Coherent Corp

Phase A — Understand the business (carried, one update)

Segments

• Datacenter & Communications is now 79% of Q4 revenue, up from 67% a year earlier. Q4 grew +58.6% YoY. Segment profit for the year was $1,330M, up 47%, a 25.2% segment margin [primary: 10-KForm 10-KA company’s audited annual report to the US regulator. The most complete thing it publishes. MD&A].
• Industrial shrank by design. The −10% comes from two sales: aerospace & defense (closed 2025-09-02) and the Munich business (2026-01-30). Segment profit still rose 4% to $423M [primary: 10-K MD&A]. June expected Industrial to drag on margins. Instead it is being made smaller and more profitable at the same time.
• Geography, by customer headquarters [primary: 10-K segment note]: North America $4.63B (65%, +30%), Europe $0.81B (11%), China $0.81B (11%, +20%), Japan $0.39B (5%), rest of world $0.47B (7%). June left geography as "not sourced". China is about one dollar in nine of revenue. It is also a manufacturing base (the 10-K lists China, Malaysia, Vietnam and others). So export-control risk hits both what Coherent sells and where it makes things.
• Customer concentration doubled. "One major customer accounted for 20%, 10% and 10% of consolidated revenue during fiscal 2026, 2025 and 2024" [primary: 10-K Note, Major Customers]. A second customer also passed 10% in FY26. The 10-K names neither. A reader will guess Nvidia. That is plausible given the March supply agreement, but it is an inference, not a disclosure, and our figures leaves the name blank.

Phase B — Measure performance

Earnings result — Q4 and full-year FY26 (reported 2026-08-12)

Graded against the May guidance. Revenue came in at the top of the $1.91–2.05B range. Non-GAAP gross margin was 40.2% against 39–41%. Non-GAAP EPS was $1.74, above the $1.52–1.72 range. It was a clean beat on every line.

Graded against the June dossier's FY26 base. June estimated ~$7.0B revenue and ~$5.50 non-GAAP EPS. The actuals were $7.12B and $5.61, so the base case landed slightly ahead.

Q1 FY27 outlook [primary: Ex-99.1 2026-08-12]: revenue $2.2–2.4B (midpoint +45% YoY), non-GAAP gross margin 39.5–41.5%, non-GAAP opex $400–420M, non-GAAP EPS $1.85–2.05. On the call, management said it expects its first $3B+ revenue quarter by the end of FY27 (June 2027). It named indium phosphide (InP) wafer supply as the only bottleneck, said assembly and test capacity is not constrained, and expects internal InP output to double year on year by the end of the September quarter. The $3B figure is an aspiration from the call, not formal guidance, and our figures records it at medium confidence.

The GAAP-to-non-GAAP gap narrowed as June expected. It was $4.05 per share in FY25 (−$0.52 vs $3.53) and $1.49 in FY26 ($4.12 vs $5.61). Share-based compensation was $193.8M and acquired-intangible amortization $280.3M [primary: Table 6]. Two one-offs flattered FY26 GAAP: a $124.1M gain on the sale of businesses and $140.1M of other income.

Comps & valuation

Price: ~$291 on 2026-09-24. That is −24% since the June dossier ($382.81), even though the fundamentals beat. It is +59% year to date. Lumentum (+151% YTD) and Applied Optoelectronics (+191% YTD) left Coherent last in its own peer group.

June's ~49x forward multiple has compressed to about 37x on the guided run-rate. The Q1 FY27 guide alone already annualises to June's FY27 base EPS of $7.80, before any of the ramp toward a $3B quarter. On earnings, the stock is cheaper than in June and the numbers are running ahead of June's base. Peer multiples are still n/a in this pass. June's comps gap stays open.

Phase C — Judge people & books

Management — a second mega-grant

On 2026-08-27 the board approved a one-off, 100% performance-based PSU grant with a combined target value of $100M. The CEO gets $50M. The CFO, CTO and chief strategy/legal officer get $15M each, and the supply-chain chief $5M [primary: 8-KForm 8-KA filing for something that happened between reports and matters enough to tell shareholders now. 2026-08-31].

• The hurdles. The stock must hold a 60-day average of $454.43 for 50% of target to vest, $542.86 for 100%, $643.61 for 150% and $757.77 for 200%. Those equal 10%, 15%, 20% and 25% annual growth over four years. Each tranche also requires Coherent's total return to beat the median of the S&P 1500 electronic-equipment index. Nothing vests before 2030, and there is a further one-year holding period.
• The implied starting price is ~$310. At today's ~$291, the first tranche needs the stock up ~56%.
• Read. The design is shareholder-friendly as these grants go: no time-vesting component, a relative-return gate, a five-year lockup, and full forfeiture on voluntary exit. But this is Anderson's second outsized package in two years. The first was the ~$101M hiring grant that made him America's highest-paid CEO for 2024. June's governance flag gets stronger. On a change in control, the hurdles are measured at the deal price with linear interpolation, and earned units then vest on a double trigger. A takeover at a premium would pay out.

Forensic — the cash-conversion break (new, primary)

June could not run this lens on filings. It flagged "inventory building ahead of revenue" as the place a demand stumble would show first. The 10-K shows that build is now the dominant fact in the accounts.

