AMAT — Research Report (2026-07-16)

TL;DR

• What it is: Applied Materials is the world's broadest maker of chip-manufacturing machines — the "picks and shovels" supplier of the AI boom.
• Take: the business has never been better; the stock price already assumes it stays that way for years. Franchise = world-class. Entry price = the risk.
• $28.4B FY25 revenue, +4% (6th straight growth year — durable, not hyper-growth)
• 50% gross margin in Q2 FY26 (best in ~25 years = pricing power)
• ~54× trailing P/E vs peer bull-cycle median ~28–32× (valuation carries the risk)
• ~1.2% FCF yield (thin cash return at this price)
• Strongest bull: Susquehanna's $900 target — AI equipment supercycle, company guiding >30% equipment growth in CY2026.
• Strongest bear: Michael Burry shorted it at $729 in June 2026 — dot-com-like sector multiples plus memory-cycle risk.
• China overhang: mid-20s% of revenue, ~$600M FY26 revenue hit from export controls, $252.5M fine settled.
• Watch next: fiscal Q3 earnings ~Aug 13, 2026, and any sign of HBM/memory capex cuts.

1. What Applied Materials is

So what: AMAT sells the machines that make chips — to nearly every chipmaker on earth — plus a growing service annuity on top.

Applied Materials (NASDAQ: AMAT, founded 1967, HQ Santa Clara) builds wafer fabrication equipment ("WFE" — the machines inside a chip factory that deposit, etch, and modify the atomic-scale layers of a semiconductor). Its portfolio is the industry's broadest: deposition (laying down thin films of material), etch (removing material precisely), ion implant, CMP (polishing), and inspection. If TSMC, Samsung, or Intel builds a fab, a large share of the machines inside come from AMAT.

Revenue split three ways in FY2025 (ended Oct 2025): Semiconductor Systems ≈ $20.3B (~72%) — the tools; Applied Global Services ≈ $6.4B (~23%, a record, +3%) — spare parts, service contracts, and upgrades on the installed base; Display and corporate ≈ $1.7B (derived remainder). Figures per press coverage of the FY25 release (GlobeNewswire, Futurum).

Customers are concentrated because chipmaking is: the FY25 10-K reports two customers at roughly 19% and 15% of revenue (historically TSMC and Samsung sit at the top), and ~89% of revenue comes from outside the US (10-K summary). China was ~40% of revenue at its peak and is now in the mid-20s% after US export restrictions (Yahoo Finance / Reuters).

2. Financials — five years

So what: steady mid-single-digit growth, expanding margins, heavy cash generation — with FY25 free cash flow dented by a deliberate doubling of capex.

All figures from SEC EDGAR companyfacts (fiscal years end late October).

Revenue CAGR FY21→FY25 is 5.3% — a compounder, not a rocket; the rocket phase started in FY26. Two flags: capex nearly doubled in FY25 (management has tied the step-up to its EPIC R&D center build-out — spending ahead of the AI cycle), and net income dipped ~2% despite record revenue. Separately, AMAT agreed in February 2026 to pay $252.5M to settle a US export-controls case over shipments that reached China's SMIC (Supply Chain Dive, Arnold & Porter).

The balance sheet is a non-issue: cash ($7.2B) roughly covers long-term debt ($6.5B) before counting the investment portfolio — effectively net cash. Backlog stood at $15.0B at FY25 year-end, split evenly between systems ($7.1B) and services ($7.1B) (10-K summary).

The FY26 acceleration (why the stock roughly quadrupled off its 52-week low): fiscal Q2 2026 (April) delivered record revenue of $7.91B, +11% YoY, at 50% gross margin; guidance for Q3 is $8.95B (+23% YoY) and EPS $3.36 (+36%). Management expects its semiconductor equipment business to grow more than 30% in calendar 2026, with advanced packaging (stacking chips together — the key enabler of AI memory) growing over 50% (Motley Fool transcript, 10-Q summary).

3. Key metrics — and why each matters here

So what: quality metrics are elite across the board; the two ugly numbers are both valuation numbers.

4. Unit economics — the number the business runs on

So what: AMAT's unit economics are "gross margin per tool dollar," plus a service annuity that cushions the cycle.

A leading-edge tool sells for millions of dollars, and the gross margin on it — now 50 cents per revenue dollar — is the purest read of pricing power: customers (even giants like TSMC) keep paying more per tool because each new chip generation needs materials-engineering steps only one or two vendors on earth can perform. That margin climbing from 46.5% (FY22) to 50% (Q2 FY26) while revenue grew is the signature of a bottleneck supplier, not a commodity one.

