop 35% during the 2008 crisis and 20% in 2018. However, the company has three layers of buffers:
1. Services Business (approximately 25-30% of revenue): Customers must maintain purchased equipment, otherwise the equipment loses value. The services business is relatively stable during economic downturns and has higher gross margins than equipment sales.
2. Counter-Cyclical Buybacks: Patel notes: "Every downturn, they aggressively buy back shares — 5%, 6%, 7% of outstanding shares." This was the case in 2008, 2018, and 2022. When the market recovers, EPS is permanently enhanced.
3. Sustained R&D Investment: Even during troughs, R&D as a percentage of revenue actually increases (from 15% to over 20%). Patel emphasizes: "They always take a long-term view — continuing deep R&D collaboration with customers, so when the market rebounds, they are more deeply embedded."
Data Chain: Since the dot-com bubble, Applied Materials has never experienced two consecutive years of revenue decline. Approximately 80% of operating cash flow is returned to shareholders.
Dylan Patel believes that geopolitical tensions are a "double-edged sword" for Applied Materials—short-term demand is boosted by subsidies, but long-term risks arise from the catch-up efforts of Chinese domestic equipment makers.
Positive Effect – Wave of Subsidies: Governments worldwide are heavily subsidizing semiconductor localization:
Patel notes: "All this money will ultimately flow to the 'shovels and picks'—that is, Applied Materials." Subsidies make the industry's downcycle smoother—even when orders are weak, subsidy funds continue to drive equipment procurement.
Negative Effect – Chinese Domestic Substitution: Chinese equipment companies (such as Naura, ACMR, etc.) are catching up rapidly. While they have not yet succeeded in high-end areas like lithography and dry etching, they have already gained significant market share in lower-end segments such as cleaning. Patel warns: "Chinese equipment makers may compete with 'irrational' pricing—unreasonable under Western logic, but perfectly rational under China's subsidy logic." If newly built Chinese fabs (accounting for 25-30% of global new capacity) heavily adopt domestic equipment, Applied Materials will lose that market.
Reader Note: As an industry analyst, Patel's assessment is based on public data and industry observations. However, from the perspective of a position holder in Applied Materials, there may be a tendency to downplay long-term risks and emphasize near-term positives. The true competitiveness of Chinese equipment makers still requires time to verify.
Dylan Patel believes that AI is the strongest structural growth driver for Applied Materials' future — because AI chips require the most advanced process nodes, and advanced process nodes require more equipment.
Mechanism Breakdown: Patel points out: "You cannot manufacture NVIDIA's AI chips without Applied Materials' equipment." The demand for semiconductors driven by AI manifests at two levels:
1. Volume Growth: AI training and inference require a large number of GPUs, HBM memory, networking chips, and more.
2. Process Node Upgrades: From 5nm to 3nm, process steps increase by approximately 30%. This means each wafer requires more equipment processing time, directly boosting equipment demand.
Historical Analogy: Patel notes that equipment companies often spot trends earlier than the industry — Applied Materials anticipated demand ahead of time during both the PC era and the mobile era. Now, they are discussing AI as "the next era of the semiconductor industry."
Data Chain: From 5nm to 3nm, process steps increase by about 30%, and equipment modifications are involved. The three major customers — TSMC, Intel, and Samsung — collectively spend over $30 billion annually on capital expenditures, of which approximately 80% is allocated to equipment procurement.
| Position | Guest Stance | Key Data |
|---|---|---|
| Applied Materials | Bullish (long-term growth + high switching costs + excellent capital allocation) | Revenue $26B, R&D $3B, 17,300 patents; gross margin ~45%, operating margin 20-30%; EPS growth ~20%; repurchased ~30% of outstanding shares over the past 8 years |
| ASML | Neutral comparison (clearer business but more concentrated) | Lithography accounts for ~20% of wafer fab equipment spending; market cap higher than Applied Materials but revenue lower |
| TSMC | Key customer/partner | Annual capex $30B+; joint R&D with equipment makers; gross margin 50-60% |
| Intel | Risk warning (failed due to neglecting equipment maker collaboration) | 10nm process failure case — insufficient collaboration with equipment makers |
| Samsung | Key customer | One of the top three equipment buyers alongside TSMC and Intel |
| Lam Research | Competitor (in certain areas) | Etch equipment market share ~15%; competes with Applied Materials in some sub-segments |
| Tokyo Electron | Unsuccessful M&A target | Japan's largest equipment maker; Applied Materials attempted a merger but was blocked |
| NVIDIA | AI demand-driven position | AI chips rely on Applied Materials equipment for manufacturing |
| Chinese equipment makers (Naura, ACMR, etc.) | Long-term risk | Have gained market share in low-end areas such as cleaning; supported by Chinese subsidies |
1. Dylan Patel believes Applied Materials is more undervalued than ASML — because it holds six or seven "small monopolies," each accounting for only 3-5% of wafer fab equipment spending, but collectively forming a stable and hard-to-replace position. ASML's story is clearer, but Applied Materials' moat is more fragmented and less visible.
2. "CMP is polishing atoms" — Patel uses this analogy to explain Applied Materials' core business: Chemical mechanical planarization (CMP) polishes wafers to sub-nanometer precision, and Applied Materials holds a 90% share in this area. This analogy vividly illustrates the fundamental difference between "materials engineering" and "lithography."
3. The core of switching costs is not technology, but "engineer time": Patel points out that TSMC may have only about 20 expert engineers in the CMP field. Asking them to spend time studying how to replace equipment, rather than optimizing existing equipment to improve yield, is economically unviable. "Do you want to make Apple happy — because you produced 5% more chips, or do you want to spend time swapping equipment?"
4. "Six Sigma will kill you in semiconductor manufacturing": Patel emphasizes that semiconductor manufacturing involves thousands of process steps. Applying Six Sigma's defect rate (3.4 per million) across thousands of steps would drive yield to zero. Therefore, equipment must achieve precision far beyond Six Sigma — this is the core barrier for equipment makers.
5. Counter-cyclical buybacks are Applied Materials' "hidden weapon": The company aggressively repurchases shares (5-7% of outstanding shares) during every downturn (2008, 2018, 2022), permanently boosting EPS when the market recovers. Patel states: "They manage the cycle, rather than being managed by it."
6. Geopolitics is a "double-edged sword" — short-term positive, long-term risk: National subsidies (US $52 billion + $30 billion in tax credits, Japan $30-40 billion, Europe $40 billion, China $200-250 billion) drive equipment demand in the short term; however, Chinese equipment makers may compete at "irrational" prices, eroding Applied Materials' market share over the long term.
7. From 5nm to 3nm, process steps increase by 30% — meaning each wafer requires more equipment processing time, directly boosting equipment demand. AI chips' insatiable appetite for advanced nodes is the strongest structural growth driver for Applied Materials.
8. "Equipment companies see trends earlier than the industry": Patel notes that Applied Materials anticipated demand ahead of the PC era and the mobile era — because customers like Intel and Apple need to place orders in advance. Now, they are discussing AI as the "next era for the semiconductor industry."