Camtek’s EPS miss is the wrong focal point: the real signal is a back-half capacity claim the Street is not crediting

The print should be read as a timing reset, not a demand rollover, and that is the variant perception. The Street had enough in the numbers to expect another uncomplicated upside quarter: revenue was estimated at $127.2 million, the company delivered $128.1 million, and the surprise was +0.7%. What was not priced in was the earnings leakage: EPS came in at $0.81 against $0.83, a -2.4% surprise, despite the top line exceeding expectations. That separation matters because investors often treat process-control beats as clean read-throughs to margin leverage, but this quarter says Camtek is spending and building ahead of the next leg. The wrong trade is to punish the EPS miss as evidence that the cycle is maturing; the better trade is to ask whether the back-half setup is credible enough to absorb one quarter of near-term revenue softness.

What actually surprised was not the record revenue itself, because management had already conditioned investors to expect AI-related inspection demand to remain the center of gravity. The surprise was that earnings did not follow the revenue beat, and that Q1 guidance immediately pulled the conversation away from the record Q4. Ramy Langer framed Q4 as a high-water mark in company terms, saying, “Fourth quarter revenues reached a quarterly record of $128 million, representing an increase of 9% year-over-year.” That wording earns attention because it pairs a record with only single-digit growth, which is exactly the tension in the stock: the business is at an all-time revenue level, but reported growth has decelerated enough that investors can plausibly argue the easy AI inspection ramp is behind it. The thesis here is the opposite: the deceleration is visible, but management’s order commentary and capacity language suggest the growth curve is being deferred into the second half rather than extinguished.

The financial trajectory supports that distinction because revenue has not rolled over; it has flattened near a new high while margins have held near the upper end of the company’s recent range. Revenue moved from $117.3 million in Q4 FY2024 to $128.1 million in Q4 FY2025, while gross margin sat at 50.0% in the latest reported quarter. The issue is slope, not level. QoQ growth had slowed to +1.7% by Q4 FY2025, so the stock could not rely on sequential acceleration to carry the multiple. But the gross margin profile did not show the deterioration one would expect if customers were pushing out tools or if mix were degrading materially. This is why the EPS miss is less damning than it looks: the top line beat, the gross margin stayed at 50.0%, and the pressure showed up below gross profit.

The margin discussion is where the market’s skepticism has some footing, because the company’s own call basis shows expenses absorbing the benefit of record sales. Moshe Eisenberg said operating expenses were $28.7 million, up from $23.1 million in the fourth quarter of last year, and operating margin was 28.6%. That combination explains why EPS missed despite the revenue beat. It also prevents a lazy bullish reading of the quarter: Camtek is not giving investors pure incremental margins at the moment. But the spending is occurring while full-year revenues reached $496.9 million and while management is talking about capacity beyond the prior ceiling, so the key question is whether the expense base is being positioned for a larger revenue run-rate. If second-half growth materializes, the same operating-expense line that hurt the EPS surprise can become the setup for better absorption; if it does not, the print becomes the first sign that the company overbuilt into a slower packaging cycle.

The Q1 guide is the cleanest near-term bear point, and it deserves to be stated plainly. Management guided revenue “around $120 million,” which sits below Q4’s $128.1 million and creates a visible pause after the record quarter. Langer’s exact phrasing, “In line with this, our revenue guidance for the first quarter is to be around $120 million,” was deliberately plain rather than promotional, and that matters because the company did not try to disguise the step-down. The offset is that Moshe Eisenberg tied the guide to a very specific cadence, saying the company expects “growth expected in the second quarter and more significant growth in the second half of 2026.” That wording is the crux of the long case: investors do not need to believe in a straight-line acceleration from Q4; they need to believe that Q1 is the trough of a delivery cadence and not the start of an order air pocket.

The order data makes that back-half claim more investable than a generic management promise. Langer disclosed a $25 million order from an IDM customer for multiple Hawk systems and said it brought the customer’s recent total to approximately $45 million. This is not enough by itself to underwrite the year, but it is a real datapoint that anchors the second-half narrative in named product demand rather than broad WFE rhetoric. It also matters that the order is for multiple Hawk systems, because Camtek’s investor debate centers on whether AI packaging inspection demand can broaden and persist beyond a narrow HBM surge. The full-year mix points in that direction: approximately 50% of revenue was driven by AI-related products, and 20% came from other advanced packaging applications. In other words, the bull case is not simply “AI helps”; it is that AI-related inspection is already half the business, while non-AI advanced packaging is large enough to keep the revenue base from depending on one single application.