The second engine: every tool sold joins an installed base that generates services revenue for decades. Applied Global Services did a record $6.4B in FY25 and holds a $7.1B backlog — equal to the systems backlog. Services are the annuity that keeps cash flowing when the equipment cycle turns down; think of it as the "take rate" AMAT collects on its own installed base, forever.

5. Valuation

So what: a world-class business trading at roughly 1.5–2× its sector's normal bull-cycle multiple — you are paying today for 2027–2028 staying perfect.

P/E is the right lens here (profitable, effectively net cash, so enterprise value ≈ market cap); EV/EBITDA is shown as a cross-check. The whole sector trades above its own history — ChartMill notes a group average P/E above 60 — and Burry's short thesis is precisely that the semis index sits at 2000-vintage extension levels.

What a reasonable buyer might pay: nobody acquires a $475B company, so the question is what a long-term investor underwrites. At the sector's bull-cycle median (~30×) on ~$14.8 of implied forward EPS, fair value is ~$445/share; the Street's targets after the June "Master Class" event run $647–$900. Today's $605 (market cap $474.7B, Yahoo Finance, Jul 15) only works if the AI equipment supercycle runs uninterrupted into 2027+ — underwrite the multiple-compression scenario first, and enter in tranches, not on momentum.

6. Value chain: bottleneck or commodity?

So what: a genuine oligopoly bottleneck — one notch below ASML's monopoly — and its real overlord is the US government, not any customer.

The WFE market is a five-company oligopoly (AMAT, ASML, Lam Research, Tokyo Electron, KLA) where each firm dominates different steps; AMAT is #1 in breadth across deposition/etch/implant. In many process steps AMAT is sole-source or one of two — that's why 50% gross margins stick even against customers as powerful as TSMC and Samsung. It is not, however, an ASML-grade chokepoint: ASML's EUV lithography machines (the light-based pattern-printing step) have zero substitutes, while most AMAT categories have at least one credible competitor.

Who has pricing power over AMAT? Not suppliers, not customers — Washington. Export controls have already cut China from ~40% to mid-20s% of revenue, imposed a ~$600M FY26 revenue hit, banned sales into China's memory and older-node markets, and extracted a $252.5M settlement (TrendForce, Reuters via Yahoo). Regulatory policy is this company's single biggest uncontrollable variable.

M&A track record (a caution flag): Varian Semiconductor (2011, $4.9B) worked; the Tokyo Electron mega-merger died on antitrust objections (2015); the Kokusai Electric deal ($3.5B) was terminated in 2021 after China's regulator never approved it — evidence that AMAT can no longer buy growth at scale, and that China holds leverage back.

7. Ownership, management, insider signals

So what: professionally managed, index-fund owned, and insiders are sellers into the melt-up — not a red flag alone, but zero insider buying at these prices is worth knowing.

There is no controlling shareholder; ownership is dominated by passive index managers (details in the DEF 14A proxy on EDGAR). CEO Gary Dickerson has run the company since 2013 — the margin expansion and services build-out happened on his watch.

Insider signal: Dickerson sold 161,321 shares over the past year and bought zero, including ~83,000 shares at $590–600 in mid-June 2026 and 78,321 more on June 30 ($55M across filings) (GuruFocus, StockTitan Form 4). Capital returns remain aggressive: $6.3B returned in FY25, ~$14B of buyback authorization outstanding, and eight straight years of dividend increases (StockTitan).

8. Bull / Base / Bear — named views

So what: the Street is bullish on fundamentals; the loudest bear is short the multiple, not the machines.

🐂 Bull — Susquehanna (Mehdi Hosseini), $900 PT, Street-high; Cantor Fitzgerald $850; B. Riley $790 Buy (Jun 26); Jefferies $770; KeyBanc $750 (Jun 29). AI datacenter buildout forces years of leading-edge, HBM, and advanced-packaging capacity additions; AMAT guides >30% equipment growth in CY26 with packaging +50%, at record 50% margins (TIKR, GuruFocus/KeyBanc, Yahoo/B. Riley).

⚖️ Base — Morgan Stanley (Shane Brett), Equal-Weight, PT $647. Downgraded from Overweight on May 18, 2026 citing risk of DRAM equipment-spend revisions in H2; later raised the target to $647 (from $502) while staying neutral — great company, full price, memory-cycle timing risk (Benzinga, TIKR). Citi sits between base and bull at $710.

