Lumentum · Optical networking · AI infrastructure · Financial analysis · Scenario analysis

Financial Analysis of Lumentum, August 2026

1. Where the business stands

Lumentum makes the optical and photonic parts that move data inside and between data centres: semiconductor laser chips, laser sub-assemblies, wavelength management systems, optical modules, optical circuit switches and industrial lasers. Fiscal 2026 ended 27 June 2026. From the first quarter of that year the company retired its two reportable segments — Cloud & Networking and Industrial Tech — and now reports as a single enterprise, with prior periods recast. What remains is a disaggregation by product type.

The quarterly path is the part worth memorising, because the annual figure understates where the business exited.

Q1, Q2 and Q4 of fiscal 2025 as reported in the corresponding quarterly Form 8-K exhibits. Q3 fiscal 2025 of $425.2m is derived as the residual against the reported $1,645.0m full year. Non-GAAP operating margin expanded in every quarter of fiscal 2026 — by 650bp, 700bp and 440bp sequentially after the first.

2. Twelve months of 8-K filings

Thirteen Forms 8-K were filed between August 2025 and August 2026. Four are earnings furnishings under Item 2.02 and four concern officers and directors. The remaining five are financing events, and they are the story of the year.

3. Five-year financial record

Fiscal years end on the Saturday closest to 30 June. Sourced from the fiscal 2026 Form 10-K for FY2024–FY2026, the fiscal 2024 Form 10-K for FY2022–FY2024 and the fiscal 2023 Form 10-K for the FY2022 balance sheet. EBITDA is operating income plus depreciation plus amortisation of acquired intangibles. Free cash flow is cash from operations less purchases of property, plant and equipment. Net debt is total debt less cash and short-term investments; negative figures are net cash. Net income ex-extinguishment adds back the $7,756.6m non-cash charge and makes no other adjustment — it still contains the $236.3m valuation-allowance release.

The five years contain a complete cycle. Fiscal 2022 was the prior peak. Fiscal 2023 and 2024 were the inventory correction: revenue fell 23.1% in fiscal 2024, gross margin compressed to 18.5%, and operating margin reached −31.9% as fixed costs sat against collapsing volume. Fiscal 2025 was the turn, and fiscal 2026 the vertical part.

What makes fiscal 2026 unusual is how little of the recovery came from spending restraint reversing. Operating expenses in absolute dollars rose only 6.7% from the fiscal 2024 trough to fiscal 2026 — $685.5m to $731.1m — while revenue more than doubled. That is the whole operating-leverage story in two numbers.

Slightly more than half of the gross-profit increase came from price, mix and absorption rather than volume. And selling, general and administrative expense grew 4.3% against 83.2% revenue growth — that single line delivered 910bp of the 2,830bp operating-margin swing.

4. Ratio analysis

Balance-sheet denominators use two-year averages where a prior year is available. ROIC applies the company's stated 16.5% long-term non-GAAP tax rate to GAAP operating income over average invested capital, defined as total debt plus equity less cash and short-term investments. Net debt to EBITDA is not meaningful in fiscal 2024 on negative EBITDA and is negative in fiscal 2022 and 2026, both net-cash years.

Five things stand out.

• Profitability is back near the fiscal 2022 peak on margin and far above it on scale. Gross margin of 41.7% compares with 46.0% in fiscal 2022, but on 76% more revenue and a very different mix. The forward-relevant figures are the 46.0% non-GAAP full-year gross margin and the 50.4% fourth quarter.
• Returns are distorted in both directions. GAAP return on equity of −240.0% is meaningless. Ex-extinguishment return on equity of 28.4% overstates the run rate, because average equity of $2,889.3m straddles a pre-raise and post-raise balance sheet. ROIC of 13.8%, computed on operating income, is the cleanest read and understates the exit rate.
• Working capital tightened sharply. The cash conversion cycle fell from 129.0 days to 85.0, driven by days payable extending from 54.1 to 82.3 and days inventory falling 13. Payables rose 152% to $567.4m. Supplier terms are financing a meaningful part of the ramp.
• Inventory is the line to watch. It rose 47.1% to $691.6m while turns improved only to 3.0x from 2.7x. Set against $2,112.5m of purchase obligations due within twelve months, that is the fiscal 2024 write-down mechanism pre-loaded.
• Leverage is effectively gone. Debt to equity fell from 2.3x to 0.4x and interest coverage moved from −8.1x to 24.1x. Cash interest paid was $15.7m on $3.0bn of revenue.

