Register D | Generated 2026-04-26 | Intel supply-chain swarm > Industry primer: Semiconductor Equipment / X-Ray Analytical Instruments (researched 2026-03-09 in industry-log)
Register D | Generated 2026-04-26 | Intel supply-chain swarm Industry primer: Semiconductor Equipment / X-Ray Analytical Instruments (researched 2026-03-09 in industry-log)
Thesis (bull case): Onto Innovation is the highest-quality pure-play on advanced packaging process control, with a defensible niche in bump/packaging inspection (Dragonfly) and OCD metrology (Atlas) at the exact inflection point where AI infrastructure is forcing the industry’s most complex packaging ramps in history. The SK Hynix $240M volume agreement through 2027 anchors near-term revenue, the Semilab acquisition adds ~$100-120M of accretive FY2026 incremental revenue, and management’s guidance of >30% advanced packaging growth in FY2026 suggests the business is re-accelerating after a soft Q3 2025. This is a capital-efficient, no-debt, $300M+ FCF generator priced at 46x forward earnings – demanding, but not unreasonable for a semicap business with genuine defensible share in the fastest-growing segment of the supply chain.
Current price (Apr 24, 2026): $307.86 Market cap: $15.31B Enterprise value: $14.69B (net cash ~$622M) Target price: Not set (this is a new research initiation). Analyst consensus average $290-317; Stifel high target $350. Conviction: Medium-High — the thesis is clear and well-anchored, but the valuation is demanding and the Q3 2025 step-down in margins raises questions about Semilab integration execution.
Full legal name: Onto Innovation Inc. Ticker: ONTO / NYSE GICS: Information Technology / Semiconductor Equipment HQ: Wilmington, Massachusetts Founded / IPO: Formed October 25, 2019 (merger of Rudolph Technologies + Nanometrics) Website: ontoinnovation.com IR: investors.ontoinnovation.com
What it does: Onto makes the measurement instruments that semiconductor fabs use to control their production processes. Every wafer that runs through a leading-edge fab needs to be measured dozens of times: is the film thickness correct? Are the gate structures the right dimensions? Is there a particle or defect that will kill yield? Onto’s tools answer these questions. Without them, fabs cannot achieve the yields that make the economics of a chip fab work.
Business lines: - Advanced Packaging: Macro-inspection and 3D bump metrology (Dragonfly family). ~35-40% of revenue. The AI packaging story lives here. - Advanced Nodes: Optical critical dimension (OCD) metrology for leading-edge logic and memory (Atlas family). ~35-40% of revenue. The GAA/2nm transition story lives here. - Specialty Devices + Service: Compound semi (SiC/GaN), older nodes, and aftermarket services/spares. ~20-30% of revenue. Semilab product lines (FAaST/CnCV/MBIR) slot in here.
Geographic revenue (FY2024): Taiwan 31%, South Korea 29%, China 12% (deliberately reduced to <3% in FY2025), US 11%, SE Asia 7%, Japan 6%, Europe 5%.
Business model: Capital equipment (one-time sale per system, ~$1-3M+ ASP per tool), supplemented by aftermarket services and spares (~20-25% recurring). No SaaS. Revenue is order-driven and lumpy; grows with fab capex cycles. Non-GAAP gross margin ~54-55%.
Investor presentation: 2024 Annual Report (Aug 2025). IR page: investors.ontoinnovation.com.
Asset-light relative to semiconductor equipment peers. Capex runs $20-32M/year (~3% of revenue), well below manufacturing-intensive semicap companies. Outsources significant mechanical assembly.
Key sites: - Wilmington, MA: Corporate HQ, core engineering, R&D - Milpitas, CA: Thin-film/OCD R&D (legacy Nanometrics) - Boise, ID: Manufacturing and engineering (legacy August Technology) - Lehi, UT: Engineering and manufacturing - Semilab USA (acquired Nov 2025): FAaST/CnCV/MBIR product lines - Taiwan, South Korea, Japan, Singapore, Malaysia, Vietnam: Field service and applications engineering
Total headcount: ~1,593 (pre-Semilab integration).
JVs: None disclosed.
Strategic partnerships: - Rigaku Holdings (268A, TSE): Co-development on hybrid x-ray/optical metrology for 3D DRAM applications, flagged via channel checks in April 2026. Not yet formally disclosed by Onto. Watch Q1 2026 earnings. - Customer co-development programs at TSMC, Samsung, Intel for process qualification — not disclosed as formal JVs but embedded in the moat.
Semiconductor manufacturing is a series of hundreds of process steps. At each step, something is deposited, etched, planarized, or implanted. At each step, the result must be measured to verify it was done correctly. If a film is 2nm thinner than spec, yield falls. If a feature is 0.5nm too wide, the transistor doesn’t switch correctly. If a particle lands on the wafer and isn’t caught, dozens of chips die.
Before process control metrology existed, fabs caught problems at the end of the line — too late, after enormous cost had been sunk. Modern inline metrology catches problems immediately after each process step, enabling real-time closed-loop control. Without it, advanced node manufacturing is economically impossible.
At advanced packaging, the same principle applies but in three dimensions: chips being stacked face-to-face (hybrid bonding) or interconnected via micro-bumps (CoWoS, EMIB) must be aligned to sub-micron precision across entire wafer surfaces. Any misalignment kills the die stack. Process control at the packaging layer is what makes these architectures manufacturable at yield.
Optical Critical Dimension (OCD) metrology, also called optical scatterometry, works as follows:
Step 1 — Illuminate: A beam of broadband polarized light is directed at the wafer surface at a controlled angle. The surface is a periodic grating structure (the transistor fins, gate lines, or contact patterns of the device being built).
Step 2 — Detect: The reflected light is captured by a spectrometer. Because the structures are periodic (like a diffraction grating), the light diffracts in ways that encode the geometry of the structures. The spectrometer captures this diffracted light across a range of wavelengths.
Step 3 — Model: The spectral response is compared against a library of theoretical responses computed by electromagnetic simulation (using Rigorous Coupled Wave Approximation, or RCWA). By finding the best match between measured spectrum and computed spectrum, the software extracts the actual dimensions of the structure: height, width, sidewall angle, top/bottom rounding.
Step 4 — Control: The extracted dimensions are fed back to the process tool (the etch, deposition, or CMP tool that created the structure). If a dimension is drifting out of spec, the process tool adjusts.
The key insight: OCD can measure sub-nanometer features with sub-nanometer precision, non-destructively, in under a second per measurement, on production wafers. It cannot image the feature (that requires SEM), but for most process control purposes, the indirect measurement is all you need, and it is far faster and cheaper than electron microscopy.
