Coherent’s beat is not the point: the market is underpricing the AI optics capacity handoff into Q3

The print changes the debate because the upside was not merely a low-quality revenue beat. What was priced in was a company expected to deliver $1,641.5 million of revenue and $1.21 of EPS, with investors already aware that AI data-center demand was improving the optical cycle. What actually surprised was $1,685.6 million of revenue, a +2.7% surprise, and $1.29 of EPS, a +6.6% surprise, followed by guidance for revenue between $1.7 billion and $1.84 billion and non-GAAP EPS between $1.28 and $1.48. The variant perception is that the earnings power is being reset by mix and capacity, not by a one-quarter snapback. Coherent’s reported sequence moved from $1,581.4 million in Q1 FY2026 to $1,686.0 million in Q2 FY2026, with gross margin moving from 36.6% to 37.0%, and the Q3 FY2026 row shows $1,805.6 million with 37.7% gross margin. That is the pattern PMs should underwrite: revenue keeps stepping up while gross margin does not give back the gains.

That trajectory matters because the market could have dismissed the quarter as a beat against a stale Street number, but the internal mix data says the business accelerating is the right one. James Anderson framed the pro forma company as growing revenue “9% sequentially and 22% year over year,” explicitly excluding the recently divested aerospace and defense business, while Sherri Luther gave the company account view as “record $1.69 billion, up 7% sequentially from the first quarter and up 17% year over year.” Those two statements are not interchangeable, and that distinction is important: the Street-comparison print was $1,685.6 million versus $1,641.5 million, while management’s call basis was $1.69 billion. The point is that both bases point the same way, even after portfolio pruning. The historical revenue line had already climbed from $1,348.1 million in Q1 FY2025 to $1,434.7 million in Q2 FY2025, $1,497.9 million in Q3 FY2025, $1,529.4 million in Q4 FY2025, $1,581.4 million in Q1 FY2026, and $1,686.0 million in Q2 FY2026. The Q3 FY2026 line at $1,805.6 million extends that staircase rather than showing a pause after the beat.

The margin story explains why this is more than optical revenue chasing hyperscale demand. Gross margin has moved from 34.1% in Q1 FY2025 to 35.5% in Q2 FY2025, 35.2% in Q3 FY2025, 36.6% in Q4 FY2025, 36.6% in Q1 FY2026, 37.0% in Q2 FY2026, and 37.7% in Q3 FY2026. Management’s non-GAAP lens is higher, with Sherri Luther saying Q2 non-GAAP gross margin was 39%, including a 24 basis point improvement compared to the prior quarter and a 77 basis point improvement compared with the year-ago quarter. The Q3 non-GAAP gross margin guide of 38.5% to 40.5% puts the midpoint at 39.5%, and Luther tied that to a longer-term target model of over 42%. The market’s risk case is usually that AI optics volume carries ramp costs, yield drag, or pricing give-up. The numbers do not support that yet: Q2 revenue beat by +2.7%, EPS beat by +6.6%, and gross margin on the historical series still moved to 37.0% from 36.6%.

The capacity signal is the clearest reason the Street may be too anchored to current-quarter EPS. Capex increased to $154 million in Q2 from $104 million in the prior quarter and $106 million in the year-ago quarter, and that spend came alongside explicit demand language around 800 gig, 1.6 T, OCS, and CPO. Anderson’s most important commitment was not the product list, but the production cadence: “we're basically at 80% of that target capacity already.” That line matters because it turns the debate from whether AI optical demand exists to whether Coherent can convert wafer starts into revenue without breaking the margin path. The Q3 revenue guide of $1.7 billion to $1.84 billion includes only $5 million of revenue from the period before the close of the Munich product division sale in January, while the divested aerospace and defense business had contributed average quarterly revenue of $25 million with gross margin well below Coherent’s corporate gross margin over the past four quarters. In other words, the guide is asking the core business to grow through lost low-margin contribution, not to mask it.

That mix shift is most visible in data center and communications, where management put numbers around the part of the business that now matters most. Anderson said the data center and communications segment now accounts for over 70% of revenue, and that Q2 segment revenue grew 11% sequentially and 34% year over year. Within that, data center revenue grew 14% sequentially and 36% year over year. Those numbers explain why the company can simultaneously absorb higher operating expenses and still lift operating margin: Q2 non-GAAP operating expenses were $321 million versus $304 million in the prior quarter and $283 million in the year-ago quarter, but operating expenses as a percentage of revenue declined to 19% from 19.2% in the prior quarter and 19.7% in the year-ago quarter. SG&A as a percentage of revenue declined to 9.6% from 9.8% in the prior quarter and 10.2% in the year-ago quarter, while non-GAAP operating margin reached 19.9% versus 19.5% in the prior quarter and 18.5% in the year-ago quarter. The operating leverage is not theoretical; it is showing up while capex is rising.

The product read is that Coherent is no longer simply a transceiver-cycle proxy, even though transceivers are still doing the heavy lifting near term. Anderson named “growth in both 800 gig and 1.6 T transceivers” alongside “new products such as OCS and CPO solutions” and demand for DCI and scale across. The 800 gig and 1.6 T point supports near-term revenue, but OCS and CPO are what can change the multiple if they convert from product ramps into material dollars. Management also said OCS revenue should grow sequentially in the current quarter and coming quarters, tied to “over $2 billion of expected addressable market opportunity” over the coming years. The sell-side trap is to wait for OCS to reach a fully disclosed revenue line before changing the framework. The better approach is to watch whether the capacity, capex, and guide remain internally consistent: $154 million of Q2 capital expenditures, Q3 revenue between $1.7 billion and $1.84 billion, and non-GAAP gross margin between 38.5% and 40.5%.

