Nova Ltd

Phase A — Understand the business

Company Overview

Nova Ltd. (incorporated 1993, Rehovot, Israel; NASDAQ since April 2000, Tel-Aviv since June 2002) designs and builds metrology and process-control systems for semiconductor manufacturing. In plain terms: when a fab etches/deposits/patterns a wafer, it must measure — at sub-nanometre precision, inline, without destroying the wafer — that each layer is dimensionally, materially and chemically correct. Nova sells the instruments (and software) that do that measuring, plus the service to keep them running. It is a "picks-and-shovels of the picks-and-shovels" play: it does not make chips; it sells the yield-assurance tooling to the people who do.

• Business model / revenue split: FY25 revenue $880.6M, of which products $705.6M (~80%) and services $175.0M (~20%). Products = capital-equipment sales tied to fab capexCapital expenditureMoney spent on long-lived things — buildings, machines, servers — rather than on running costs. cycles; services = extended warranty, time-and-materials, service contracts, install-base upgrades — a recurring annuity that grew for a 13th straight sequential quarter into Q1 2026.
• Customers: all leading manufacturers in logic, foundry, memory and packaging — the world's largest IC makers. Highly concentrated (Lens 4). Sold mainly to chipmakers, occasionally to other process-equipment OEMs.
• Product franchises: Dimensional metrology (the legacy scatterometry/OCD franchise), Materials metrology (incl. XPS / Sentronics) and Chemical metrology — Nova's stated three-pillar strategy. Named platforms surfacing in recent results: Metrion (record DRAM sales), WMC, Semdex (HBMHigh-bandwidth memoryMemory stacked in layers beside a processor so data reaches it faster. The usual bottleneck in AI chips. bookings).
• Contract structure: capital-sale (not take-or-pay, not subscription); cyclical with fab capex. The service layer is the recurring ballast.

Supply Chain

Map: specialised optical/electron/X-ray components & subsystems → Nova assembly (Israel, USA, Germany) → chip manufacturer (the fab) → end devices (AI accelerators, HBM, smartphones, etc.).

• Upstream (inputs): Nova integrates lasers, X-ray sources, electron optics & detection, vacuum systems, spectroscopic and electrochemical modules, plus heavy in-house software/algorithms (physical modelling + ML). It is a systems integrator of precision photonics/electron-optics rather than a fab itself — so its upstream chokepoints are specialised component vendors (named suppliers are not disclosed in the 20-F; n/a — not disclosed, a genuine gap).
• Manufacturing nodes: Israel (primary), USA, and Germany via Sentronics (acquired Jan 30 2025). Concentration of inventory/WIP in these "Manufacturing Facilities" is itself flagged as a risk in the filing.
• Downstream (the buyers — names): the customer base is the fab oligopoly — TSMC, Samsung, SK Hynix, Micron, Intel, and the leading Chinese foundries/memory makers (SMIC, CXMT, YMTC-type buyers). Geographically, FY25 sales went China 33% / Taiwan 29% / Korea 16% / USA 9% / Other 13% — i.e. the downstream is literally TSMC (Taiwan), the Korean memory duo (Korea), and the Chinese fab build-out (China).
• Chokepoints / single-source dependencies: the binding constraint is customer concentration, not supplier concentration — Customer A is 23% of revenue (Lens 4). The second chokepoint is geopolitical: a third of revenue routes through China under a tightening US/Dutch export-control regime (Lens 10/13).