Onto’s competitive advantage in OCD is in the electromagnetic modeling software: specifically, the Ai Diffract and SpectraProbe suites that extend spectral ellipsometry to measure 3D nanosheet (GAA) structures. KLA’s spectral ellipsometry tools reportedly cannot handle the 3D modeling complexity required for gate-all-around transistors, giving Onto a specific window of advantage at 3nm/2nm nodes.
Dragonfly is a macro-defect inspection system — it uses optical imaging (not electron beams) to scan large areas of a wafer or panel quickly and flag anomalies.
Step 1 — Illuminate: Multiple illumination channels (bright field, dark field, varied angles) are applied to the wafer or package surface.
Step 2 — Capture: High-speed cameras capture reflected/scattered light. At the G5 generation, the system can detect defects down to 150nm at throughputs up to 5x the prior generation.
Step 3 — Classify: AI-assisted classification software distinguishes real defects (particles, scratches, cracks) from false positives. The 3D capability adds height measurement, critical for detecting Cu bump height variations or voiding in die stacks.
Step 4 — Report: Defect maps are generated and fed into yield management systems.
In advanced packaging, Dragonfly’s relevance is bump inspection: verifying that every micro-bump on a 2.5D/3D package is the correct height, has no bridging, and has no voiding. With hundreds of thousands of bumps on a single Nvidia H100 substrate, even a fraction of a percent defect rate is catastrophic for yield. The Dragonfly G3’s $240M agreement with SK Hynix is specifically for this inspection of HBM die stacks.
| Metric | Why It Matters | ONTO’s Position |
|---|---|---|
| Defect sensitivity (min detectable feature) | Determines which process steps the tool can monitor | G5: 150nm; G3: submicron |
| Throughput (wafers/hour) | Determines cost of ownership for the fab | G5 is 5x prior gen |
| Precision / repeatability (nm) | Determines if measurements are trustworthy for process control | Sub-nm for OCD |
| Recipe portability | Can recipes be transferred across tool fleets | Yes — fleet management software |
| Total cost of ownership (TCO) | What the fab pays per measurement | Key buying criterion vs KLA |
| Time-to-production (qualification) | How fast a new recipe is ready | Competitive differentiator |
What it does: Measures the dimensions and properties of structures patterned onto the wafer during front-end fabrication: gate heights, feature widths, film thicknesses, sidewall profiles.
How it works: Spectroscopic ellipsometry and OCD (as described in Section 3).
Generation status: - Atlas III+: Deployed at leading fabs for FinFET processes - Atlas V: Current HVM tool; target market FinFET, GAA logic, DRAM, 3D NAND - Atlas G6: Launched April 2025. Next-gen for 2nm/AI device nodes.
ASP: Not publicly disclosed. Semicap OCD tools typically range $2-5M per unit.
Competitive position: Onto has been selected at TSMC, Samsung, and Intel for GAA-specific OCD. SemiAnalysis noted this as a genuine share gain vs. KLA at the GAA transition. The Ai Diffract 3D modeling software is the differentiator.
Customers: TSMC, Samsung, SK Hynix, Intel, Micron.
Attach rate / service tail: Recipe libraries, modeling software licenses, annual service contracts (~15-20% of system ASP per year).
What it does: Detects macroscopic defects and performs 3D metrology on wafers and packages. Critical for bump inspection in advanced packaging.
Generation status: - Dragonfly G3: Current volume tool; $240M+ volume agreement with SK Hynix for HBM3/HBM4 inspection. Combines 2D defect inspection with 3D bump height measurement. - Dragonfly G5: Launched Q1 2026. Sub-150nm defect sensitivity, 5x throughput improvement.
ASP: Not disclosed. The $240M agreement covers multiple units through 2027; implied per-unit pricing in the $1-3M range for the G3. G5 will likely command a premium.
Application specificity: Unlike Atlas (process control for patterning), Dragonfly is most important at the advanced packaging back-end. Every HBM die stack needs bump inspection before bonding. Every CoWoS interposer needs surface quality verification. Every panel-level package needs macro defect detection.
Competitive threat: Camtek (CAMT) competes directly in APAC at potentially lower price points. KLA does not have a strong Dragonfly analog.
FAaST: Inline wafer contamination monitoring. Uses rapid fluorescence and spectroscopic techniques to detect metal contamination at the ppb level. Critical for logic and power device fabs where metallic contamination kills yield.
CnCV: Materials characterization. Composition and thickness of thin films, dopant profiling. Extends Onto’s reach into electrical characterization.
MBIR: Surface charge metrology. Measures interface trap density and oxide charge at semiconductor surfaces. Important for SiC and GaN power devices.
Strategic fit: These three product families allow Onto to offer a “measure everything” capability: dimensional control (Atlas), defect detection (Dragonfly), and now materials/electrical properties (Semilab). No single competitor can match this complete process control suite.
FY2026 contribution: ~$100-120M (H2 weighted per management guidance).
[Si wafers] → [Chemicals/gases] → [Semicap equipment (litho, etch, CVD, CMP)] → [Process control ★] → [Chip fabs] → [Packaged chips] → [End systems]Onto operates at the process control layer (★). It is not the “main” fab equipment (that’s ASML, Lam, Applied, TEL) — it is the measurement and control layer that wraps every process step. This is a less glamorous but critically important niche: you cannot run a fab without process control.
Revenue pool: The process control layer (metrology + inspection) is estimated at $7-12B TAM depending on scope. KLA captures the largest share (~$8B revenue in FY2025). Onto is at ~$1B, suggesting ~8-12% of the TAM.
