Onto’s miss was not the story; packaging capacity and Q4 EPS reset were

The actionable read from this quarter is that Onto Innovation printed a messy revenue line but gave enough hard numbers to treat Q3 FY2025 as a capacity-transition quarter rather than a demand-break quarter. What was priced in was straightforward: the Street wanted $220.2 million of revenue and $0.89 of EPS. What actually surprised was more nuanced: revenue came in at $218.2 million, a -0.9% miss, while EPS came in at $0.92, a +2.9% beat. That split matters because the share debate is not whether Q3 was optically clean, it was not, with revenue down -14.0% QoQ and -13.5% YoY in the quarterly history. The debate is whether the miss signals a fading process-control cycle or a mix-timing air pocket before a packaging-led rebound. Management’s Q4 framework points to the latter: revenue is guided to $250 million to $265 million, operating margins to 24% to 26%, and EPS to $1.18 to $1.33. The variant perception is that the Street may have anchored on the -0.9% top-line miss and the Q3 revenue deceleration, while the company gave a more investable signal in Q4 conversion and customer tool demand.

That interpretation only works if Q3 margins and earnings show that the revenue miss did not expose structural deleverage. On the Street-comparison basis, the evidence is the +2.9% EPS beat on a -0.9% revenue miss. On management’s own call basis, CFO Brian Roberts said “Revenue for the quarter was slightly ahead of the midpoint of our previous guidance range at $218.2 million,” and also said adjusted EPS was “towards the high end of our guidance range at $0.92.” The wording is useful because it separates external disappointment from internal execution against a prior range: investors who expected $220.2 million were disappointed, but the company did not describe Q3 as a demand shock. The gross-margin data are harder to reconcile because the quarterly history shows Q3 FY2025 gross margin at 50.7%, while Roberts said “Gross margin for Q3 2025 was 54% and includes approximately a 1 percentage point impact related to tariffs.” Those are different reporting bases in the data pack, so the defensible conclusion is not that margin was clean, but that margin commentary was better than the GAAP-like history alone implies. The stock call depends on that distinction: if investors use the 50.7% history line as the only margin anchor, Q4’s margin expansion sounds less compelling; if they use management’s non-GAAP operating framework, the Q4 EPS guide becomes the more relevant signal.

The financial trajectory reinforces the trough argument because revenue history had already shown a sharp downshift before management guided a Q4 rebound. Revenue moved from $266.6 million in Q1 FY2025 to $253.6 million in Q2 FY2025 and then $218.2 million in Q3 FY2025, with sequential growth rates of +1.0%, -4.9%, and -14.0%. That was the part the market could reasonably fear: Q3 FY2025 looked like the third consecutive quarter in which the revenue cadence deteriorated. But the guide calls for $250 million to $265 million in Q4, described as “15% to 21% sequential growth,” and CEO Michael Plisinski put the midpoint growth at “approximately 18%.” The key is that the Q4 guide does not merely recover the revenue missed versus the $220.2 million Street estimate; it implies a return toward the $253.6 million Q2 FY2025 level and back into the range that investors had seen before the Q3 air pocket. The quarterly history also shows that subsequent reported periods reached $266.9 million in Q4 FY2025 and $291.9 million in Q1 FY2026, with revenue QoQ of +22.3% and +9.4%, respectively, which supports the argument that Q3 was not the start of a straight-line revenue reset.

The source of that rebound is more important than the rebound itself because Onto’s debate is increasingly about advanced packaging exposure rather than generic wafer-fab equipment beta. Plisinski said the “greatest contributor” to Q4 growth is 2.5D packaging customers, where revenue is expected to “nearly double from the third quarter,” driven by Dragonfly system demand. That is the strongest sentence in the call because it names the segment, the sequential shape, and the product. Q3 specialty devices and advanced packaging revenue was $113 million, or approximately 52% of revenue, and Roberts guided a rebound to approximately $150 million in Q4, with the market expected to finish slightly higher than $500 million for the full year. The market may miss that the Q4 rebound is not evenly distributed across all end markets; it is concentrated in packaging, where Onto also says customer discussions point to “as much as 20% more tools” for 2026 capacity needs across 2D subsurface and 3Di inspection technologies. That 20% number is not a booking, and management called discussions “early stages,” but it gives PMs a concrete watchpoint: the upside case is less about one quarter of catch-up and more about customers adding tool counts into 2026.