Competitive Advantages (moats)

Nova is the clear #2 in process control behind KLA — it holds roughly 25% of thin-film and critical-dimension metrology, second only to KLA, in a market KLA dominates with ~54-55% overall process-control share. The moat is real but narrower than KLA's:

• Switching costs / recipe lock-in (the core moat): metrology tools are qualified into a customer's process recipe at a specific node; once Nova's tool is the reference measurement for a fab's 2nm or advanced-DRAM line, ripping it out mid-node risks yield. This is why the install base compounds and services grow every quarter. Durable.
• Technical breadth (the widening moat): Nova deliberately diversified from Dimensional-only into Materials + Chemical metrology and bolted ML onto physical modelling — expanding its served market and making it a multi-technique supplier rather than a one-trick OCD vendor. The Sentronics (Materials/XPS) deal extends this.
• Bargaining power — asymmetric. Over customers: moderate — the fabs are giant and concentrated (they need a metrology answer, but KLA is the default, so Nova competes hard on differentiation/price). Over suppliers: high — it's a systems integrator buying components. Net: Nova needs the big fabs more than they need Nova, which caps pricing power vs. a true monopolist.
• R&D intensity as a moat-maintenance cost: FY25 R&D (net) $143.4M = 16.3% of revenue; ~682 R&D staff incl. ~163 PhDs. This is the price of staying #2 in a market where process-control intensity keeps rising — not optional.

Moat verdict: a strong specialist moat (recipe lock-in + multi-technique breadth) sitting underneath KLA's category-defining one. Nova is the best house on a street KLA owns.

Segments

Nova reports as one operating segment (CODM = the CEO, who manages on consolidated operating margin and net income) — so there is no GAAP product-line P&L. Disaggregation available:

Geographic (% of sales): China 33% (was 39% '24 / 36% '23 — declining share, export-control + local-competition headwind), Taiwan 29% (was 20% / 18% — rising fast, the TSMC 2nm ramp), Korea 16%, USA 9%, Other 13%.

Demand mix (Q1 2026, web): Memory = 34% of product revenue and the standout driver, with advanced DRAM ~two-thirds of the memory business and record Metrion sales; HBM bookings on WMC/Semdex flagged robust. Read: the growth engine has rotated from foundry/logic toward AI-memory (HBM/advanced DRAM), which is the single most important fact in the forward story.

¹ FY23 product GM stated as "60%" in the FY25 20-F MD&A narrative; the consolidated gross profit % was 57% in FY23/FY25 and 58% in FY24.

---

Phase B — Measure performance

Earnings Result

Two prints matter: the FY2025 annual (on-disk) and the Q1 2026 (web).

FY2025 (20-F):

• Revenue $880.6M, +31% YoY. Gross profit $505.2M (57% GM). Operating income $253.5M (28.8% op margin). Net income $259.2M (29.4% net margin) — net income > operating income because of $49.8M financial income (interest on a ~$1.65B cash/securities pile).
• EPS: basic $8.61, diluted $7.96 (FY24 diluted $5.75; FY23 $4.28) — diluted EPS +38% YoY.
• Opex discipline: R&D $143.4M, S&M $82.2M, G&A $26.1M; total opex grew 26% vs revenue +31% → operating leverage.
• Cash flow: operating cash flow $245.6M; capex $27.7M ⇒ FCF ≈ $217.9M (~25% FCFFree cash flowCash left after paying to run and maintain the business. Unlike profit, it is hard to flatter with accounting choices. margin). Tax rate ~14.5% (Israeli preferred-enterprise regime).

Q1 2026 (reported 2026-05-14):

• Revenue $235.3M (+10% YoY, +6% QoQ) — beat the $222-232M guide. GAAP net income $69.3M, diluted EPS $2.04; non-GAAP EPS $2.33 (also above guide). GM 57.7%; non-GAAP operating margin 34%.
• Q2 2026 guide: revenue $245-255M, GAAP EPS $2.10-2.24, non-GAAP EPS $2.34-2.48.
• Balance-sheet flags (FY25 20-F): inventory $183.7M (+17% YoY, outpacing the +10% Q1 revenue rate — watch), receivables $151.9M (+9%, in line). Net cash ~$914M after the convert (cash+deposits+securities $1,646M − $731.7M notes).
• Market reaction: stock fell ~8% on the Q1 print despite the beat — a classic "priced for more" reaction at a 35-48x multiple. The stock then ran to a $615.99 ATH by mid-June before pulling back to ~$554.