Value capture: Process control tools have strong pricing power because: 1. The customer cannot afford a defect that slips through 2. The tool is typically $1-5M vs. the $50-200M+ litho/etch tools it supports — but it protects billions of dollars of wafer starts 3. Switching costs are high (requalification of recipes = 6-18 months)
Suppliers: Onto sources optics, lasers, electronics, and precision mechanics from diversified suppliers. No confirmed single-source critical supplier. Laser and optical component suppliers include standard semicap vendors (II-VI/Coherent, IPG, etc.).
| Supplier | Ticker | Layer | Bypass-ability | Supplier MC vs ONTO | Market-pricing |
|---|---|---|---|---|---|
| Coherent (optical components) | COHR (NYSE) | Optics/lasers | Partial (multiple optics vendors) | ~$14B vs $15B (roughly equal) | Priced-in |
| II-VI / Coherent (after merger) | COHR | Photonics | Partial | Same as above | Priced-in |
| Precision mechanics suppliers | Private | Mechanical subsystems | Yes (multiple vendors) | Not applicable | N/A |
| Rigaku (x-ray sources) | 268A (TSE) | X-ray detection | Partial (co-development signal) | ~$3.7B vs $15B (smaller) | Potentially under-priced |
Bottleneck verdict: No clear single-source upstream bottleneck that represents an investable secondary position. Rigaku’s co-development with Onto on hybrid x-ray/optical metrology is worth monitoring — if this becomes a revenue-significant product, Rigaku could be a secondary beneficiary. See [[268A]] in vault.
| # | Customer | Ticker | Est. Revenue Share | Relationship Type | Contract / Qualification Details |
|---|---|---|---|---|---|
| 1 | TSMC | TSM (NYSE) | ~25-30% est. | OEM end-user (foundry) | Multi-product qualification: Atlas for GAA/OCD, Dragonfly for advanced packaging. TSMC co-qualification of Atlas for 3nm/2nm processes. |
| 2 | SK Hynix | 000660 (KRX) | ~15-20% est. | Volume agreement through 2027 | $240M+ Dragonfly G3 volume purchase agreement; $60M specifically for 3D bump metrology for HBM4 ramp |
| 3 | Samsung Electronics | 005930 (KRX) | ~10-15% est. | Multi-product (foundry + memory) | Atlas for GAA; Dragonfly for packaging; hybrid bonding evaluations ongoing |
| 4 | Intel | INTC (NASDAQ) | ~5-10% est. | Strategic OEM | EMIB bump inspection; Foveros hybrid bonding metrology; Intel’s advanced packaging capex key to this relationship |
| 5 | Micron Technology | MU (NASDAQ) | ~5% est. | Memory | DRAM and NAND process control; HBM exposure |
Concentration risk: Taiwan ~31% of FY2024 revenue (predominantly TSMC). Korea ~29% (Samsung + SK Hynix). Combined top-2 geography = ~60%. Customer names not disclosed in SEC filings. The SK Hynix volume agreement is the most concrete public anchor.
What happens if the largest customer walks: If TSMC (~25-30%) reduced tool orders by 50%, it would cost $125-150M in revenue (~12-15% of total). That would hurt, but with Semilab adding $100-120M FY2026 and advanced packaging growing >30%, other segments would largely offset over 12-18 months. The business would not face an existential threat.
Strategic partnerships: Beyond volume agreements, Onto participates in customer-funded co-development programs where the fab pays for recipe development and early access. These are not disclosed but are standard in the semicap industry and create deep switching costs.
Why it matters: AI chips are the most complex manufactured objects humans have ever made. A single Nvidia Blackwell GPU involves multiple chips bonded together in a 2.5D package with HBM memory dies stacked on top, all connected through thousands of micro-bumps aligned to sub-micron precision. If any step of this process drifts, yield collapses. The only way to maintain yield is continuous inline process control measurement. That is what Onto sells.
End-use applications: 1. AI accelerator packaging (CoWoS, SoIC, EMIB): Die-level bump inspection, overlay metrology 2. HBM memory stacking: 3D bump height inspection, die-stacking alignment 3. Logic at 3nm/2nm (GAA): OCD metrology for nanosheet gate structures 4. 3D NAND memory: Film thickness, etch depth control 5. Power semiconductors (SiC/GaN): Surface defect inspection, materials characterization (Semilab)
TAM: | Segment | TAM Estimate | CAGR | |—|—|—| | Overall process control (metrology + inspection) | $7-12B (2025) | ~5-8% | | Onto’s stated “AI Era” TAM | $12.3B (projected) | Higher | | Advanced packaging specific subset | ~$2-3B growing to $5B+ by 2030 | ~15-20% |
Onto’s SAM: primarily optical metrology, macro inspection, and advanced packaging. KLA owns electron beam inspection and much of the wafer defect review segment.
Market share: Onto has ~8-12% of the total process control TAM. In advanced packaging inspection specifically, Onto has meaningful share (Dragonfly is the leading macro-inspection platform for packaging); in OCD metrology, Onto has ~20-25% share with share gains at GAA nodes.
Demand inflection: AI infrastructure capex is the most powerful demand driver the semiconductor equipment industry has seen in a decade. TSMC, Samsung, SK Hynix, and Intel are all spending at record levels on advanced packaging capacity to serve Nvidia, AMD, Google, Amazon, Microsoft. Every CoWoS unit, every HBM stack, every EMIB module requires process control inspection. Advanced packaging equipment spend is growing faster (~15-20% CAGR) than front-end capex.
Supply constraint: There is no meaningful supply constraint in process control equipment itself – Onto can expand production. The constraint is in the measurement recipes and applications expertise required to qualify tools at new nodes. This is a human capital constraint, not a hardware one, and it plays to incumbents with established customer relationships.
Inventory cycle: Semiconductor capex had a soft cycle in 2023 (revenue declined 19%) and began recovering in 2024 (+21%). FY2025 was flat (+1.8%) at the overall level, but the softness was concentrated in Q3 2025 (revenue -13.5% YoY), driven by a reset in advanced packaging orders between old and new tool generations (G3 to G5 transition effect + Semilab integration). Q4 2025 re-accelerated to record $267M; backlog doubled.
Coming shortage/glut: Looking 12-36 months out, the sector is set up for a shortage of advanced packaging capacity. Every AI cloud customer (Amazon, Google, Microsoft, Meta) is building out AI inference infrastructure. This is a multi-year capex cycle, not a one-quarter event.
What has changed in the last 24 months: 1. AI packaging became mandatory. Every Nvidia GPU above H100 requires CoWoS-L or larger. Every AMD MI300 requires advanced packaging. This wasn’t true two years ago — “standard” packaging was still the mainstream. 2. HBM became a primary memory architecture. HBM3 and HBM4 require die-stacking to tolerances that demand dedicated inspection tools. The $240M SK Hynix agreement is evidence that the memory industry has accepted this as a permanent, high-intensity need. 3. GAA transistors entered high-volume manufacturing. Samsung’s 3nm and TSMC’s 2nm both use gate-all-around nanosheet transistors. These require a new class of OCD metrology that KLA’s tools struggle with but Atlas G6 specifically addresses.
What the sell-side is missing: The consensus focuses on Onto as a play on advanced packaging (Dragonfly/HBM), but underweights the GAA OCD opportunity at logic nodes. SemiAnalysis estimated logic OCD alone could represent $400M+ annual revenue by 2025. If that trajectory continues to 2nm and 1.4nm, the Atlas franchise could be as large as Dragonfly over a 3-5 year horizon.
