Silicon & Steel Case Study Edition
FY2025 revenue
~$28B
WFE market share
~18-22%
Big-5 of WFE
~75%
CMP + implant
No. 1
Annual R&D
~$3.3B
I.The Signal

The company behind the companies.

Applied Materials is the largest maker of the machines that build chips, and most people outside the industry have never heard of it. It calls what it does materials engineering: depositing films a few atoms thick, etching them, polishing them flat, and firing dopant ions into silicon to set its electrical behavior. Nearly every transistor and memory cell on Earth passes through its tools at some point. Revenue reached roughly $28B in FY2025 (up from $27.2B in FY2024), it ranks No. 1 in wafer-fab equipment, and has been doing this since 1967.

AI made it more central, not less. The 2nm-era gate-all-around transistor, HBM's stacked through-silicon vias, backside power delivery, and advanced packaging all add new materials-engineering steps, which is exactly Applied's wheelhouse. So equipment intensity per wafer rises with every AI node, and a company that sells the picks and shovels captures that spend no matter which fab wins the customer. It is the quietest critical link in the whole chain.

Silicon wafer
A finished wafer cycles through deposition, etch, CMP and implant hundreds of times. Applied builds the tools for most of those steps. Photo: Inductiveload / Wikimedia Commons (Public domain)
II.What an Applied Materials tool actually does
Players

Applied Materials (US) the largest wafer-fab-equipment maker; deposition, etch, CMP, implant, inspection.

TSMC, Samsung, Intel, SK Hynix, Micron the fabs that buy its tools to build every logic and memory chip.

Lam Research, Tokyo Electron, KLA, ASML the other equipment giants it competes or coexists with.

VAT (vacuum valves), MKS, Advanced Energy, Edwards the subsystem suppliers Applied itself depends on.

US BIS the export-control authority that gates what Applied can sell to China.

How it moves

1. A blank silicon wafer enters the fab.

2. Deposition tools lay down ultra-thin films via physical vapor deposition (PVD), chemical vapor deposition (CVD), epitaxy, and atomic layer deposition (ALD), Applied's core franchise.

3. Etch and CMP tools carve and polish those films flat, layer by layer.

4. Ion implantation and annealing tune the electrical properties of the silicon.

5. Inspection and metrology check each step; the wafer cycles through hundreds of times before it is a chip.

Neutral process view. Applied touches most non-lithography steps; ASML owns the lithography step.

TRADE NOTE
Watch Applied's China revenue mix and its ICAPS line, not just the headline number. China has recently been ~35-45% of revenue on the mature-node buildout, so a mix normalizing toward ~25% signals export controls biting. ICAPS (its mature-node franchise for IoT, comms, auto, power and sensors) strength offsets leading-edge cyclicality. Wafer-fab-equipment spend also leads chip output by 12-18 months, so Applied's order book is an early read on the whole cycle.
III.Deep Dive: Why a 1967 toolmaker became an AI linchpin

AI is, underneath, a materials-engineering problem.

Every recent AI hardware advance shows up first as a new materials step. Gate-all-around transistors at 2nm need new epitaxy and atom-by-atom selective deposition. HBM stacks need through-silicon-via etch and deposition plus hybrid bonding (the packaging crunch we covered in The CoWoS Ceiling). Backside power delivery reroutes the wiring under the transistor. More performance now comes from more layers and more 3D integration, and each layer is another deposition, etch and CMP cycle. Applied's breadth, it touches nearly every step except lithography, means it captures that rising intensity regardless of which fab or chemistry wins, and in CMP and ion implantation it holds a near-monopoly.

The dependency runs both ways, and it is geographic. TSMC, Samsung, Intel, SK Hynix and Micron cannot build leading-edge logic or memory without Applied and its peers (Lam, Tokyo Electron, KLA, ASML), so the real chokepoint behind the foundry chokepoint is four or five equipment makers. Those five split roughly three-quarters of wafer-fab equipment between them: ASML around 22-25% (it owns lithography outright), Applied around 18-22%, Tokyo Electron 13-15%, Lam 12-13%, and KLA 6-7%, with Applied the broadest of the group and the clear leader in deposition, CMP and ion implant. Applied, in turn, books a large slice of revenue from China's mature-node expansion, even as US export controls wall its most advanced tools off from Chinese customers. That tension, China as both biggest customer and biggest restricted market, is the central question in its outlook.