Unusual vs. own history: the financing line. In Sep 2025 Nova issued $750M of 0% convertible senior notes due 2030 and bought ~$53M of capped calls; the older 2025 converts (~$181M) fully converted, adding 2.43M shares. So the share count and the balance sheet both took a structural step.

Earnings Calls (sentiment trend)

No transcripts on disk (transcripts=0). From web call coverage:

• What management keeps saying: "record," "memory," "advanced DRAM," "HBM," "share gains," and the $1B-revenue-in-2027 target (now described as "on track," with 2027 orders already underway and a new Asia production facility online by end-2026). Tone over the last ~3-4 calls has shifted from cyclical-recovery framing (2024) to confident secular-growth framing (2025-26) as memory inflected.
• What they stopped emphasising: macro/cyclical caution and China softness — China share is quietly declining (39%→33%) but the narrative now leads with Taiwan/memory, not China risk.
• Sentiment read: clearly bullish/expansionary management voice; the risk is that the tone has caught up with (or exceeded) the valuation, leaving little tolerance for an in-line quarter.

Comps

Peer set = the process-control / metrology-inspection group. Multiples are, dated; where unsourced, n/a. Do not treat as precise — sources conflict and dates vary.

Read: Nova trades in line with the smaller specialists (Onto/Camtek at ~35x) and at a discount to KLA's ~70x trailing — but KLA's multiple is partly a quality/monopoly premium and partly trailing-vs-forward optics. On forward earnings the group clusters ~35x; on the higher ~48x quote Nova would be the expensive one. Nova's distinguishing features in the comp set: highest revenue growth rate (+31%), zero dividend, strong 24% ROE, and a fortress net-cash balance sheet that none of the smaller peers match. KLAC's ~38.7x EV/EBITDA being "151% above its 10-year median" tells you the entire group is rich, not just Nova.

Stock-Price Catalysts (what moves NVMI >5%)

From the pattern over the cycle:

• Earnings prints & guidance are the dominant mover — and the reaction is asymmetric to expectations, not absolutes: Q1 2026 fell ~8% on a beat because the bar was higher.
• Memory/HBM datapoints — the rotation to memory (34% of product rev) makes NVMI increasingly a DRAM/HBM-capex proxy; HBM bookings commentary moves it.
• The $1B-2027 target & the new Asia fab — milestone catalysts management has explicitly dangled.
• China export-control headlines — a third of revenue; new BIS/Dutch rules are a recurring down-catalyst.
• The AI-capex macro — as a second-derivative AI-infrastructure name, NVMI swings with the NVDA/TSMC/memory complex and with any "AI-capex-peaking" scare.
• Recent tape: ATH $615.99 (2026-06-15) → ~$554 (2026-06-17), i.e. a ~10% pullback from the high inside a week — momentum-driven, on no disclosed company-specific news.

---

Phase C — Judge people & books

Management

• CEO: Gabriel ("Gaby") Waisman — appointed CEO & President Feb 2023; previously Nova's Chief Business Officer since 2016, 20+ years in global public-company roles across Israel/US/Asia. An internal-promotion, commercially-grown CEO, not a founder.
• Executive Chairman: Eitan Oppenhaim — CEO 2013-2023 (joined 2010); ran the transformation from a ~$200M sub-scale OCD vendor into the diversified $880M multi-technique #2. Now Exec Chairman — so the architect of the growth era is still at the top table. Strong, quantified track record: revenue roughly 4x'd over his tenure with margin expansion.
• Skin in the game / insider ownership: insider holdings not on disk (our figures absent) — n/a precisely; Nova is a widely-held, no-controlling-shareholder Israeli tech (no founder block disclosed in the 20-F cover).
• Capital-allocation history: disciplined and shareholder-aware: modest, steady buybacks ($25M program '18; $30M '24; $35M '25 at cost) — used to offset SBC dilution, not lever the balance sheet. No dividend (correct for a 31%-grower). The big '25 move — $750M 0% convertible + capped calls — is a sophisticated, low-cost financing (zero coupon, dilutionDilutionIssuing new shares, so each existing share owns a smaller slice of the same company. capped) to fund growth/M&A and the Asia fab; not value-destructive, but it did add the 2.43M-share convert overhang. M&A: the Sentronics tuck-in (~$56M) extends Materials metrology — small, strategic, on-thesis.
• Capital-allocation scorecard: ROE 24.2%, ROIC well above cost of capital given ~$253M operating income on a largely cash-funded balance sheet. Reinvest-first, buy-back-to-offset-dilution, opportunistic tuck-in M&A. Grade: A-.
• Red flags: none material. Comp not flagged as excessive; no related-party deals in the filing; the only "promotional" risk is the management tone outrunning results at a rich multiple.
• Archetype: professional operators, founder-grade continuity (the long-tenure Chairman + internally-grown CEO) — the right profile for a scaling specialist that needs execution, not founder-visionary risk-taking.

