Semiconductors
PrivateA toll-booth on every wafer fab (>85% of WFE process steps) wrapped in $4B of Atotech leverage — the AI-WFE cycle is paying down the debt fast, but at ~39x forward non-GAAP EPS and ~2x the median target, the cycle is already in the price. WATCHING, not buying, until a WFE wobble resets the multiple.
Research
The verdict
A toll-booth on every wafer fab (>85% of WFE process steps) wrapped in $4B of Atotech leverage — the AI-WFE cycle is paying down the debt fast, but at ~39x forward non-GAAP EPS and ~2x the median target, the cycle is already in the price. WATCHING, not buying, until a WFE wobble resets the multiple.
Primary sources
Source documents — open to read in full
MKS is the broadest critical-subsystem supplier to the wafer-fabrication-equipment (WFE) ecosystem — the company's own framing is "Surround the Wafer®," and it claims to address over 85% of the WFE market by content. It does not build the etch/deposition tools themselves (that's Applied Materials, Lam, Tokyo Electron); it sells the guts those tools are built from — vacuum pressure/flow control, gas and reactive-gas delivery, RF/microwave power generation, lasers, optics, precision motion, and (since Atotech) the wet-process chemistry and plating used downstream in advanced packaging. It is, structurally, a picks-and-shovels supplier to the picks-and-shovels makers — a toll on the toolmakers.
Three reportable divisions:
Three end-markets (cross-cut the divisions): Semiconductor $1,696M (43%), Electronics & Packaging $1,111M (28%), Specialty Industrial $1,124M (29%).
Customers: thousands worldwide; top-10 = 35% of revenue (up from 32% in 2024, 30% in 2023); no single customer >10% of revenue in any of 2025/2024/2023. The large customers are concentrated in semiconductors — i.e. AMAT, Lam, TEL and the device makers (TSMC/Samsung/SK Hynix/Intel). Applied Materials publicly flagged a $250M supply-chain hit from MKS's 2023 ransomware outage — direct confirmation that MKS is an embedded, hard-to-substitute AMAT supplier.
Payment terms: mostly point-in-time revenue on shipment, net 30–60 days; plating equipment (MSD) recognized over time (cost-to-cost) with lead times up to 12 months; service/extended-warranty recognized ratably. ~12.6% of revenue is service (recurring-ish); the MSD chemistry consumable stream is the real recurring-revenue thesis but isn't broken out as a clean ARR figure.
Upstream → MKS → end customer, named where the filing names them:
Geographic chokepoint: 81% of revenue is international, with China the single largest country and South Korea/Singapore/Taiwan/Japan close behind. China grew +$155M in 2025 — a tailwind that doubles as the largest geopolitical risk (see Lens 13).
Moat 1 — breadth + "copy-exact" lock-in (the real one). Once an MKS subsystem is designed into a customer's qualified process tool, "copy-exact" rules make it extremely costly to swap — re-qualification can take quarters and risk yield. No competitor "competes across all product lines"; MKS is the only one selling the full Surround-the-Wafer stack. That breadth is the durable edge: a fab buying MKS for pressure control, RF power, and reactive-gas has high switching friction across all three.
Moat 2 — IP + process know-how. 585 US patents + 2,217 foreign patents (expiring through 2044), 150 pending US applications, 15 MSD technology centers. Decades of vacuum/plasma/RF domain expertise that is hard to replicate.
Moat 3 — recurring chemistry consumables (Atotech thesis). MSD's plating chemistries are consumed per-wafer/per-board, giving a razor-blade annuity tied to production volume (not just capex cycles) — the strategic rationale for the $5B+ Atotech deal: smooth the notoriously cyclical equipment revenue with a consumables stream at the Interconnect.
Bargaining power — asymmetric. Over customers: moderate-to-strong on the embedded subsystems (copy-exact), weaker where competitors exist niche-by-niche. Over suppliers: weakened by sole-source inputs and copy-exact constraints that prevent dual-sourcing. Net: the moat is real but narrower than a monopolist's — MKS faces named competition in every product line (Lens 13), just never one competitor across all of them.