[primary: 10-QForm 10-QThe quarterly version of the annual report. Lighter, and not audited. balance sheets; quarterly cash flow derived from year-to-date totals (10-Qs, 10-K)]

• Operating cash flow was $79.5M on $805M of net income, down from $633.6M the year before. The 10-K's own explanation: "a significant increase in inventory levels to support higher revenue growth" [primary: 10-K MD&A]. Inventory rose +80% while revenue rose +22.5%. Inventory days went from ~139 to ~212.
• Suppliers carried part of the build. Accounts payable rose from $847M to $1,905M (+$1.06B). Without that, operating cash flow would have been about −$1B. A payables increase that size tends to reverse. If it does, it becomes a cash outflow.
• Capex went up 2.5x, to $1.10B, and half of it landed in Q4 alone ($556M). That is the InP and Sherman, Texas build-out. The 10-K notes a $50M CHIPS Act preliminary memorandum of terms for Sherman.
• Purchase commitments: $3.4B due in FY27 and $8.4B thereafter, $11.8B in total [primary: 10-K Note 13]. Against FY26 cost of goods of $4.45B, that is 2.6x a year's COGS already committed. The 10-K says it does not expect "a significant amount of liquidated damages" on these commitments. This is the same shape as the purchase-obligation risk in the Arista refresh: suppliers are locked in for years ahead of demand.
• Dilution. Average basic shares rose from 155.5M (Q4 FY25) to 195.7M (Q4 FY26), +26%. Two things did it. The Bain-backed Series B preferred converted in fiscal Q2 (it had been $2.48B of mezzanine equity). Nvidia bought 7,788,161 shares at $256.80 on 2026-03-02 [primary: 10-K equity note]. The conversion also removed a 5% payment-in-kind dividend ($129.9M in FY25, $35.1M in FY26), so the common shareholders gained that back.
• Nvidia's six-month lock-up ended around 2026-09-02 [inference: six months from the March closing, per the 10-K's lock-up terms]. Nvidia holds ~4.0%, below the 10% line that would require it to report sales on Form 4Form 4The filing an insider must make within two business days of buying or selling their own company’s shares.. Any selling would first show up in its quarterly 13F.
• Legal and regulatory. 10-K Item 3 lists only "various claims and lawsuits incidental to its business", none material [primary]. The SEC EFTS re-fetch found 0 litigation releases and 0 AAERs [primary: regulatory-findings.md, 2026-09-25]. June's "verification incomplete" is now verified clear.
• Goodwill and intangibles fell to $7.26B from $7.68B through amortization and divestitures. That is 40% of total assets, down from 51% as the balance sheet grew. The impairment fault line June named is still there but matters less.

Forensic verdict. Nothing here suggests the accounts are wrong. They show a company converting earnings into inventory, fabs and supplier commitments at a scale it has never run before, and funding the gap with equity. This is ordinary for a supply-constrained ramp. If demand pauses, the balance sheet will show it first: inventory, payables and $11.8B of commitments.

Phase D — Project & stress-test

Forward projection

FY27 (July 2026 – June 2027).

• Base. Q1 at guide ($2.3B, $1.95 EPS), then ~+9% a quarter toward ~$2.9B in Q4. That gives about $10.4B revenue and ~$8.80 non-GAAP EPS.
• Bull. The $3B quarter arrives on time and 6-inch InP yields lift gross margin above 42%: ~$11B / ~$10.30. That is close to June's bull EPS ($10.50), a year early.
• Bear. InP output stalls, or a hyperscaler digests orders after Q1. Revenue flat at the Q1 run-rate, ~$9.2B / ~$7.60, with an inventory write-down on top.

June's FY27 EPS paths ($5.50–6.00 bear / $7.80 base / $10.50 bull) all move up. June's base is now this dossier's bear.

Brier forecast: NOT logged (unattended run). Candidate: "COHR reports a quarter with revenue ≥ $3.0B within FY27 (by the 10-Q/10-K for the quarter ending 2027-06-30), p≈0.45." The 0.45 reflects that management has named InP as the single constraint and is doubling it. They have hit the top of the revenue range for two quarters running, but a $3B quarter needs roughly +9% sequential growth every quarter.

Cash, the number June did not project. On base-case FY27 earnings of $1.8B, free cash flow stays negative if inventory keeps growing faster than revenue and capex stays at the Q4 annualised rate ($2.2B). The $2.0B short-term investment and cash cushion ($1.99B) covers roughly one more year like FY26 without new financing. Watch Q1 FY27 operating cash flow (10-Q due early November) as the single most informative line.

Bull vs bear (updated)

Bull, updated. June's thesis held on every operating measure. Revenue and EPS beat, Datacenter & Communications grew 40%, the Industrial clean-up worked, and GAAP profit arrived. Management says InP is the only bottleneck and is doubling it. The stock trades at ~37x a run-rate that already equals June's FY27 base, and it is last in its peer group. If the cash conversion normalises once the fabs are built, the multiple has room.

Bear, updated. The June bear was about valuation. The new bear is about the balance sheet. Coherent is running a working-capital and capex build that consumed $1B of cash in a record year. It has signed $11.8B of supplier commitments. One customer is now 20% of revenue, and that customer is probably also a 4% shareholder whose lock-up just ended. The stock's 24% fall despite the beat suggests the market has started pricing this [inference: no source attributes the fall to cash flow].

Skeptic (what changed)

• "The earnings aren't cash yet." $805M of net income, $80M of operating cash. Until one quarter shows operating cash flow above net income, EPS is an accrual.
• "The inventory is a bet on the order book." Inventory days of ~212 in a business that says it is supply-constrained is a strange combination. Some of it is InP work-in-progress on long cycle times, which is ordinary. The 10-K does not split raw materials, work-in-progress and finished goods in the text read here, so this is an open item.
• "The customer holds the leash, and now the shares." A 20% customer concentration plus a funding relationship with the same probable counterparty means Coherent's cash plan depends on one company's roadmap. Nvidia also funded Lumentum ($2B), carried from June.