Forensic Red Flags

Forensic read of the FY2025 20-F. This is a clean set of books. Specifics:

• Revenue recognition: standard ASC 606 capital-equipment + service; deferred revenue $67.2M (down from $72.9M) — no channel-stuffing tell; the deferred-revenue decline alongside revenue growth is mild and explained by timing, not a red flag.
• Cash vs. earnings: operating cash flow $245.6M vs. net income $259.2M — a ~0.95x cash-conversion. Slightly below 1.0x, driven by inventory build (+$19.0M) and other working-capital uses, not by aggressive accruals. Healthy.
• Receivables/inventory vs. revenue: receivables +9% vs revenue +31% (conservative — collecting faster than growing). Inventory +17% (FY25) and the absolute level keeps rising — the one line to watch; in a capital-equipment up-cycle this is normal pre-positioning, but if memory demand stalls it becomes the first impairment risk.
• SBC / non-GAAP: SBC $26.3M = 3.0% of revenue — modest by semi-equipment standards. The non-GAAP-to-GAAP EPS gap (Q1'26 $2.33 vs $2.04) is ~14%, driven mostly by SBC and intangible amortisation — reasonable, not abusive.
• Goodwill/intangibles: goodwill $90.8M (+$42.5M from Sentronics), intangibles $45.8M — small relative to $2.36B assets; low impairment risk; one reporting unit, tested annually.
• Convertible accounting: the $750M 0% convert sits at $731.7M net (non-current); EPS dilution handled via if-converted; capped calls economically offset upside dilution. Transparent.
• Internal controls: SOX 404(b) attestation clean; Sentronics (acquired Jan '25) excluded from the first-year ICFR assessment (standard, ~2% of assets / ~6% of revenue).

Regulatory findings (required sub-section):

• SEC Litigation Releases / AAERs: None. EDGAR EFTS search (LR + AAER, 2021-06-20 → 2026-06-20) returned 0 findings.
• Non-SEC enforcement (FTC/DOJ/etc.): None found. Targeted web search ("Nova Ltd" + FTC/DOJ/SEC/settlement/fine/penalty/lawsuit/class-action, 2024-2026) surfaced only unrelated "Nova" entities (Fashion Nova FTC; the NovaTech crypto fraud) — no hits on NVMI the semiconductor company.
• Item 8 / Legal Proceedings (the company's own disclosure): the 20-F carries only ordinary-course commitments & contingencies (Note 13); no material litigation disclosed.
• Conclusion: No material regulatory or legal findings — verified via SEC EDGAR EFTS (LR, AAER), targeted web search, and the 20-F's own legal/contingency disclosure as of 2026-06-20. The regulatory risk that matters here is prospective export-control, not enforcement (see Lens 13).

---

Phase D — Project & stress-test

Forward Projection

Build bottom-up from FY2025 actuals + Q1 2026 + the Q2 guide + management's $1B-2027 target. All outputs; inputs labelled. Fiscal years = FY2026 / FY2027 / FY2028 (Dec year-ends).

Anchors: FY25 revenue $880.6M, GAAP diluted EPS $7.96. Q1'26 $235.3M; Q2'26 guide midpoint $250M. H1'26 run-rate ⇒ ~$485M ⇒ a ~$1.0B FY26 is plausible if H2 holds the memory ramp. Management guides $1B for 2027, implying they may be conservative on 2026.

