Layer 2 · CAMT — Camtek Ltd.
One-line thesis
Camtek is the pure-play HBM bump-inspection franchise — a small-cap with the cleanest single-thesis HBM-cycle exposure in the equipment cohort, the post-Q1 -7% sell-off creates an entry, and the H2 26 ramp guided at >25% above H1 is the cleanest forward-visibility in the small-cap space.
What Camtek physically does
Camtek makes optical-inspection tools that examine the µbumps (microscopic solder bumps, typically 25-40µm in diameter) and the through-silicon-via openings on HBM die wafers and on the underlying logic + interposer wafers before, during, and after the stacking process. The flagship product is the Eagle G6 automated optical inspection platform (and the newer Eagle XT variant) — a multi-camera, multi-illumination optical inspection tool that scans 300mm wafers at high throughput looking for bump-height anomalies, bump-coplanarity defects, via-opening blockages, residue contamination, and any defect that would cause a TC-bonded HBM stack to fail mechanically or electrically.
The physical inspection challenge is non-trivial. A 12-Hi HBM3e stack contains roughly 12,000 µbumps per layer, with each bump approximately 25-30µm in diameter and the entire bump array required to remain within ±1µm coplanarity tolerance across the full die surface. A single anomalous bump destroys the entire stack — and HBM stacks cost upwards of $300 each at HBM3e generation, rising toward $500-700 at HBM4. The yield economics make in-line bump inspection an obligatory step rather than a sampling step, which is why every HBM-producing fab runs Camtek (and Onto) tools at multiple inspection passes per wafer cycle.
In the HBM stacking flow specifically, Camtek tools sit at three inspection steps: pre-bond (inspecting the bumps on each die before stacking begins), post-bond (verifying the bond integrity after each layer), and incoming-wafer inspection at the OSAT for CoWoS-L underfill quality. In CoWoS-L specifically — the packaging platform for NVIDIA Blackwell and Rubin — Camtek tools also inspect the redistribution layer (RDL) bumps at the silicon interposer level, and the next-generation Eagle platform under development carries explicit hybrid-bonding-pad alignment metrology to position for the HBM5 / HBM6 hybrid-bonding transition.
In rough numbers, a single HBM4 capacity expansion step (e.g., a new 25k-wafer-month HBM4 line at SK Hynix M16) requires roughly $20-30M of Camtek equipment, plus several million of recurring service revenue per year for software updates, recipe development, and consumables. The Eagle platform is the dominant tool at this inspection tier and Camtek holds approximately 70-80% share among the HBM customer set, with a particularly deep installed-base position at SK Hynix that has been built across HBM2, HBM3, and HBM3e generations and is now extending into HBM4.
The product roadmap matters: the Eagle XT introduced in late 2025 added higher-throughput multi-camera scanning specifically designed for the 16-Hi HBM stacks coming on HBM4 generation, and the next-generation tool (rumoured Eagle G7 or under a new product name) will integrate the hybrid-bonding alignment metrology under development. Camtek's bet on the hybrid-bonding inspection paradigm is the principal R&D priority and the principal terminal-risk mitigation.
The financial print
Q1 2026 (reported May 12 2026) printed revenue of $121.7M against $119.7M consensus — beat — and EPS of $0.70 against $0.69 (slight beat). The guide for Q2 was raised to $129-131M and management commented that the H2 2026 revenue should run >25% above H1, on the back of HBM4 capacity additions at SK Hynix, Samsung, and Micron. The stock sold off -7.4% pre-market on the print, principally on operating margin compression to 22.4% (versus historical 28%+) — a reflection of mix-shift toward the lower-margin HBM-specific tools and increased R&D ahead of HBM5 / hybrid-bonding inspection tool development.
The margin compression deserves attention. Camtek's historical gross margin has run 49-51% non-GAAP with operating margin in the 28-32% range. The Q1 print at 22.4% operating margin is roughly 600 basis points below the prior-year quarter — driven by three factors: (1) a one-time mix-shift toward HBM-specific tools that carry lower margins than Camtek's historical advanced-packaging products, (2) elevated R&D spend related to the hybrid-bonding inspection roadmap and the next-generation Eagle XT platform, and (3) higher engineering deployment costs as Camtek scales its customer-support headcount into Korean and Taiwanese HBM lines. Management has guided that operating margin should recover toward 25-27% in H2 2026 as the volume ramp normalises and the R&D step-up moderates. Whether that guidance holds is the principal financial-risk variable.
The stock at $192 is 1m -7.4%, 3m +5% — meaningfully behind the broader semi equipment cohort over the same window. The underperformance is doing the entry-point work: the relative compression versus KLAC (+28% over the same window), LRCX (+24%), and ONTO (+18%) creates the asymmetric R/R that the bull case requires. Q2 prints August 11. Consensus is the guided $129-131M revenue range plus EPS around $0.85-0.95. The setup post-pullback is materially better — the stock has compressed back to a multiple closer to ONTO and BESI on a forward-earnings basis, and the H2 revenue ramp visibility is the most concrete in the equipment cohort, with the SK Hynix HBM4 line build-out, the Samsung PyeongTaek HBM4 ramp, and the Micron Idaho IPF1 capacity step all contributing through the August-October order book.
Customer mix
SK Hynix is the largest customer at approximately 35-40% of revenue (HBM3e and HBM4 line inspection), with Samsung Memory at 20-25% (PyeongTaek HBM4 build-out), Micron at 15-20% (Idaho HBM ramp), and TSMC + the OSAT cohort (ASE, Amkor, SPIL) at the balance — roughly 15-20% combined for CoWoS-L underfill and RDL inspection. The customer concentration in HBM is real but the underlying demand is structurally underpinned by the HBM unit growth forecast of 60-80% in 2026.