Data center hall
The AI build-out at the far end of Applied's tools: a data-center hall full of accelerators. Photo: Wikimedia Commons (GFDL 1.2)
IV.What it means for the C-suite
For the CEO
If your product depends on leading-edge logic, HBM or advanced packaging, your true single points of failure are four or five equipment makers, not the fabs. Applied, Lam, Tokyo Electron, KLA and ASML are the chokepoint behind the chokepoint, and a tool-line constraint shows up in your supply 12-18 months before a chip shortage does. Map your exposure to the equipment layer, not just the foundry logo on the datasheet.
For the CFO
Applied is a clean read on the whole cycle, because wafer-fab-equipment spend leads chip output by 12-18 months. Its China mix (recently ~35-45%) and its ICAPS strength tell you whether 2026 capex is broadening across mature nodes or concentrating at the leading edge. Model the export-control tail explicitly: a scenario where advanced-tool revenue into China goes to zero is the single largest swing factor in its, and the sector's, numbers.
For the CSCO
Applied's own bottleneck is subsystems, not silicon: VAT vacuum valves (Switzerland), MKS and Advanced Energy power and RF (US), Edwards vacuum pumps (UK). A tool ships only as fast as its scarcest subsystem. If your plan depends on a fab capacity expansion, the lead time that actually bites is the equipment maker's component base, so push for tool-level, not just cleanroom-level, delivery visibility in your supplier reviews.
V.Where Applied stands among the giants
Positioned to win  ▲
Applied itself, on the broadest portfolio in the industry, CMP and ion-implant leadership, and a large ICAPS franchise that steadies the cycle. The subsystem suppliers (VAT, MKS, Advanced Energy, Edwards) riding the same WFE growth. And peers Lam (etch and deposition, NAND and HBM), Tokyo Electron, and KLA (process control), all structurally lifted by AI capex.
Under pressure  ▼
Applied's leading-edge China revenue, sitting under tightening US export controls. Any single niche where a focused rival out-specializes it, ASML in lithography, KLA in inspection. And any chipmaker single-sourced on one tool vendor for a critical step it can't second-source quickly.

The startup opening. The opening is in the gaps the giants underserve: metrology and yield-analytics software layered on top of the tools, used-tool refurbishment and second-source subsystems to ease the component bottleneck, and specialty materials and precursors for ALD and selective deposition. It is picks-and-shovels for the picks-and-shovels, a smaller market that the majors don't prioritize and that gates their own throughput.

VII.Glossary

WFE: wafer-fab equipment, the machines that turn blank wafers into chips. Applied is the largest maker of them by revenue.

Deposition: laying down films a few atoms thick on a wafer (by physical vapor deposition (PVD), chemical vapor deposition (CVD), epitaxy, or atomic layer deposition (ALD)). Applied's core franchise.

CMP: chemical mechanical planarization, polishing each layer perfectly flat before the next is built. Applied's near-monopoly franchise.

Ion implantation: firing dopant ions into silicon to set its electrical properties. Applied dominates this step too.

GAA: gate-all-around, the 2nm-era transistor that wraps the gate fully around the channel, needing new deposition and etch steps.

ICAPS: Applied's name for IoT, Communications, Auto, Power and Sensors, the mature-node markets that are a big, steadier chunk of its revenue.

VIII.The 'Applied' Confusion

Okay, quick story, because it’s half the reason I wrote this.

There are about six public companies named “Applied [something],” and they do completely different things:

Applied Materials: the machines that make the chips (what you just read).
Applied Intuition: self-driving-car software.
Applied Digital: data centers and AI infrastructure.
Applied Optoelectronics: fiber-optic networking gear.
Applied Industrial Technologies: industrial parts distribution.
Applied DNA Sciences: molecular tagging and biotech.

This actually happened a few weeks ago. A founder I admire (he runs a big AI-infrastructure startup) asked me directly: what do you make of Applied Materials? Easy, I thought, and launched in, full confidence, simulation stacks, sensor suites, the entire self-driving roadmap. He nodded along for a while, then said, “I think this might be a different Applied company.”

He was right. I’d been talking about Applied Intuition the whole time. To his credit, he let me finish before pointing it out, which somehow made it worse.

But hey, now we all know there are six of these. So these days I just ask which Applied first. Lesson learned.

One honest admission.
The physical supply chain behind AI is changing faster than any one person can track, and I get things wrong. If you work in this space and I've missed something (or flat-out botched it), just reply and tell me. Better yet, if you're in the Bay Area, let's grab a coffee. This newsletter is me thinking out loud, and it's far better when you think back.
Silicon & Steel Intelligence Desk · Supply Chain Strategy & Semiconductor Analysis.
Corrections & coffee: [email protected]

Reply

Avatar

or to participate

Keep Reading