The AI semiconductor cycle is in the early stages of a multi-year capex up-cycle driven by AI infrastructure build-out. Onto is positioned at two specific nodes within this cycle: (1) advanced packaging inspection, where every HBM stack and CoWoS module requires its tools, and (2) OCD metrology at GAA nodes, where a technology transition has created a specific window for share gain vs. KLA. The Semilab acquisition adds a third leg (materials characterization) that enlarges the SAM and is accretive to margins. The business has no debt, generates $300M+ in FCF, and has backlog at two quarters of revenue. The “why now” is concrete: AI packaging is structural, not cyclical, and Onto’s tools are qualification-locked into the ramps happening now.
| Name | Title | Tenure | Background |
|---|---|---|---|
| Michael P. Plisinski | President & CEO | 11 years (CEO from 2015 at Rudolph; ONTO from Oct 2019) | Built career in semiconductor capital equipment; founded Counterpoint Solutions (acquired by August Technology 2003); EVP/COO Rudolph Technologies before CEO; >25 years in semicap; B.S. Computer Science UMass; Harvard AMP |
| Brian K. Roberts | CFO | Since June 16, 2025 | 20 years CFO experience; 11 years at public companies; most recently CFO Sensata Technologies (NYSE: ST, ~$4B revenue); “track record of scaling high-growth organizations” per company release |
| Shirley Chen | SVP, Customer Success | Since June 16, 2025 | 25 years semiconductor experience; 18 years progressive sales leadership at KLA (KLAC) and Thermo-Fisher |
| Yoon Ah Oh | SVP & General Counsel | Multi-year | Legal and governance leadership |
Key executive changes: Both the CFO and SVP Customer Success are new as of June 2025, representing a partial refreshing of the C-suite below the CEO. The prior CFO (Mark Sefcik) and prior sales leader departed. The timing (mid-2025, post-Semilab acquisition announcement) suggests these changes were partly restructuring-related and partly succession/professionalization as the company crossed $1B in revenue.
Founder-led vs. professional management: Plisinski is not a founder — he is a professional semicap executive. He has run the business with a consistent strategy (advanced packaging + advanced nodes + specialty devices) for a decade. The Rudolph/Nanometrics merger he orchestrated in 2019 was viewed as a “merger for survival” against KLA dominance — a reasonable strategic move that has played out well.
| Name | Role | Shares Owned | % of Outstanding | Est. Value | How Acquired |
|---|---|---|---|---|---|
| Michael P. Plisinski | CEO | ~150,492 (post-Jan 2025 sales) | ~0.31% | ~$46M at $307 | Mix of grants/PSUs + some open-market holding |
| Brian K. Roberts | CFO | ~3,096 + PSUs (recent grant) | Very small | ~$1M | Grants (new in role) |
| Yoon Ah Oh | SVP & GC | ~724 shares (recent PSU vest) | Minimal | ~$222K | PSU vesting |
Insider buying vs. selling: CEO Plisinski sold 85,000 shares in January-February 2025 for approximately $18M total ($219-211/share average). These were executed under pre-established 10b5-1 plans (automatic diversification). No disclosed open-market purchases by management in the last 12 months. The selling was at prices significantly below the current $308 — Plisinski diversified near the cycle bottom.
Assessment: Insider ownership is low (~0.86% aggregate) and the CEO has been selling systematically. This is not a red flag in isolation — post-merger management teams routinely have low ownership, and 10b5-1 sales are standard diversification. However, the lack of any open-market purchases is a yellow flag: management is not demonstrating conviction through personal capital.
| Name | Holdings in ONTO | Where Is the Majority? |
|---|---|---|
| Plisinski | ~$46M (150K shares) | Likely ONTO is not the majority of net worth given the $18M+ in stock sales; has diversified |
| Roberts | Minimal (new in role) | Elsewhere (prior Sensata equity/comp) |
No evidence of shell entities, complex cross-holdings, or related-party structures in available public records. The company’s 10-K related-party disclosures do not flag material insider transactions.
No red flags identified. The company has: - No dual-class share structure - No disclosed poison pill - Standard NYSE governance - No activist 13D filings - No disclosed related-party transactions of concern in SEC filings (DEF 14A/10-K)
The Semilab acquisition was a straightforward product-line purchase from a European instruments company. No insider-controlled entities were involved.
M&A track record: - Rudolph/Nanometrics merger (2019): The defining transaction. Combined two sub-scale process control companies to create a credible #3 vs. KLA. Revenue has since grown from ~$450M to $1B+. The merger has created value. - Inspectrology acquisition: Small bolt-on adding overlay metrology capability. Details not fully public; appears to have been smoothly integrated. - Semilab product lines (Nov 2025, ~$495M): Largest acquisition. Thesis is extension into materials characterization — adjacent, logical, accretive per management. Execution is being tested in FY2026; $100-120M revenue guidance is the initial signal. The Q3 2025 GAAP margin compression reflects purchase accounting amortization from this deal.
Buyback track record: Limited buyback activity. Board authorized $100M in Nov 2020; modest usage, resulting in essentially flat share count. The Semilab acquisition consumed cash that might otherwise have gone to buybacks.
Capex efficiency: Capex is only ~3% of revenue ($18-32M/year). Every dollar of capex drives significant revenue. This is a genuinely asset-light model.
Equity issuance: Minimal. Share count flat at ~49M for 5 years. No ATM. No equity issuance to fund the Semilab deal.
Capital allocation grade: B+. The Rudolph/Nanometrics merger was an A-grade strategic move. The Semilab acquisition is B (logical strategically, expensive at $495M for $120M annual revenue = ~4x revenue multiple; time will tell if it reaches the growth rates that justify the price). Buyback activity has been modest — the company has not returned capital aggressively but has not destroyed it either.
CEO total comp: Plisinski’s comp includes base salary, annual cash bonus, and long-term equity (RSUs + PSUs). PSUs vested at 200% in FY2024 due to strong TSR performance — indicating that at least part of the equity program is TSR-indexed.
SBC as % of revenue (FY2024): ~$60-70M SBC / $987M revenue = ~6-7%. This is in the normal range for a mid-cap semicap company.
Incentive metrics: PSU vesting is TSR-based (evidenced by 200% vesting on strong 2024 stock performance). This aligns management with shareholders for long-term returns.
Does management eat their own cooking? The CEO holds ~$46M in stock. But he has been systematically selling, not buying. No net addition to personal position in the last 12 months.