Where the moat is thin: PSD's lithography/metrology/inspection products lost share in 2025 (sales fell) — the photonics franchise is the most contested, least defensible leg.
By division:
| Division | FY2025 rev | FY2024 rev | YoY | FY2025 seg. GP% | FY2024 seg. GP% |
|---|---|---|---|---|---|
| Vacuum (VSD) | $1,579M | $1,384M | +14.1% | ~42.9%* | 42.9% |
| Photonics (PSD) | $1,029M | $1,021M | +0.8% | ~44.9%* | 44.9% |
| Materials (MSD) | $1,323M | $1,181M | +12.0% | ~56.1%* | 56.1% |
| Total | $3,931M | $3,586M | +9.6% | 46.7% | 47.6% |
*FY2025 segment GP% read from the segment reconciliation table; division-level GP% moved <1pt vs FY2024 — the structure (MSD ~56%, VSD ~43%, PSD ~45%) is stable.
By end-market:
| End-market | FY2025 | % | FY2024 | YoY | Driver |
|---|---|---|---|---|---|
| Semiconductor | $1,696M | 43% | $1,498M | +13% | logic/foundry strength, NAND upgrades, VSD service |
| Electronics & Packaging | $1,111M | 28% | $922M | +20% | MSD chemistry/equipment + PSD PCB via-drilling |
| Specialty Industrial | $1,124M | 29% | $1,166M | −4% | weaker industrial at VSD/MSD |
Trend read: the two accelerating legs (Semi +13%, E&P +20%) are both AI/advanced-packaging-driven — VSD into logic/foundry capex, MSD/PSD into the package-substrate and HDI build-out for AI accelerators. Specialty Industrial is the decelerating ballast (−4%), dragging on the blend. Geography: international rose to 81% (from 78%), China +$155M. Three-year revenue arc: $3,622M (2023) → $3,586M (2024) → $3,931M (2025) — i.e. MKS only just cleared its 2023 level in 2025; this is a company emerging from a two-year semi-capex trough, not one compounding from strength.
The print:
What drove it: Semiconductor revenue $466M (+13% YoY, +7% QoQ) — plasma/vacuum/RF power for deposition & etch surged on AI capacity; Electronics & Packaging $321M (+6% QoQ); the residual ~$291M Specialty Industrial. Management explicitly credited AI-related demand.
Guidance/outlook: management guided revenue toward ~$1.08B (next Q) and ~$1.14B further out — sequential acceleration, AI tone unmistakably positive at JPMorgan.
Balance-sheet flags:
Market reaction: stock +~10% premarket on the print — the market rewarded the AI-demand beat and EBITDA guide-beat. Read: expectations were already high and MKS cleared them.
Unusual vs own history: the $18M "fees and expenses related to debt activities" + $5M extinguishment loss in Q1 are refinancing one-offs; a $3M legal settlement appeared. None structural.
No transcripts on the research-layer shelf (transcripts/ empty) — this lens is ``-grounded and shallower than ideal; flag for the next refresh to ingest Q1'26 + the prior 3 quarters.
Trajectory (from web summaries):
Recurring phrases: "Surround the Wafer," "Optimize the Interconnect," "foundational technology," "broadest critical subsystem provider." What they stopped saying: the 2023 ransomware/recovery framing has fully dropped out; the narrative has rotated from crisis + integration → deleveraging → AI growth. Tone is at a multi-year high — which is itself a mild contrarian caution.
