Coherent’s miss is the wrong signal: mix and product cycle, not demand, drove the print

The print means investors should not treat the revenue miss as a demand break. What was priced in was a beat-and-raise AI optical story: consensus sat at $1,547.7 million, and the setup expected datacenter strength to carry reported sales above that bar. What actually surprised was the composition of the result: revenue came in at $1,529.4 million, a -1.2% miss, while EPS came in at $1.00, an +8.7% beat. That combination is not what a weakening optical cycle usually looks like. It says the company is controlling product mix, cost, and below-the-line conversion better than revenue-only investors expected, while management removed approximately $20 million of Aerospace and defense revenue from the forward outlook because of the pending sale. The variant perception is that the stock should be judged less on whether Q4 cleared a stretched revenue estimate and more on whether the higher-margin optical portfolio can keep gross margin near the fiscal 2026 guide as lower-margin divestiture revenue exits.

That distinction matters because the financial trajectory has changed from recovery to operating leverage, even though the headline sales beat failed. Revenue has climbed from the FY2024 trough to a record $1.53 billion in Q4 FY2025, while gross margin reached 36.6% on the company’s reported quarterly history. The key is not that sales were up, since that was already embedded in estimates; the key is that gross margin has moved from the low-30s band into the mid-30s while revenue growth has become less dependent on broad industrial recovery. CFO Sherri R. Luther’s wording is useful because it ties the record to the two demand pools that matter for the next leg: “Fourth quarter revenue was a record $1.53 billion, up 2% sequentially from the third quarter and up 16% year-over-year, driven by growth in AI datacenter demand, coupled with the continuing recovery in telecom.” That sentence commits management to a dual-cycle explanation, AI and telecom, rather than a one-product inventory rebound.

The revenue chart is why the bear case cannot stop at the -1.2% miss. Coherent’s sales base has reset higher, but the sequential growth rate has naturally slowed as the company laps the first AI optical ramp. Management itself acknowledged the deceleration through the question from Vivek Arya, who framed data center and communications sequential growth as having gone “from 9% in March to 5% in June,” with September implied “at or somewhat below this number.” That is the cleanest articulation of what was priced in versus what surprised: investors wanted acceleration from 1.6T and OCS, while the reported and guided cadence looks more like digestion before the next product step. The mistake would be to equate that cadence with demand saturation. James Robert Anderson said data center revenue still grew 3% sequentially and 38% year-over-year in Q4, while communications grew 11% sequentially and 42% year-over-year. Those are not numbers that describe a broken cycle; they describe a cycle shifting from initial ramp to mix-sensitive execution.

The margin guide reinforces that interpretation because it points up even as the company removes lower-quality revenue. Luther guided revenue to $1.46 billion to $1.6 billion and non-GAAP gross margin to 37.5% to 39.5%. If management were seeing a plain demand problem, it would be difficult to guide gross margin above the reported quarterly 36.6% level while also excluding approximately $20 million in Aerospace and defense revenue. The divestiture detail matters: Luther said the business contributed average quarterly revenue of approximately $50 million over the past 4 quarters with a gross margin below Coherent’s average gross margin. That is not just portfolio cleanup language; it mechanically shifts the mix toward the optical franchises that are carrying AI datacenter and communications growth. The market saw the missing revenue dollars. The better read is that management is willing to sacrifice some reported revenue to raise the quality of the remaining sales base.

The EPS beat is the strongest evidence that quality already matters more than absolute revenue in this print. Luther said, “Fourth quarter non-GAAP earnings per diluted share was $1 compared to $0.91 in the prior quarter and $0.51 in the year ago quarter.” Use that as the company’s own accounting basis, separate from the Street comparison basis, because the surprise math belongs to the print. The operating point is that EPS doubled year-over-year on management’s non-GAAP basis despite revenue landing below consensus. Non-GAAP operating margin was 18% in Q4, compared with 15.4% in the year-ago quarter, and full-year non-GAAP operating margin was 17.8%, up 472 basis points from 2024. That is the leverage the market may be underpricing: the optical cycle does not need every quarter to be a revenue beat if mix, divestiture, and product ramps keep the margin floor rising.

The legitimate concern is that opex is also rising, so the margin thesis depends on revenue quality rather than cost starvation. Fourth quarter non-GAAP operating expenses were $307 million, and full-year 2025 non-GAAP operating expenses increased to $1.17 billion from $998 million in 2024. That is not a trivial spend step-up, and it means the EPS beat is not simply a product of underinvestment. The company is funding the portfolio while still improving operating margin, which supports Anderson’s claim that fiscal 2025 revenue increased by approximately 23% year-over-year to $5.81 billion, driven by data center and communications. The spending increase is the price of participation in 100-gig, 400-gig and 800-gig ZR/ZR+ Coherent transceivers, and the test for fiscal 2026 is whether those products scale fast enough to keep opex from compressing incremental margin.