What the price implies: at ~$554 and base FY26 GAAP EPS ~$10, the forward P/E ≈ 55x on GAAP / ~48x on non-GAAP — demanding. To "grow into" ~30x by FY28 the company needs the bull EPS (~$13.5), i.e. the market is asking for the super-cycle, not the base. One source pegged the implied ask at a ~35% 5-year earnings CAGR — directionally consistent with "priced for the bull case."

No our model create is run — per --watchlist rules, the Brier forecast is only logged on genuine conviction commitment, which is a human-gated our position log step. Candidate forecast to log there: "NVMI FY2027 GAAP diluted EPS ≥ $10.50, p≈0.45, resolves 2027-12-31."

Bull vs Bear

Bull case. Nova is the #2 in a market whose intensity is structurally rising — process control went from 5.9% → 8.6% of WFE in seven years, and the hard part of the AI buildout (HBM stacking, advanced DRAM, 2nm GAA + backside power) is exactly where metrology demand compounds fastest (sub-3nm = 9.25% CAGR, the fastest sub-segment). Nova has out-grown the process-control market (27.5% product CAGR vs ~16% sector, 2020-25 ) by widening from Dimensional into Materials + Chemical metrology, and it now rides the single best capex theme in tech (AI memory) with record memory bookings. Fortress balance sheet ($914M net cash), 24% ROE, 25% FCF margins, disciplined capital allocation, and a credible $1B-2027 target with orders already in hand. If HBM/advanced-DRAM capex stays elevated through 2027, the bull EPS (~$13.5 by FY28) is reachable and the multiple is justified.

Bear case (permanent-impairment risks).

1. Customer concentration → demand cliff. Customer A = 23% of revenue (up from 18%), top-5 = 51%. A single large customer cutting/delaying capex (a TSMC or memory-maker air-pocket) takes a fifth of revenue with it — and metrology is capex, so it's volatile, not annuity.
2. China (33% of sales) is a structural one-way risk. Export controls can shut the high-end overnight; local Chinese metrology champions (favoured by policy) erode the rest. China share is already declining 39%→33% — the bear says this continues to single digits.
3. Valuation is the risk. At ~35-48x forward the stock has priced in the super-cycle. An in-line quarter (it beat Q1 and still fell 8%) or a memory-capex pause de-rates a 45x multiple violently — a 35% growth-CAGR ask leaves no margin for error.

Pre-mortem (18 months out, thesis broke): It's late 2027. HBM capex over-shot in 2025-26 and the memory makers are digesting — Nova's memory line (the whole growth story) goes from +record to flat. Simultaneously a BIS rule tightens China advanced-node tooling and Chinese locals take the mature-node metrology share. Revenue stalls near ~$950M, the $1B-2027 target slips, and a stock that was 48x forward re-rates to 25x on now-flat earnings — a ~50% drawdown that has nothing to do with the business being "bad" and everything to do with the price.

Contrarian view (what the market refuses to see): the bull tape treats NVMI as a secular compounder, but a third of its revenue is the most politically fragile demand in the industry (China) and a fifth is one customer. The market is paying a monopoly-quality multiple for a #2 with concentration risk. The contrarian flip side: if you strip out the China third and value only the "allied-fab + memory" two-thirds, the durable business is smaller and cheaper than the headline cap implies — but also cleaner, and that's the version that survives a China-decoupling.

Devil's Advocate (short-seller)

Dismantling the bull case.