The mix argument also explains why the advanced-node data should not be read as a contradiction, even though it was softer in Q3. Advanced nodes generated $54 million, or 25% of Q3 revenue, as DRAM and NAND revenue decreased sequentially from Q2. That is not what a clean AI infrastructure bull story usually looks like, and it is the main reason to avoid a simplistic “AI wins, buy everything” conclusion. But Roberts also said full-year 2025 advanced nodes revenue is expected to double to approximately $300 million from $148.5 million in full-year 2024. In other words, Q3’s $54 million advanced-node contribution was weak sequentially, while the full-year advanced-node guide still calls for roughly a doubling on management’s own basis. The conflict is real: Q3 memory-related advanced-node revenue softened, while full-year advanced-node revenue remains guided to approximately $300 million. The right interpretation is that Onto’s near-term earnings inflection is packaging-led, not memory-led, even if advanced-node revenue remains part of the full-year growth algorithm.

That customer mix has direct read-throughs for TSMC, Samsung, and Intel, but the magnitudes matter. For TSMC, which appears in the data pack as a customer for process control metrology and inspection, Onto’s Q4 commentary around 2.5D packaging and Dragonfly is a positive signal specifically tied to packaging capacity, with specialty device and advanced packaging expected to move from $113 million in Q3 to approximately $150 million in Q4. For Samsung, listed as a customer for metrology and inspection for GAA nanosheet process, the relevant number is not the packaging rebound but the advanced-node framework: Q3 advanced nodes were $54 million, or 25% of revenue, while full-year 2025 advanced nodes are expected to reach approximately $300 million versus $148.5 million in full-year 2024. For Intel, listed as a customer for advanced logic metrology and inspection, the same advanced-node numbers are the read-through, but Q3’s DRAM and NAND sequential decrease means the call does not isolate logic strength from memory weakness. The supply-chain paragraph is also notable for what it lacks: the data pack lists no suppliers to Onto, so there is no named upstream beneficiary to underwrite from this print. The named customer read-through is therefore concentrated on buyers of Onto’s inspection and metrology capacity rather than on a mapped supplier chain.

The competitive context makes the quarter look less like a broad process-control outlier and more like a company-specific transition with mixed execution. In the peer table, ONTO’s latest reported quarter shows $291.9 million of revenue, 50.1% gross margin, and +9.5% revenue YoY. That compares with KLAC at $3,415.1 million of revenue, 61.1% gross margin, and +11.5% revenue YoY, and NVMI at $235.3 million of revenue, 57.7% gross margin, and +10.3% revenue YoY. The peer point is not that Onto screens cheapest or best; the data pack gives no valuation. The point is that Onto’s scale is closer to NVMI than KLAC, while its 50.1% gross margin trails NVMI’s 57.7% and KLAC’s 61.1%. That makes the Q4 operating-margin guide of 24% to 26% important: Onto needs the packaging rebound and international factory ramp to show up in conversion, not just revenue. If that happens, the market can look through the Q3 top-line miss. If it does not, the margin gap versus NVMI and KLAC remains the cleaner comparative bear case.

The factory ramp is the operational hinge between the demand commentary and the margin guide. Plisinski said Onto shipped over 30% of third-quarter tools from extended factories in Asia and is “on pace to be capable of shipping over 60% of our production demand from our international locations by the end of the first quarter of 2026.” That commitment matters because Q4 guidance includes both growth and tariff pressure: Roberts said the Q4 gross-margin expectation includes an anticipated percentage point impact of tariffs, or approximately $2.5 million of cost, primarily due to inbound tariffs on raw material imports. Management also expects approximately 50 basis points of sequential gross-margin improvement at the midpoint of Q4 revenue guidance, despite that tariff drag. This is where the bull case has a measurable operational test. If Onto can move from over 30% of Q3 tools shipped from extended Asian factories toward over 60% of production demand capacity by the end of Q1 2026, then the margin bridge has a plausible manufacturing basis. If the ramp slips, the $2.5 million tariff cost and the Q4 operating-expense burden become harder to offset.