That product-cycle point is where the second-order read-through is most important for customers and suppliers. For NVIDIA, Coherent’s 38% year-over-year data center growth supports continued pull for optical and SiPho components tied to CPO, but the 3% sequential growth rate says near-term capacity additions are not yet a vertical ramp. For Broadcom, the same CPO optical component exposure benefits from Coherent guiding non-GAAP gross margin to 37.5% to 39.5%, because that range implies pricing and mix discipline rather than a component glut. For Zhongji Innolight, the read-through is more competitive and supply-linked: Coherent’s growth in EML and CW laser chips for transceivers points to sustained demand in the optical engine, but the Street revenue miss shows end-system timing can still slip around a strong component cycle. On the supplier side, AXT Inc has the cleanest semiconductor materials read-through through InP substrates and compound semi wafers, while Fabrinet’s optical transceiver manufacturing exposure should follow the ZR/ZR+ ramp management expects to increase its revenue contribution throughout fiscal ’26 and beyond. 5N Plus is tied to the materials intensity of the same ramp through high-purity semiconductor metals, but Coherent’s removal of approximately $20 million of Aerospace and defense revenue means the cleanest demand signal is optical, not broad specialty materials.

The peer context argues the same way: Coherent is no longer just a recovery laggard inside photonics, but it is also not the highest-beta optical name. In the latest peer table, COHR shows $1,805.6 million of revenue, 37.7% gross margin, and +20.5% revenue YoY. That puts it near the margin level of 6521.T at 37.0%, but below LITE’s 42.4% gross margin, while LITE also shows +90.1% revenue YoY. The implication is not that Coherent should trade like the fastest rebounder. It should trade like a scale player whose margin is migrating toward higher-quality optical peers as AI datacenter and telecom recovery replace lower-margin businesses. Against AAOI at 29.1% gross margin and LASR at 33.1%, Coherent’s 37.7% margin supports a premium to lower-margin optical cyclicals, but the revenue miss argues against paying for uninterrupted acceleration.

The call delivery supports the thesis, with one important caveat. The tone history shows Q4 FY2025 guidance_tone at 0.44, above Q3 FY2025 at 0.29, while qa_evasiveness fell to 9.2 from 53.5. That is a notable combination: management became more constructive on guidance while answering questions more directly. The caveat is that uncertainty rose to 64.3, which conflicts with the lower evasiveness and suggests the team was direct about a still-variable ramp rather than pretending visibility is complete. That is exactly the tone investors should want in an optical transition quarter. Management is not hiding the timing issue around September, the pending sale, and the product ramps, but it is guiding margin in a range that requires better mix.

The tone chart also helps separate prepared enthusiasm from Q&A stress. Prepared_sentiment recovered to 0.40 in Q4 FY2025 after 0.00 in Q3 FY2025, while qa_sentiment stayed roughly stable at 0.33. That tells us the improved narrative was not just a scripted victory lap, because Q&A did not collapse when analysts pressed on deceleration. More importantly, ai_optimism returned to 0.60, close to the earlier Q2 FY2025 level of 0.59, after dipping to 0.44 in Q3 FY2025. The language and the numbers align: the company is presenting AI optical demand as durable, but not promising a straight-line sequential acceleration every quarter. The market often punishes that distinction in the first trade after a revenue miss; it tends to reward it later if margin and product ramps keep showing up.

Debt reduction adds another piece to the equity story, but it should not be overstated. Luther said fiscal 2025 total debt payments were $437 million, nearly 2x the amount made in fiscal 2024. That matters because optical component cycles can demand working capital and capacity investment at the wrong time, and lower leverage gives management more room to fund transceiver and CPO programs. It does not, by itself, change the demand thesis. The demand evidence is still the 51% fiscal ’25 growth in data center and communications and the Q4 communications growth of 42% year-over-year. The balance sheet work simply lowers the probability that Coherent has to underinvest just as 1.6T, OCS, and ZR/ZR+ move through customer qualification and production ramps.

The best pushback to the bull case is that the September guide is too wide to prove acceleration. Revenue guidance of $1.46 billion to $1.6 billion brackets a range that could look like either digestion or continued growth depending on where the quarter lands. EPS guidance of $0.93 to $1.13 also leaves room for a quarter that is only modestly above the Q4 non-GAAP EPS base of $1.00. That is why the thesis is not “the miss does not matter.” The thesis is that the miss matters less than the margin and mix signal, and that the market is likely to misprice Coherent if it demands a linear revenue ramp in optical AI. The company has already shown it can beat EPS despite missing revenue, and the next print must show that this was not a one-quarter mix artifact.

What to watch next is specific. The thesis is confirmed if September-quarter revenue lands in the upper half of the $1.46 billion to $1.6 billion guide while non-GAAP gross margin holds inside 37.5% to 39.5%, because that would show the divestiture and optical mix are offsetting the removed approximately $20 million of Aerospace and defense revenue. It is also confirmed if data center sequential growth reaccelerates from the cited 3% Q4 rate or if communications sustains growth near the 11% sequential Q4 pace as ZR/ZR+ contributes more revenue. The thesis breaks if revenue lands near the bottom of the guide and gross margin fails to reach 37.5%, because that would imply the EPS beat was not repeatable mix leverage. The next call also needs qa_evasiveness to stay closer to Q4 FY2025’s 9.2 than Q3 FY2025’s 53.5; if management becomes less direct while uncertainty remains elevated, the market will be right to demand a lower multiple on the optical ramp.