Full board composition requires DEF 14A (not fetched due to 403 restrictions on IR pages). Based on available data: - Board Chair: Steven R. Roth - CEO/Director: Michael Plisinski - Standard mix of independent directors with semiconductor industry backgrounds - No dual-class shares; no poison pill disclosed; standard NYSE governance - No publicly visible activist activity
| Dimension | Rating | Key Finding |
|---|---|---|
| Skin in the Game | Yellow | Low insider ownership (~0.86%); CEO has been selling systematically under 10b5-1 |
| Holdings Concentration | Yellow | CEO’s majority of net worth likely not in ONTO after diversification sales |
| Shell / Cross-Holdings | Green | No red flags; clean corporate structure |
| Capital Allocation | Green/Yellow | Rudolph/Nanometrics merger excellent; Semilab expensive but logical; buybacks limited |
| Compensation Alignment | Green | PSUs are TSR-indexed; SBC in normal range |
| Governance Quality | Green | Standard NYSE; no poison pill; no dual-class |
| Litigation / Enforcement | Green | No public litigation of concern |
| Overall Management Grade | B+ | Competent professional management; strategic track record is good; insider selling is the main yellow flag |
| Company | Ticker | Segments Overlapping | FY2025 Revenue | Share | Moat Type | Pure-Play? |
|---|---|---|---|---|---|---|
| KLA Corporation | KLAC | Wafer inspection, OCD, overlay | ~$10.6B | ~50% process control | Dominant installed base; analytics software; breadth | Partial (process control is 75%+ of rev) |
| Applied Materials | AMAT | OCD/VeritySEM metrology, process integration | ~$27B total | ~10-15% process control | Process integration; fab bundling; breadth | No (process control is a small fraction) |
| Nova Measuring | NVMI (NASDAQ) | OCD, XPS, optical/x-ray metrology | ~$800M | ~8-10% | Secondary OCD position; XPS/optical combo | Yes |
| Camtek | CAMT (NASDAQ) | Advanced packaging inspection | ~$350M | ~5% (packaging) | APAC-focused; competitive pricing; throughput | Yes |
| Rigaku Holdings | 268A (TSE) | X-ray metrology | ~$600M (semicap subset) | Niche x-ray | X-ray measurement expertise; scientific instrument brand | No (semicap is subset) |
Competitive dynamics: - KLA is the 800-pound gorilla. Its dominance across wafer inspection, overlay, and process control is structural. Onto cannot match KLA’s breadth or installed base. - Where Onto wins: (1) GAA OCD metrology where Ai Diffract 3D modeling software outperforms KLA’s tools at 3nm/2nm nodes; (2) advanced packaging macro inspection where Dragonfly has no direct KLA equivalent. - Where Onto is vulnerable: (1) Nova is a credible secondary OCD player and may benefit if KLA extends more aggressively into advanced packaging; (2) Camtek can compete on price in packaging inspection at APAC OSATs.
Competitive moat analysis: - Switching costs: The strongest moat. Once a fab qualifies Onto’s tool and builds process recipes atop it, switching requires 6-18 months of re-qualification. This is the primary barrier. - IP and software: Ai Diffract modeling software is a genuine technical differentiation vs. KLA at GAA nodes. This is a time-limited advantage — KLA will eventually close the gap as GAA recipes become standardized. - Packaging specialization: Dragonfly has first-mover advantage in advanced packaging inspection. This is more durable than OCD differentiation because the packaging inspection market is less developed and the competitive field is less crowded. - Semilab materials characterization: Unique capability — no single competitor can offer OCD + inspection + materials characterization in one vendor. This is a genuine moat extension.
5-year lock-up test: Yes, with moderate discomfort. Onto is mission-critical to its customers, has no debt, generates strong FCF, and addresses a structurally growing market. The main risk is KLA encroaching on the packaging niche — but the 5-year secular trend (AI packaging growing) is a strong tailwind. I would hold this business locked up for 5 years.
Unique economic engine: Onto’s economic engine is: tools installed at leading-edge fabs generate a recurring service/spares tail (~20-25% of revenue) while new tool placements ride the fab capex cycle. The incremental cost to add one more OCD recipe is essentially zero (software), while the value delivered (yield protection) is enormous. The source of uniqueness is the recipe library and customer integration, which takes years to build and cannot be easily replicated.
Blank-check disruptor: Yes — KLA could theoretically develop a Dragonfly equivalent and bundle it with its advanced packaging inspection suite. KLA has the capital. The protection is: (1) recipe libraries and customer qualification cycles create significant drag even if KLA builds equivalent hardware; (2) advanced packaging inspection is not KLA’s historical core strength; (3) Onto has a 5-7 year head start in packaging metrology. A $1B+ KLA investment could close the gap in 3-5 years, but not tomorrow.
Quality verdict: High-quality / durable with the caveat that the OCD moat is time-limited (GAA differentiation window will close as the node matures) while the packaging moat is more durable.
Industry structure: Highly consolidated. KLA has ~50% share. Applied Materials + Nova + Onto + Camtek + a few others share the remaining 50%. Consolidation history: Rudolph + Nanometrics = Onto (2019); KLA absorbed most of its smaller competitors decades ago.
Barriers to entry: Extremely high. A new process control equipment company would need to: 1. Develop the electromagnetic modeling software (10+ years, PhD-intensive) 2. Build the optical hardware (precision sub-nm alignment, 300mm wafer handling) 3. Qualify at a leading fab (18-36 months, requires customer cooperation) 4. Build a service and applications engineering network globally
This is not a market where a startup can enter. The threat is from existing equipment companies expanding scope (KLA into packaging inspection; Applied expanding OCD).
Cyclical or secular: Both. Secular growth driven by chip complexity increases (more process steps needed). Cyclical overlaid due to fab capex spending cycles. The AI infrastructure cycle is creating a secular demand shift that partially decouples from the traditional semiconductor cycle.
Where are we in the cycle: Coming out of a trough. FY2023 was the bottom (-19% revenue). FY2024 recovered (+21%). FY2025 was flat (+1.8%) at the aggregate level but with an H2 2025 dip (Q3 2025 -13.5% YoY) related to order timing. Q4 2025 re-accelerated to record $267M and backlog doubled. FY2026 is guided for significant re-acceleration (+26% analyst consensus).
Historical cycle data: Semicap downturns typically last 12-24 months; recoveries last 24-36 months. The current recovery (from FY2023 trough) is now 2 years old. AI infrastructure demand suggests this recovery could be longer and more durable than prior cycles.