Competitive context
The direct competitor is Onto Innovation (also in this Tier A list) — Onto's bump-inspection tools (the Dragonfly platform inherited from Rudolph Technologies and the Atlas in-die metrology) compete head-on with Camtek's Eagle line. Onto is structurally larger and has the broader product portfolio (metrology + inspection across multiple application areas) while Camtek is the pure-play HBM bump-inspection specialist with the deeper customer-specific recipe library.
The competitive dynamic is closer to a duopoly than a single-vendor monopoly: Camtek + Onto collectively hold ~90%+ of the global bump-inspection market, with Hitachi High-Tech and a handful of Korean local vendors (Kornic, Park Systems) in single-digit share niches. The two companies divide the customer set: Camtek is dominant at SK Hynix, Onto is dominant at Micron, both compete at Samsung. The structural valuation question — discussed under Onto below — is whether the duopoly compresses to a single winner over the next 3-5 years or whether the HBM cycle is large enough to sustain both at premium economics.
Camtek's structural advantages within the duopoly are three. First, the SK Hynix relationship is multi-decade — Camtek's tools have been the reference inspection platform at the M14 / M16 HBM lines through HBM2 / HBM3 / HBM3e generations, and the recipe-development relationship is locked in at a level that an Onto displacement attempt cannot easily replicate inside a single product transition. Second, the Israeli engineering base (Camtek is headquartered in Migdal Haemek, Israel) operates with materially lower fixed costs than Onto's New York / Massachusetts cost structure, contributing to historical gross margin advantages. Third, the product focus is sharper — Camtek's R&D is entirely on the bump-inspection / advanced-packaging vertical, while Onto's spend is divided across bump-inspection, in-die metrology, lithography, and the broader portfolio.
The downside of the focus is precisely the customer concentration risk: in any scenario where SK Hynix HBM share compresses (per the Hanmi share-loss discussion, ASMPT is taking marginal HBM4 socket at SK Hynix), Camtek's customer exposure is more concentrated than Onto's. The H1 26 print suggested this is not yet impacting the order book — but the second-derivative read on SK Hynix capex through 2027 is the variable to monitor.
Terminal risk
The terminal risk is the same as Hanmi's at the equipment-cohort level: hybrid bonding adoption for HBM5 / HBM6 generations changes the inspection paradigm. If HBM moves to hybrid bonding (BESI-led, ASMPT-led) starting in 2027-28, the µbump inspection franchise compresses as fewer wafers carry traditional µbumps. Camtek's mitigation is the hybrid-bonding alignment metrology roadmap (under development); whether Camtek can win at the next-generation inspection tier is the binary question on a 3-5 year horizon.
Bull case
The base case is FY26 revenue of $530-550M (versus FY25 around $430M), FY27 revenue of $620-680M, gross margin holding 48-50% and operating margin recovering from the Q1 22% trough toward 28-30% as the H2 mix normalizes. At 30x forward EPS — premium to historical Camtek multiple but consistent with HBM-pure-play exposure — that prints the stock at $235-265 in 2027, +22-38% from spot.
The upside case is HBM unit growth landing at the +80% end of the Trendforce range and Camtek capturing full share at HBM4 + early HBM5 — which lifts FY27 revenue toward $750M and EPS toward $7.00. At 32x forward, the stock prints $260-290 in 2027.
The smaller-cap nature of Camtek versus its competitors creates additional upside leverage on share-take scenarios. A single 25k-wafer-month HBM4 capacity expansion at Samsung PyeongTaek would contribute roughly $20-30M of incremental Camtek tool revenue — meaningful at the ~$500M base. If Samsung's HBM4 program ramps materially faster than the consensus path (which assumes Samsung remains a quarter behind SK Hynix through 2026-27), Camtek's revenue line takes a step-function uplift that the small-cap valuation amplifies. Conversely, the small-cap nature also means a single program cancellation has 5-10% revenue impact — the asymmetric exposure cuts both ways.
Gap / bear case
Three concerns. First, operating margin compression to 22% in Q1 is the warning sign that the customer mix is shifting toward lower-margin tools — if this persists rather than normalising, the FY26 EPS print disappoints despite revenue beating. Second, the duopoly with Onto means pricing discipline is harder to maintain than at single-vendor franchises like KLAC. Third, valuation premium to ONTO has historically run 20-30% on forward multiples and is currently elevated — a re-rating toward Onto-level multiples would compress Camtek by 15-20%.
Optionality
Two real options. First, the SK Hynix Wide TC Bonder qualification cycle in H2 26 brings net new tool order volume that Camtek captures in inspection alongside Hanmi's bonder side. Second, the hybrid-bonding inspection roadmap — if Camtek can credibly position for the HBM5 generation alongside BESI hybrid bonding, the long-tail terminal risk diminishes and the multiple expands.
The trade
CAMT — BUY 8/10. Entry: $190-195 post-Q1-pullback — a better entry than pre-print. Add on $175-180 if the broader equipment cohort consolidates further. Position size: 1.5-2.5% of NLV — smaller than the large-cap equipment names because of the single-customer-concentration risk and small-cap volatility. Stop: daily close below $158 (the December 2025 base and the 200-day moving average). Catalyst date: August 11 Q2 print; SK Hynix Q2 print July 24 (read-through on HBM4 capex); September SEMICON Taiwan commentary; HBM5 hybrid-bonding inspection product disclosure expected late 2026. Trim/exit triggers: operating margin compression below 22% on a second quarter; any verifiable hybrid-bonding inspection loss at SK Hynix or Samsung; revenue guide cut below the H2 +25% step. Conviction: the cleanest pure-play HBM bump-inspection name and the best small-cap equipment exposure. Pair with ONTO at 3-4% combined for the bump-inspection duopoly trade.