Peer set: semiconductor subsystem/materials suppliers + the WFE OEMs MKS sells into.
| Company | Ticker | Mkt cap | Fwd P/E | EV/EBITDA | Notes |
|---|---|---|---|---|---|
| MKS Inc. | MKSI | ~$26–27B | ~39x | ~13x | see derivation |
| Advanced Energy | AEIS | n/a | n/a | ~45x (peak Mar-2026) | closest pure subsystem peer (RF power) |
| Entegris | ENTG | n/a | n/a | EBITDA margin ~26.5–27.5% guided | materials peer; multiple not sourced |
| Lam Research | LRCX | n/a | ~50x | n/a | WFE OEM customer |
| Applied Materials | AMAT | n/a | ~47x (trailing ~53x) | n/a | WFE OEM customer; ~$500 stock |
| Coherent | COHR | n/a | n/a | n/a | photonics peer |
| US semi industry avg | — | — | ~33x fwd | — | benchmark |
MKSI multiple derivation: price ~$388–410 × non-GAAP FY2026 EPS consensus ~$10.00 → fwd P/E ≈ 39x (range 35–41x across the price band). On trailing GAAP EPS $4.39 → ~89x (not the right lens given the amortization drag). EV ≈ $26.5B equity + $4.05B debt − $0.57B cash ≈ $30.0B; on FY2026E adj. EBITDA ~$1.15B → EV/EBITDA ≈ 13x.
Read: MKSI's ~39x forward non-GAAP P/E sits below AMAT (~47x) and Lam (~50x) and above the ~33x industry average. The discount to the OEMs is rational — MKS carries $4B of leverage and the Atotech overhang, and its growth is a derivative of the OEMs' growth. The premium to the industry reflects the AI cycle. It is not cheap on any absolute measure; it is cheap only relative to the toolmakers it supplies.
Mostly ; pattern-reading is .
What the market actually reacts to for MKSI: (1) the WFE/semi-capex cycle (the dominant beta — MKS is a high-beta WFE play), (2) balance-sheet/deleveraging milestones (idiosyncratic, the 2023–2026 re-rating story), and (3) AI-capacity headlines. It reacts less to any single customer (no >10% concentration) and more to the aggregate capex tape. This is a cyclical + balance-sheet stock, now trading as an AI-capex stock.
insider-transactions.csv not on shelf — n/a for precise insider %. Comp is RSU-heavy with performance-based Adjusted-EBITDA RSUs (0–200%) and 3-yr relative-TSR RSUs — incentive design is reasonably aligned to operating performance and shareholder return. No 10b5-1 adoptions/terminations by directors/officers in Q4 2025.Forensic lens. Every figure labeled.
Income statement:
Balance sheet:
Cash flow:
Leverage (the headline risk):
Audit/controls: PwC, auditor since 1981; clean opinion + effective ICFR for FY2025; single critical audit matter = revenue recognition (routine for this model); no changes/disagreements with accountants. No restatements.
Regulatory findings (required sub-section):
Built bottom-up from Q1 2026 actuals + guidance. No forecast.ts create (watchlist mode). Every input labeled; outputs ``.
Anchors: Q1'26 revenue $1,078M, adj. EBITDA $277M (25.7%), non-GAAP EPS $2.30. Guidance points to sequential acceleration (~$1.08B → ~$1.14B). ~71M diluted shares (incl. convert dilution). WFE backdrop: +9% (WFE) to +18% (300mm) in 2026, +7–14% in 2027.
| FY2026E | FY2027E | FY2028E | |
|---|---|---|---|
| Revenue | ~$4.45B (+13%) | ~$4.85B (+9%) | ~$5.15B (+6%) |
| revenue logic | Q1 $1.078B run-rate + guided ramp; AI/WFE +13% blended | WFE +7–9% + MSD consumables share gain | cycle matures, +mid-single |
| Adj. EBITDA margin | ~26% | ~27% | ~27.5% |
| Non-GAAP EPS — base | ~$10.00 | ~$11.75 | ~$13.00 |
| EPS logic | $2.30 Q1 × ~4 + interest savings ($27M/yr) | rev growth + ~50bps margin + ~$60M/yr debt paydown lowering interest | operating leverage + continued de-lever |
| consensus check | matches Street ~$10.00 (range $9.02–$10.96) | — | — |
Base call (for tracking, not logged): MKSI FY2026 non-GAAP EPS ≥ $10.00, ~60% probability — in line with consensus, slight upside skew given Q1 momentum, but tariff/China is a real two-sided tail. (Resolves ~2027-02 with the FY2026 10-K.)