KLA’s packaging push: KLA has been expanding into advanced packaging inspection. Any KLA product announcement targeting Dragonfly’s core use cases (bump inspection, packaging defect detection) would be a direct threat. KLA has the customer relationships, scale, and capital to credibly enter.
Camtek acceleration: Camtek (CAMT) is growing fast in packaging inspection ($350M revenue, growing ~30%+ in 2024). Camtek has a lower-cost offering that is competitive in APAC markets (OSATs, smaller memory makers). If Camtek captures the cost-sensitive end of the market, it pressures Onto’s pricing.
AI-enabled process control disruption: It is theoretically possible that future AI-driven process control (wafer-to-wafer virtual metrology, AI prediction of process drift without physical measurement) could reduce the demand for hardware metrology tools. This is a 10+ year threat, not a near-term concern.
China alternatives: China is building its own process control equipment industry (Naura, AMEC, others) to reduce dependence on US/foreign suppliers. China is already <3% of Onto’s revenue, so direct China revenue loss is not the concern — the concern is whether Chinese manufacturers can eventually qualify Chinese alternatives and remove the global TAM growth from that geography.
1. Organic revenue growth: FY2024 grew 21% organically (no M&A in that year). FY2025 grew 1.8% (essentially flat) — the slowdown reflects a digestion phase after Q3 2025 weakness. FY2026 is guided to grow ~26% with Semilab adding $100-120M. Core business (ex-Semilab) growing approximately 12-14% in H1 2026 per guidance. The durability of organic growth is strong given the secular AI packaging demand; the year-over-year comparison in H2 2026 will also benefit from a weak H2 2025 comp.
2. Margins: Non-GAAP gross margin has been consistent at 54-55%. GAAP margin compressed in FY2025 due to Semilab purchase accounting amortization. Operating margin compressed in FY2025 for the same reason: FY2024 EBIT margin 18.9% → FY2025 13.2%. This is a one-time step-down, not structural margin erosion. Non-GAAP operating margin was ~25% in Q4 2025. FY2026 should see margin recovery as integration costs normalize and Semilab revenue ramps.
3. Capital intensity: Very low. Capex $18-32M/year (~3% of revenue). FCF margin expanded from 11.8% (FY2022) to 29.8% (FY2025). This is an asset-light business getting more efficient over time.
4. Capital deployment: Flat share count; no debt; cash pile used for Semilab ($495M). The balance sheet is now leaner ($640M net cash vs. $852M pre-acquisition) but still very strong. No equity raise, no dilution.
| Q1 2024 | Q2 2024 | Q3 2024 | Q4 2024 | Q1 2025 | Q2 2025 | Q3 2025 | Q4 2025 | |
|---|---|---|---|---|---|---|---|---|
| Revenue | $228.9M | $242.3M | $252.2M | $263.9M | $266.6M | $253.6M | $218.2M | $266.9M |
| YoY growth % | +15.7% | +21.0% | +21.0% | +25.5% | +16.4% | +4.7% | -13.5% | +1.1% |
| QoQ % | - | +5.9% | +4.1% | +4.6% | +1.0% | -4.9% | -13.9% | +22.3% |
| 2nd derivative (YoY ∆) | - | +5.3pp | 0pp | +4.5pp | -9.1pp | -16.3pp | -34.5pp | +14.6pp |
Assessment: The second derivative turned sharply negative in Q2-Q3 2025 — revenue deceleration was severe. Q4 2025 saw a meaningful recovery (+22.3% QoQ). The Q3 2025 trough was driven by: (1) digestion of Semilab integration, (2) customer ordering patterns between Dragonfly G3 and G5 generations, (3) macro uncertainty. Q4 2025 re-acceleration and the backlog doubling suggest the second derivative has turned positive again. Q1 2026 guidance ($275-285M, implied +4-7% QoQ) and Q2 2026 guidance (>$300M, implied +5-10% QoQ further) confirm the positive trajectory.
Implied exit rate: At Q2 2026 guidance of $300M+, annualized run rate = $1.2B+, consistent with analyst FY2026 consensus of $1.27B.
| Metric | Value |
|---|---|
| Market cap | $15.31B |
| Enterprise value | $14.69B |
| P/E (TTM) | 110.7x |
| Forward P/E (FY2026E) | ~46x |
| EV/EBITDA (TTM) | 53.8x |
| P/FCF (TTM) | 51.1x |
| EV/Revenue (TTM) | 14.6x |
| P/Sales | 15.2x |
| FCF yield | ~2.0% |
| Dividend yield | None |
| 52-week range | $85.88 – $316.00 |
| Beta (5Y) | 1.47 |
| PEG Ratio | 1.45 |
Valuation context: The TTM P/E of 111x is heavily distorted by the FY2025 GAAP margin compression from Semilab purchase accounting. The forward P/E of 46x on $6.65 consensus EPS for FY2026 is more meaningful. KLA trades at ~30x forward earnings; Applied Materials at ~22x. Onto commands a premium to KLA partly because it is a smaller, purer-play with higher growth expectations. At 46x forward, the stock is pricing in a strong FY2026 delivery (26% revenue growth, $6.65 EPS). A miss would be punishing.
| Metric | FY2023 | FY2024 | FY2025 | LTM | FY2026E |
|---|---|---|---|---|---|
| Revenue | $816M | $987M | $1,005M | ~$1,005M | $1,270M |
| YoY growth | -19% | +21% | +1.8% | +1.8% | +26% |
| Gross profit | $420M | $515M | $500M | $500M | ~$690ME |
| Gross margin % | 51.5% | 52.2% | 49.7% | 49.7% | ~54%E |
| EBIT | $116M | $187M | $133M | $133M | ~$290ME |
| EBIT margin % | 14.2% | 18.9% | 13.2% | 13.2% | ~23%E |
| Net income | $121M | $202M | $137M | $137M | ~$325ME |
| Net margin % | 14.8% | 20.4% | 13.6% | 13.6% | ~26%E |
| EPS (diluted) | $2.46 | $4.06 | $2.78 | $2.78 | $6.65E |
Note: FY2025 compression is Semilab purchase accounting (amortization of acquired intangibles). Non-GAAP EPS for FY2025 was higher than GAAP. FY2026E from analyst consensus (8 analysts).
| Metric | FY2023 | FY2024 | FY2025 | LTM | FY2026E |
|---|---|---|---|---|---|
| Operating cash flow | $172M | $246M | $328M | $328M | – |
| Capex | ($23M) | ($32M) | ($29M) | ($29M) | – |
| Free cash flow | $149M | $214M | $300M | $300M | – |
| FCF margin % | 18.3% | 21.7% | 29.8% | 29.8% | – |
| Net cash (no debt) | $698M | $852M | $640M | $622M | – |
| Net debt / EBITDA | N/A | N/A | N/A | N/A | – |
| ROIC | ~12% | ~18% | ~14% | 14.2% | ~22%E |
Balance sheet note: The decline in net cash from $852M to $622M reflects the $495M Semilab acquisition offset by $300M FCF generation in FY2025. The business is cash generative enough to fully fund operations, R&D, and moderate M&A without any external financing.