• Where the money structurally breaks: Nova sells capital equipment into a cyclical capex budget controlled by ~5 customers. Strip the cycle and you have a great business; mis-time the cycle and a 31%-grower is a -20% decliner — and the multiple assumes the up-cycle is permanent.
• Revenue concentration / the shift: 23% in one customer, 33% in one country. If Customer A in-sources or dual-sources metrology, or if China is export-control-fenced, the growth story (not just a quarter) is impaired. The China share has fallen three straight years — that's not noise, it's a trend the bulls are narrating around.
• Why the moat is weaker than bulls think: Nova is #2, not #1. KLA (~54%) sets the reference and out-spends Nova ~3-4x on R&D in absolute dollars. In any segment where KLA chooses to push, Nova defends with price. The "multi-technique breadth" is real but is also KLA's strategy, executed with a far bigger chequebook.
• Most dangerous competitor bulls underestimate: not KLA — Onto Innovation in advanced packaging, the fastest-growing adjacency (Onto guiding >30-50% advanced-packaging growth, a >$240M HBM purchase agreement). If packaging metrology becomes the prize and Onto owns it, Nova's memory-led growth narrative has a hole. Plus policy-backed Chinese local metrology for the mature-node third.
• Worst capital-allocation reads: none egregious — but the $750M convert + 2.43M-share conversion quietly diluted holders, and a 0% convert is cheap only if the stock cooperates; at a 48x multiple that's a bet.
• Assumptions that must hold for ~$554: (a) memory/HBM capex stays at super-cycle levels through 2027; (b) China doesn't fall below ~25% of sales; (c) op-margin holds ~30%+; (d) no large-customer air-pocket. Break any one and the multiple, not the business, is the loss.
• If growth disappoints 20-30%: FY28 EPS drops from base ~$10.8 toward bear ~$7.5; a sympathetic de-rate to ~25-28x ⇒ ~$190-210 fair value, i.e. ~60% downside from ~$554. That is the asymmetry the short presses.
• Single permanent-impairment scenario & plausibility: full US/allied export-control fence on China advanced-node + Chinese-local displacement of the mature-node third ⇒ permanent loss of ~25-30% of revenue with no near-term replacement. Plausibility: moderate (20-35% over 3 years) — high enough to demand a margin of safety the current price doesn't offer.

Management Questions (ordered by information value)

1. Of FY25 revenue, how much is HBM/advanced-DRAM-specific, and what is your assumed memory-capex trajectory underpinning the $1B-2027 target — what happens to that number if memory capex is flat in 2027?
2. Customer A is now 23% of revenue (up from 18%). Is this concentration rising further, and what is your single-customer exposure ceiling before it changes how you run the business?
3. China fell from 39%→33% of sales in two years. Decompose that: how much is export-control-driven vs. local-competitor share loss, and where do you see China settling?
4. Which Chinese domestic metrology competitors do you now lose deals to at mature nodes, and on what dimension (price, policy preference, "good-enough" spec)?
5. In advanced-packaging metrology — the fastest-growing adjacency where Onto is aggressive — what is Nova's share today and your right-to-win versus Onto and KLA?
6. KLA outspends you several-fold on R&D in absolute dollars. In which specific segments are you gaining share on KLA, and where are you structurally ceding?
7. The $750M 0% convert + capped calls: walk through the dilution math at various share prices, and what the proceeds actually fund (Asia fab? M&A? buffer?).
8. Inventory rose +17% in FY25 while Q1'26 revenue grew +10%. How much is HBM/2nm pre-positioning vs. demand you may not realise, and what's the write-down risk if memory pauses?
9. What is the service-revenue ceiling as a % of mix, and its steady-state gross margin — how much of the model can become recurring annuity?
10. The new Asia production facility (online end-2026): capacity added, capex, gross-margin impact, and which customers/geos it de-risks?
11. Sentronics (Materials/XPS): integration status, revenue contribution trajectory, and is more Materials/Chemical M&A coming?
12. At ~30%+ operating margin, where is the incremental operating leverage — can margins expand from here, or is R&D intensity a structural cap?
13. Capital-return policy: at ~$914M net cash and 25% FCF margins, why no dividend, and at what cash level does buyback/dividend scale up?
14. What is your win-rate at the 2nm/18A node versus prior nodes — are you holding, gaining, or losing reference-tool position at the leading edge?
15. If forced to name the one variable that most determines whether Nova is a $2B or a $1B revenue company by 2030, what is it?

---