That mix has direct read-throughs for Camtek’s customers, and the numbers make the implications sharper than a generic advanced-packaging comment. For TSMC, Camtek’s approximately 50% AI-related revenue mix suggests inspection intensity remains tied to AI packaging capacity additions rather than only wafer starts. For ASE Group, the 20% contribution from other advanced packaging applications supports continued inspection demand outside the most visible AI product lines. Samsung and SK Hynix should be read through the HBM and packaging lens: Camtek’s $45 million recent total from one IDM customer does not name them, so it cannot be assigned to either, but it does show that large integrated device manufacturers are still placing multi-system orders into this equipment category. There are no named suppliers in the data pack, so the supply-chain implication is customer-side concentration and packaging capex intensity, not a parts or component read-through.

The peer context also argues against treating Camtek’s Q4 as a company-specific demand break. In the process-control group, Onto Innovation posted +9.5% revenue YoY with 50.1% gross margin, which is close to Camtek’s +9.2% revenue YoY and 50.0% gross margin. KLA remains structurally larger and higher-margin at $3,415.1 million of revenue and 61.1% gross margin, so Camtek is not being valued as a broad process-control franchise with KLA-like economics. The more relevant comparison is that Camtek’s current growth and margin profile line up with advanced-packaging inspection peers rather than with a collapsing niche supplier. That matters for positioning because a stock can derate on decelerating growth, but it should not be priced as if its end-market signal has diverged materially from the process-control cohort when the closest peer markers still show positive growth.

The call delivery reinforces the idea that management is more confident in the guide path than the headline EPS miss suggests, though it is not a clean read. The tone history shows guidance_tone rising to 0.78 in Q1 FY2026 from 0.72 in Q4 FY2025, while ai_optimism increased to 0.45 from 0.28. That is a meaningful rhetorical shift toward the forward story. The conflicting number is uncertainty, which rose to 47.4 from 42.3, so management sounded more constructive while also carrying more ambiguity. I would not ignore that conflict: it fits the substance of the quarter, where Q1 is soft, second-half growth is emphasized, and the company is asking investors to bridge from a near-term revenue step-down to a larger capacity target. The more important delivery point is that qa_evasiveness fell to 20.3 from 43.5, which suggests the higher uncertainty was not simply avoidance in Q&A.

The capacity language is the strongest variant-perception datapoint because it changes the scale of what management is willing to underwrite. When asked about prior capacity potential, Langer responded, “I think it is well over $700 million in capacity.” The phrase matters because it is not framed as a distant ambition or a vague market-size comment; it is a direct answer to a capacity question. Put beside full-year revenue of $496.9 million, the capacity statement implies management is planning around a materially larger business than the one just reported, although the data pack does not give a timing bridge to that level. This is also where the downside case is cleanest: if the company has built operating expenses for a revenue base it cannot fill, Q4’s EPS miss is not a one-off absorption issue. But if Q2 improves and the second half inflects as guided, investors will likely revisit whether a company at roughly the $0.5 billion revenue milestone deserves to be valued on a subscale-growth framework.

The distinction between company-accounting language and Street-comparison numbers also matters for interpretation. The Street-comparison print was $128.1 million of revenue and $0.81 of EPS, which produced the +0.7% revenue surprise and -2.4% EPS surprise. On the call, Eisenberg described net income for the fourth quarter of 2025 as $40.7 million or $0.81 per diluted share, while also citing cash and equivalents, including deposits and marketable securities, of $851.1 million. The cash figure is not the main thesis, but it reduces the probability that Camtek’s spending ahead of demand becomes a balance-sheet problem. The operating question is not liquidity; it is whether expenses and capacity additions are being matched to real customer pull. Accounts receivable being down by $22 million to $90.8 million helps at the margin, because it argues against the record quarter being financed through looser collections, but it does not answer the second-half demand question by itself.

The market may therefore be mispricing the quarter by over-indexing on the EPS miss and underweighting the explicit cadence management gave for 2026. A clean print would have had a revenue beat, EPS beat, and Q1 guide that kept sequential momentum intact. Camtek delivered only the first of those three. That is why the stock should not be chased as if nothing changed. But the evidence does not support the more bearish conclusion that the AI packaging inspection cycle has peaked: AI-related products were approximately 50% of full-year revenue, other advanced packaging was 20%, and a single IDM’s recent Hawk orders totaled approximately $45 million. The company is asking investors to accept a first-quarter pause, but it is doing so while giving more specific evidence of product demand than many semiconductor equipment companies provide at this stage of a cycle.

What to watch next is straightforward and numerically bounded. The thesis is confirmed if Q1 revenue lands around $120 million and management still points to growth in the second quarter with more significant growth in the second half of 2026. It is strengthened if gross margin stays near the recent 50.0% level while operating expenses do not keep rising from $28.7 million without a matching revenue acceleration. It is broken if the Q2 commentary fails to show growth after the Q1 reset, if the $45 million IDM Hawk order does not translate into visible shipment momentum, or if management backs away from capacity “well over $700 million.” The next quarter is not about another record; it is about whether the guided trough behaves like a trough. If Q1 comes in near $120 million and the second-half language remains intact, the EPS miss should be treated as the cost of building into demand, not the start of an earnings-quality problem.