ROIC: 14.2% (TTM). FY2024 peak was ~18%. FY2025 compression reflects Semilab goodwill/intangibles entering the capital base.
WACC estimate: For a semicap company with beta ~1.47, cost of equity ~10-12% (CAPM), no debt, WACC is approximately equal to cost of equity: ~10-12%.
ROIC vs. WACC: ROIC ~14% vs. WACC ~10-11% = spread of ~3-4%. Value-creating, but the spread has compressed from ~8-9% at FY2022 peak. FY2026E ROIC of ~22% would represent a meaningful spread expansion if achieved.
Using quarterly data (YoY delta):
| Q1 2024 vs Q1 2023 | Q2 2024 vs Q2 2023 | Q3 2024 vs Q3 2023 | Q4 2024 vs Q4 2023 | |
|---|---|---|---|---|
| Delta Revenue (YoY) | ~+$29M | ~+$38M | ~+$52M | ~+$46M |
| Delta Gross Profit (YoY) | ~+$18M | ~+$27M | ~+$36M | ~+$34M |
| Incremental GM | ~62% | ~71% | ~69% | ~74% |
| Delta EBIT (YoY) | ~+$8M | ~+$15M | ~+$22M | ~+$18M |
| Incremental EBIT Margin | ~28% | ~39% | ~42% | ~39% |
| Q1 2025 vs Q1 2024 | Q2 2025 vs Q2 2024 | Q3 2025 vs Q3 2024 | Q4 2025 vs Q4 2024 | |
|---|---|---|---|---|
| Delta Revenue (YoY) | +$37.7M | +$11.3M | -$34.0M | +$3.0M |
| Delta Gross Profit (YoY) | +$24.9M | -$6.1M | -$25.8M | -$8.6M |
| Incremental GM | ~66% | Negative | ~76% (on decline) | Negative |
| Delta EBIT (YoY) | +$20.4M | -$16.6M | -$29.4M | -$28.6M |
| Incremental EBIT Margin | ~54% | Negative | Negative | Negative |
What the incrementals tell us: - 2024 incrementals were excellent: 60-74% incremental gross margins demonstrate strong operating leverage. As revenue grew through 2024, every incremental dollar drove ~$0.65-0.74 in gross profit. - 2025 incrementals deteriorated sharply in Q2-Q4 2025: The Semilab acquisition introduced significant purchase accounting charges (amortization of acquired intangibles) that depressed reported GAAP margins. Non-GAAP incrementals in the same period would look better. - The Q3 2025 revenue decline (-$34M YoY) with -$25.8M gross profit decline implies incremental GM was ~76% even on the downside — meaning the cost structure deleveraged at a high rate. This is expected for a software-heavy company (fixed costs + high incremental margins = high deleverage on revenue declines). - FY2026 restoration: If revenue accelerates as guided (+26%), incrementals should return to the 60-70% gross / 35-45% EBIT range seen in strong 2024 quarters.
Distortions: Semilab purchase accounting is the primary distortion. Non-GAAP margins (~54-55% gross, ~25% operating) are cleaner signals of underlying business economics.
Peer comparison:
| Company | Ticker | Fwd P/E | EV/EBITDA | EV/Rev | FCF Yield | Growth (NTM) |
|---|---|---|---|---|---|---|
| Onto Innovation | ONTO | ~46x | 53.8x | 14.6x | 2.0% | +26%E |
| KLA Corporation | KLAC | ~30x | ~22x | ~10x | 3.5% | +12-15%E |
| Nova Measuring | NVMI | ~35x | ~28x | ~10x | 2.8% | +15-20%E |
| Camtek | CAMT | ~28x | ~20x | ~7x | 4.0% | +25-30%E |
| Applied Materials | AMAT | ~22x | ~18x | ~6x | 4.0% | +8-12%E |
Onto trades at a premium to all peers on every multiple. The premium is partially justified by: 1. Higher revenue growth outlook (Semilab contributing) 2. Purer-play on AI packaging vs. diversified players (AMAT) 3. Small/mid-cap premium (more re-rating potential)
Implied expectations at current price: At $308 and FY2026E EPS of $6.65, the market is pricing roughly 46x for ~26% revenue growth and ~139% EPS growth. The EPS re-rating from $2.78 to $6.65 is largely normalization from depressed FY2025 GAAP levels (Semilab amortization normalizes over 3-7 years). The revenue growth is more conservative.
Where is the upside: If advanced packaging revenue grows >30% in FY2026 (as guided) AND the Semilab lines track toward $120M+ (high end of guidance), FY2026 revenue could exceed $1.3-1.4B. At the same multiples, the stock would be worth $350-400+.
DCF sensibilities: A rough DCF using $300M FCF (FY2025 base), 20% growth for 3 years, then decelerating to 10% terminal growth rate, at 10% discount rate, yields an intrinsic value in the $280-320 range — broadly consistent with where the stock is trading. Upside requires either faster growth or multiple expansion.