The Semilab transaction adds another layer to that operating leverage debate because it consumes cash but arrives at a lower stated value than the original deal terms. Roberts said that once the acquisition closes, expected “in the coming weeks,” Onto will pay Semilab $432.3 million in cash and issue 641,771 shares of common stock. He also said the total transaction value, based on Onto’s closing price as of June 27, 2025, is approximately $495 million, a decrease of about $50 million from the original terms of the deal. The cash-generation data make that funding less alarming than it might otherwise be: cash from operations increased sequentially to $83 million from $58 million in Q2. The investment case still requires discipline, because Q4 operating expenses are guided to approximately $77 million, and the extra week adds approximately $3 million in incremental operating expenses, representing approximately 120 basis points of operating margin. The offset is that operating margins are still guided to 24% to 26% in Q4, which means management is asking investors to accept near-term deal integration and opex noise while underwriting a revenue rebound and better conversion.

The call delivery supports the thesis, but not with unqualified confidence, and the tone history is useful precisely because it shows a split between prepared confidence and Q&A restraint. For Q3 FY2025, sentiment was 0.24, guidance_tone was 0.65, prepared_sentiment was 0.48, Q&A sentiment was 0.09, uncertainty was 80.3, and qa_evasiveness was -66.3. Compared with Q2 FY2025, sentiment fell from 0.37 to 0.24, while guidance_tone rose from 0.41 to 0.65; that is the profile of a call where management’s forward framework was more constructive than the quarter’s overall language. The Q&A sentiment of 0.09, down from 0.27 in Q2 FY2025, says analysts did not receive the same warmth in interaction that the prepared guide conveyed. That conflict is not fatal to the thesis, but it prevents overstatement: management sounded most confident when framing Q4 and 2026 packaging demand, while the live exchange retained caution. The later tone history also shows Q4 FY2025 sentiment at 0.42 and Q1 FY2026 at 0.47, with Q1 FY2026 uncertainty down to 72.4 from 90.9 in Q4 FY2025, which is consistent with the idea that the Q3 uncertainty was transitional rather than permanent.

The EPS guide is the cleanest investable number because it puts a profit range around the rebound rather than leaving investors to infer leverage. Brian Roberts gave the commitment directly: “Earnings per share for the fourth quarter is expected in the range of $1.18 to $1.33 per share, assuming an estimated tax rate of approximately 13% to 15% and about 49.4 million shares outstanding.” That range is the counterweight to the Q3 Street revenue miss. The company beat Q3 EPS by +2.9% despite revenue missing by -0.9%, then guided Q4 EPS well above the Q3 actual $0.92 on the Street-comparison basis and above the Q3 diluted EPS of $0.57 in the quarterly history. The distinction between EPS bases matters, but the direction is consistent: management is telling investors that Q4 revenue growth of 15% to 21% should come with operating margins of 24% to 26%. In a semiconductor equipment name, that is the difference between a backlog air pocket and a margin reset. The market may be pricing the former as if it were the latter.

What to watch next is therefore concrete. First, Q4 revenue must land within the $250 million to $265 million guide, with the midpoint growth of approximately 18% and the stated 15% to 21% sequential growth intact; anything below that would break the trough thesis. Second, specialty device and advanced packaging should approach the approximately $150 million Q4 target after $113 million in Q3, and management should either repeat or sharpen the “nearly double” 2.5D packaging language tied to Dragonfly demand. Third, operating margin needs to print inside the 24% to 26% range despite approximately $77 million of operating expenses, approximately $3 million of extra-week operating expense, and approximately $2.5 million of tariff cost. Fourth, the international factory ramp should progress from over 30% of Q3 tools shipped from extended Asian factories toward the stated ability to ship over 60% of production demand from international locations by the end of the first quarter of 2026. Fifth, by the next call, the Semilab closing expected “in the coming weeks” should be resolved around the stated $432.3 million cash payment and 641,771 shares. Confirmation would be Q4 EPS inside $1.18 to $1.33 with packaging near approximately $150 million and no slippage in the factory ramp; the thesis breaks if Q4 revenue misses $250 million, operating margin falls below 24%, or management walks back the “as much as 20% more tools” 2026 capacity signal.