Bull case. MKS is the indispensable, broadest content supplier to a WFE cycle entering a multi-year AI/HBM/advanced-packaging up-leg (record $156B equipment sales projected by 2027 ). It addresses >85% of WFE with copy-exact lock-in, so it captures the up-cycle with operating leverage and now carries a higher-margin recurring chemistry annuity (MSD ~56% GM) that dampens the trough. The 2024–2026 deleveraging has de-risked the balance sheet (net leverage ~3x and falling, ~$27M/yr fresh interest savings), so each turn of the cycle now converts to EPS faster than the last — earnings + multiple can compound together. At ~39x forward it trades at a discount to the AMAT/Lam tools it sells into, despite arguably better cyclical-trough protection via consumables.
Bear case (permanent-impairment risks). (1) It's a derivative cyclical at a cycle-high multiple — ~39x forward non-GAAP and ~2x the median analyst target means a lot of the AI up-cycle is already priced; a WFE air-pocket (China digestion, memory wobble) takes both EPS and the multiple down. (2) $4.63B goodwill+intangibles (53% of assets) already impaired once — a demand cut that lowers MSD/PSD forecasts triggers another writedown and a GAAP-loss headline. (3) China concentration meets a hardening export-control / 25%-tariff regime — the single largest country is also the single largest policy risk, and it grew +$155M in 2025 (so the at-risk base is rising).
Pre-mortem (18 months out, thesis broke). WFE 2027 came in flat-to-down as China front-loaded buying into 2025–26 and then digested; memory capex paused; a new BIS rule or retaliatory tariff cut MKS's China shipments 15–20%. Revenue stalled near $4.4B, non-GAAP EPS slipped toward $8.50, the multiple compressed from ~39x to ~22x, and the stock halved from ~$390 to ~$200 (back toward the bear-target band). The Oct-2027 goodwill test forced a partial MSD impairment, putting a GAAP loss in the headline.
Are multiples too high? For a derivative cyclical, yes — uncomfortably. ~39x forward is a secular-growth multiple on a cyclical earnings stream that only just exceeded its 2023 revenue. The bull needs the AI cycle to be durable and MSD consumables to genuinely re-rate MKS toward a less-cyclical multiple. That can happen — but you're paying for it now.
Contrarian view (what the market refuses to see). Two-sided. The bullish contrarian: the consumables annuity is structurally de-cyclicalizing MKS and the market is still pricing it as a pure WFE beta — the re-rate is real and unfinished. The bearish contrarian (which I weight higher at this price): the market is extrapolating a China-front-loaded, AI-front-loaded 2025–26 capex surge as a clean secular trend, and is under-pricing the digestion air-pocket + export-control tail that historically follows every WFE spike — on a balance sheet still carrying $4B of debt.
Skeptical short dismantling the bull.
A licensing-fortified cash machine being paid ~20x forward NOT to lose a $7B Apple leg it is already losing — the rerate only comes if Snapdragon-X PCs + custom data-center silicon replace Apple faster than handsets fade, and the tape (rev −3% YoY, Q3 guide −7%) says it isn't there yet.
A genuinely elite fabless analog compounder (43% 5yr ROIC) that has become a high-beta levered call on Nvidia's power-delivery socket — priced at ~112x trailing / ~49x NTM EV-EBITDA (peer median ~21x) while carrying an unremediated material weakness, an adverse ICFR opinion, and a live securities class action. Own the franchise, not this multiple; the gap between the Vera Rubin "70% share" dream and the Blackwell "allocation-at-risk" reality is the whole trade.