That same product mix has direct read-throughs for the supply chain and customers, and the magnitudes matter. For NVIDIA, the print reinforces that optical and SiPho components for CPO are moving from roadmap language toward production urgency, because Coherent is already at 80% of target wafer-start capacity and is guiding Q3 revenue between $1.7 billion and $1.84 billion despite the Munich sale contributing only $5 million before close. For Broadcom, which is listed as a customer for optical components for Broadcom CPO, the relevance is that Coherent’s data center business grew 14% sequentially and 36% year over year, a direct signal that the optical attach around AI fabrics is not waiting for one architecture to become dominant. Zhongji Innolight, which uses EML and CW laser chips for transceivers, sees the same demand pulse through 800 gig and 1.6 T. On the supplier side, AXT Inc, as a provider of InP substrates and compound semi wafers, is the most exposed named supplier to wafer-start intensity, while 5N Plus benefits from demand for high-purity semiconductor metals including Ge, In, Se, and Te, and Fabrinet gets a positive transceiver manufacturing read from Coherent’s data center and communications segment now exceeding 70% of revenue.

The peer comparison also argues against treating Coherent as a generic photonics rebound. In the latest peer table, Coherent shows $1,805.6 million of revenue, 37.7% gross margin, and +20.5% revenue YoY. That gross margin is below LITE at 42.4%, but Coherent’s revenue base is larger than LITE’s $808.4 million, and its growth is well above 6923.T at +3.4%, 6965.T at +7.9%, and 6925.T at +5.5%. It is not the fastest-growing smaller optical name, since LITE grew +90.1%, AAOI grew +51.4%, and LASR grew +55.2%, but those came off revenue bases of $808.4 million, $151.1 million, and $80.2 million. The portfolio implication is different: Coherent offers scale exposure to AI optics with gross margin moving toward 37.7%, not just maximum beta to the optical upcycle. That distinction matters for PMs deciding between owning the broad AI interconnect chain and chasing smaller, higher-volatility optical names.

The call delivery supports the thesis, but it is not uniformly clean. The tone history shows Q2 FY2026 sentiment at 0.42, guidance_tone at 0.52, tone_confidence at 0.32, prepared_sentiment at 0.66, qa_sentiment at 0.34, ai_optimism at 0.70, uncertainty at 52.3, and qa_evasiveness at 45.8. Versus Q1 FY2026, guidance_tone fell from 0.58 to 0.52 while uncertainty rose from 42.1 to 52.3 and qa_evasiveness rose from 18.0 to 45.8. That is the conflict: prepared messaging improved from 0.61 to 0.66 and ai_optimism rose from 0.67 to 0.70, but Q&A quality deteriorated. The Q3 FY2026 tone line helps resolve some of it, with guidance_tone at 0.59, uncertainty at 45.9, and qa_evasiveness at 6.4, but tone_confidence falls to 0.25. The right read is not that management eliminated risk; it is that the directional guide and product demand language are stronger than the Q2 Q&A delivery.

That delivery nuance is especially relevant because management did not pretend the cost model is already at the target. Luther explicitly said the investor day SG&A target was 8% of revenue, while Q2 SG&A was 9.6% of revenue. She also said the R&D target was 10% of revenue, and Q2 total non-GAAP operating expenses were $321 million, with Q3 guided between $320 million and $340 million. The leverage case therefore requires revenue growth to do work while SG&A continues to compress from 9.6% toward 8% and gross margin moves from 39% non-GAAP toward the over 42% long-term model. That is why the Q3 guide is so important: revenue between $1.7 billion and $1.84 billion, non-GAAP gross margin between 38.5% and 40.5%, and EPS between $1.28 and $1.48 leave room for both capacity ramp and operating-expense absorption. If operating expenses land near $340 million without revenue near the upper end of the guide, the leverage story weakens.

The stock debate after this print should center on whether Coherent is entering a capacity-monetization phase or merely pulling forward AI optics demand. The evidence favors capacity monetization: data center and communications is over 70% of revenue, data center grew 14% sequentially and 36% year over year, capex stepped to $154 million, wafer starts are already at 80% of target capacity, and the Q3 revenue guide absorbs divestiture noise. The counterargument is that Q2 historical diluted EPS was $0.76 even as non-GAAP EPS was $1.29, and the tone history shows Q2 qa_evasiveness at 45.8. Those are real friction points. But the Street comparison is on the non-GAAP EPS basis, where the company beat $1.21 with $1.29, and the forward gross margin guide of 38.5% to 40.5% gives the company a concrete way to prove that the AI mix is accretive rather than just bigger.

What to watch next is deliberately narrow. For the next quarter, the thesis is confirmed if revenue lands inside or above the $1.7 billion to $1.84 billion guide, non-GAAP gross margin holds within 38.5% to 40.5%, and EPS tracks the $1.28 to $1.48 range while operating expenses remain within $320 million to $340 million. The break points are equally clear: revenue near $1.7 billion with operating expenses near $340 million would challenge the operating-leverage case, gross margin below 38.5% would imply ramp or mix friction, and any retreat from the 80% wafer-start capacity marker would weaken the capacity-monetization view. Product confirmation should come from sequential OCS growth in the current quarter and coming quarters, plus continued strength in 800 gig and 1.6 T transceivers. The next call also needs better delivery: Q3 FY2026 tone history should hold guidance_tone near 0.59, keep uncertainty closer to 45.9 than 52.3, and avoid a return of qa_evasiveness toward 45.8.