🐻 Bear — Michael Burry (Scion), disclosed short at $729.40 on June 30, 2026. Thesis: the semis index trades at 2000-bubble extension levels; AMAT at ~41× forward earnings is priced above faster-growing chip designers; an SK Hynix signal of slowing HBM expansion is the pin. Structural kicker: US export controls keep shrinking the China market (TIKR, TrendForce). No major sell-side house currently publishes a Sell rating — the institutional bear case is expressed through short positioning and neutral ratings, not downgrades.

9. Catalysts & what would prove the thesis wrong

So what: the next two quarters of memory-capex news decide whether this is mid-supercycle or the top.

Catalysts: fiscal Q3 earnings (~Aug 13, 2026) vs the $8.95B guide · HBM/DRAM capex announcements from SK Hynix, Micron, Samsung · any loosening or tightening of China export rules · advanced-packaging order momentum · continued gross-margin prints ≥50%.

The bull thesis breaks if: memory makers cut 2027 capex (watch SK Hynix first — Burry's pin) · gross margin slips back below ~48% (pricing power was cyclical, not structural) · backlog shrinks two quarters in a row · China ICAPS demand falls faster than the "flat to slightly up" company guide (Motley Fool transcript) · Washington broadens controls beyond memory and legacy nodes.

📚 What this company teaches

1. Sell shovels in a gold rush — but grade the shovel-sellers. Equipment oligopolies capture margin no matter which chipmaker wins the AI race: AMAT wins whether the leader is TSMC, Samsung, or Intel, which is structurally safer than picking the winning chip. But note the gradient within the layer: ASML (monopoly) > AMAT/Lam/KLA (oligopoly) > commodity toolmakers.
2. A great business and a great stock are different objects. Every fundamental here is elite (34% ROE, 50% GM, net cash) and the stock can still be dangerous, because a 1.2% FCF yield means valuation — not operations — carries the risk. The Burry short versus the $900 bull target is a disagreement about price; both sides agree the machines are magnificent.
3. Geopolitics is a P&L line, not a headline. For semicap, export policy behaves exactly like losing a top customer: China went from ~40% to mid-20s% of revenue, taking a $600M annual haircut and a $252.5M fine with it. When a company's most powerful counterparty is a government, model regulation the way you model customer concentration.

Self-test: If SK Hynix cuts HBM capex 30% in early 2027, trace the order in which AMAT's numbers move — systems orders → backlog → Semiconductor Systems revenue → gross margin — then explain which line barely moves at all (hint: §4), and what that means for where the P/E should settle.

📖 Glossary

• P/E (price-to-earnings) — price per $1 of annual profit; "trailing" uses the last 12 months, "forward/NTM" the next 12 months' estimate.
• EV/EBITDA — whole-business price (equity + debt − cash) versus operating cash profit; comparable across different debt loads.
• FCF (free cash flow) — operating cash flow minus capex; the cash actually left over for owners each year.
• FCF yield — FCF divided by market cap; the "interest rate" the business pays you at today's price.
• OCF / capex — cash generated by operations / cash spent on plants and equipment.
• Net cash / net debt — cash minus debt; net cash means all borrowings could be repaid tomorrow.
• Gross margin (GPM) — revenue minus direct production cost, as % of revenue; the cleanest pricing-power gauge.
• ROE — profit as a % of shareholders' capital; how hard retained money works.
• CAGR — compound annual growth rate; smoothed per-year growth over a period.
• Buyback — a company repurchasing and retiring its own shares, raising each remaining share's slice of profit.
• Backlog — signed orders not yet delivered or recognized as revenue.
• WFE (wafer fab equipment) — the machines inside a chip factory; AMAT's market.
• Deposition / etch — adding thin layers of material to a silicon wafer / precisely removing them; AMAT's core steps.
• EUV — extreme-ultraviolet lithography, the light-based pattern-printing step; ASML's monopoly, not an AMAT product.
• Foundry — a factory that manufactures chips designed by others (TSMC is the largest).
• ICAPS — AMAT's term for mature-node chips (IoT, comms, auto, power, sensors) — the non-cutting-edge market where China spends heavily.
• HBM / DRAM — high-bandwidth memory (stacked memory feeding AI processors) / standard working memory; the capex swing factor for equipment demand.
• Advanced packaging — connecting multiple chips into one unit (how HBM is stacked); AMAT's fastest-growing area, guided +50% in CY26.
• Equal-Weight / Overweight — analyst-speak for "hold" / "buy" relative to the sector.
• Short position — a bet that profits when a stock falls.
• BIS — US Bureau of Industry and Security, the export-controls enforcer that fined AMAT $252.5M.
• Take rate — the % a platform keeps of each transaction it enables; used here by analogy for AMAT's services revenue on its installed base.