| Tailwind | Mechanism | Durability |
|---|---|---|
| AI accelerator packaging | Every Nvidia/AMD/Google AI chip requires CoWoS/HBM inspection — Onto’s Dragonfly is a required tool | 3-10 year |
| HBM memory scaling | HBM4 and beyond require tighter die-stacking tolerances; Onto holds $240M agreement | 3-7 year |
| GAA/2nm transistors | New OCD metrology required at every fab adopting gate-all-around; Atlas G6 targets this | 2-5 year |
| Advanced packaging CapEx cycle | Intel ($4B), TSMC ($3.6B), Samsung ($2B), OSATs all investing | 3-5 year |
| Power semi (SiC/GaN) growth | EV and industrial electrification driving SiC/GaN capacity; Semilab FAaST/CnCV/MBIR serve this | 5-10 year |
SK Hynix Volume Purchase Agreement: - Counterparty: SK Hynix (world’s #2 memory maker; 000660.KS) - Value: $240M+ through 2027; $60M specifically for 3D bump metrology systems - What it covers: Dragonfly G3 systems for HBM3/HBM4 die stacking inspection; 3D bump height measurement - Status: Awarded, contracts signed; revenue flowing over 2025-2027 - Revenue impact: Approximately $80-100M/year implied run-rate over the remaining term
| Risk | Likelihood | Existing Mitigants | Mgmt De-risk Plan | Can It Be Closed? |
|---|---|---|---|---|
| Semicap capex cyclicality | High (structural) | Diversified across logic/memory/packaging; service revenue ~20-25%; SK Hynix contract anchors 2025-2027 | Customer diversification; new product categories (Semilab) expand revenue base | No — inherent in capital equipment business; only partially mitigated |
| KLA competitive encroachment on packaging | Medium | Dragonfly head start (5-7 years); customer recipe libraries; Semilab materials differentiation | Atlas G6 + Dragonfly G5 new-product cycle; pricing competitiveness; deepening customer co-development | Partially — cannot prevent KLA from entering, but switching costs and head start protect for 3-5 years |
| Semilab integration execution | Medium | Product line acquisition (not full company merger); experienced acquisition team; immediate EPS accretion | FY2026 revenue guidance $100-120M; integration of FAaST/CnCV/MBIR into Onto sales force and service network | Closable — FY2026 execution will validate; monitoring revenue vs. $100-120M guidance is the primary metric |
| Valuation risk (46x forward P/E) | High (if consensus misses) | Strong growth guidance; backlog doubled; SK Hynix anchor; Semilab incremental | Management guided Q1 2026 $275-285M and Q2 >$300M — clear near-term targets | Not closable at current valuation — any execution miss would re-price the stock significantly |
| Intel execution risk | Medium | Intel is estimated <10% of revenue; not a single-point-of-failure | New customer development (cloud OSATs, SK Hynix, Samsung) reduces Intel dependency | Partially closable through customer diversification |
| China export control escalation | Low (largely resolved) | China <3% of FY2025 revenue; already de-emphasized | No strategic reliance on China revenue | Largely closed |
| CEO key-person risk | Low-Medium | Competent succession bench (new CFO Roberts is credible); Plisinski not sole architect of current strategy | PSU-based retention; board succession awareness assumed (not publicly disclosed) | Partially — no public succession plan; risk is manageable given the depth of the management team |
What would make the thesis wrong: 1. KLA announces a competitive Dragonfly equivalent and starts winning at leading-edge customers 2. AI capex cycle moderates sharply (macro recession, GPU demand correction) 3. Semilab integration fails — FY2026 revenue comes in well below $100M guidance 4. Intel’s advanced packaging business stalls, reducing a key growth customer 5. GAAP margin never recovers from Semilab amortization to the level consensus expects
Downside target: If consensus EPS of $6.65 is missed by 20% ($5.32 EPS) and multiple compresses to 35x (KLA-like), the stock would trade at $186 — approximately 40% below current price. This is a significant risk given the demanding valuation.
| Holder | Type | Who They Are | Shares | % Outstanding | Source |
|---|---|---|---|---|---|
| Vanguard Group | Passive index | World’s largest asset manager; holds as index constituent | 5,576,305 | ~11.3% | 13F |
| William Blair Inv. Mgmt | Active growth | Chicago growth/quality equity manager; significant conviction position | 1,726,587 | ~3.5% | 13F |
| Geode Capital | Passive/quant | Fidelity’s indexing arm; +45% recent increase (index constituent weighting change) | 1,127,532 | ~2.3% | 13F |
| Wellington Management | Active multi-strategy | Major Boston active manager | 1,218,548 | ~2.5% | 13F |
| Paradigm Capital | Active small/mid | NY-based active equity, tech specialist | 1,125,838 | ~2.3% | 13F |
| AQR Capital | Quantitative | Major quant hedge fund; +315% recent increase (factor model signal) | 615,651 | ~1.2% | 13F |
| Fisher Asset Management | Active growth | Ken Fisher’s global equity platform | 701,999 | ~1.4% | 13F |
| D.E. Shaw | Quant HF | Major quant fund; -48% reduction | 517,189 | ~1.0% | 13F |
| Millennium Management | Multi-strat HF | Multi-strategy hedge fund; tactical | 495,000 | ~1.0% | 13F |
Total institutional: ~93-95% of outstanding shares. Insider aggregate: ~0.86%. Activist positions: None disclosed (no 13D filers). Short interest: ~9.0% of float; 2.88 days to cover (Dec 2025). Modest.
Notable flows: AQR added 315% (quantitative factor signal), Geode added 45% (index inclusion/weighting), D.E. Shaw reduced 48% (tactical exit). William Blair’s ~3.5% position as an active manager is the most notable conviction holder.
| Metric | Value |
|---|---|
| Total analysts | 8 |
| Strong Buy | 5 |
| Buy | 3 |
| Hold | 0 |
| Sell | 0 |
| Avg price target | $290-317 |
| High target | $350 (Stifel, Apr 2026) |
| Low target | $160 |
| FY2026E Revenue consensus | $1.27B |
| FY2026E EPS consensus | $6.65 |
Assessment: 100% buy-side consensus with no sells or holds is unusual and can signal analyst coverage captured by deal flow or management optimism. However, with only 8 analysts, the sample is small and represents the genuine conviction of a relatively small coverage community. The $350 Stifel target is the highest after a recent upgrade. The stock’s current price ($308) is at the high end of analyst range before the Stifel upgrade.
Conviction level: Medium-High
Position sizing rationale: At 46x forward earnings and 53x EV/EBITDA, this is a high-multiple story requiring execution. For a conviction-weighted portfolio, ONTO merits a 3-5% position as part of a broader AI infrastructure/semicap basket. The premium valuation does not support a top-decile position unless the AI packaging thesis accelerates beyond consensus.
Entry strategy: The stock has already re-rated significantly (+156% in 52 weeks). Dollar-cost averaging or scaling in over 2-4 quarters as Q1-Q2 2026 results confirm the revenue re-acceleration is lower risk than a single entry at current prices.
Stop-loss / re-evaluation triggers: - Q1 2026 revenue misses $275M guidance by >5% → reassess - Semilab FY2026 revenue tracking below $80M → reassess - Any KLA announcement of competitive Dragonfly equivalent → reassess - FY2026 GAAP gross margin fails to recover above 50% → reassess
Add vs. trim: - Add: Q2 2026 revenue exceeds $310M+; Semilab revenue tracking above $110M; Atlas G6 wins at TSMC for 2nm - Trim: Stock exceeds $350 without fundamental catalyst; any two of the stop-loss triggers fire