5. Sector analysis

A company growing 83% with a 30% operating margin has to be read against its sector, because in optics both of those numbers are cyclical rather than structural. Four listed comparables, each at its own most recent completed fiscal year.

All figures from SEC XBRL. Coherent gross profit and operating income are derived — revenue less cost of goods sold, then less R&D, SG&A, restructuring and impairment — because Coherent presents neither subtotal; the derivation reconciles exactly to reported pre-tax income. Coherent return on equity uses net income attributable to Coherent over average total shareholders' equity, a period spanning the fiscal 2026 Series B preferred conversion. Fabrinet does not disclose R&D separately. Peer ratios use ending balances rather than the two-year averages applied to Lumentum elsewhere, so they are directionally comparable rather than identical in method.

Where Lumentum sits

• Fastest grower of the scaled names. 83.2% against Fabrinet's 35.7%, Coherent's 22.5% and Ciena's 18.8%. Only Applied Optoelectronics matched the rate, from a base one-seventh the size and with a −12.0% operating margin.
• Highest operating margin in the group by a wide gap. 17.4% GAAP and 29.8% non-GAAP against Coherent at 10.9%, Fabrinet at 10.0% and Ciena at 4.1%. The comparison is not like-for-like — Fabrinet is a contract manufacturer whose 12.0% gross margin is structural, and Ciena carries a systems-vendor cost base — but among the vertically integrated photonics names Lumentum is now the margin leader.
• The scale gap to Coherent has halved and is still real. Coherent's $7,118m is 2.4x Lumentum's revenue, down from 3.5x two years ago. Coherent is spending to defend it, with $1,102.9m of fiscal 2026 capital expenditure that drove its own free cash flow to −$1,023.4m.
• The sector is spending its cash flow on capacity. Combined capital expenditure across the five names was $2,126.6m against $1,719.3m of combined operating cash flow. Lumentum, Coherent and Applied Optoelectronics each spent more than 15% of revenue. Ciena, the one company not chasing AI-driven optical volume, is the only one with large positive free cash flow.
• Inventory is building everywhere. Fabrinet +75.8%, Coherent +79.5%, Applied Optoelectronics +107.8%, Lumentum +47.1%. The whole supply chain is positioned for continued growth.

What the sector implies for the forecast

Capacity is being added industry-wide, not only at Lumentum. $2.1bn of combined capital expenditure in a single year against a combined revenue base of $20.0bn. If AI optical demand grows into it, the supply constraint eases and pricing normalises — which caps the optimistic gross margin below what the 50.4% fourth quarter might suggest on its own. If demand pauses, five companies de-absorb simultaneously, which is what makes the downside a margin event rather than only a revenue event.

The margin premium is a mix and vertical-integration premium, and it is contestable. Coherent has the larger indium phosphide and laser-chip franchise and 2.4x the revenue over which to amortise engineering. Lumentum's R&D at 11.8% of revenue is the lowest point in its own five-year range, and in absolute dollars Coherent outspends it roughly two to one, $723.0m against $356.5m. Sustaining a 40%-plus operating margin against that is the central assumption of the optimistic cases below.

The differentiated products have no direct listed comparable. Optical circuit switches and co-packaged-optics laser sources are where Lumentum's disclosed backlog and orders sit, and neither Coherent nor Fabrinet reports a competing line. That is the structural reason the bull case can carry a higher margin than the sector.

6. Capital structure and the $7.8bn charge

Lumentum exchanged roughly 10.7m common shares for $1,124.9m of convertible principal plus the conversion value above it. Under US GAAP the fair value of shares delivered in excess of the carrying amount of the debt is a loss on extinguishment. Because the share price had risen far above conversion prices set between $69.54 and $187.77, that excess was $7.8bn.

Nothing about it is an operating or a cash event. The mirror entry is a $10,443.3m increase in additional paid-in capital. The economic substance is dilution — about 10.7m shares, of which only 1.4m were incremental to what the original conversion terms would have required — not value destruction. Cash interest paid during the year was $15.7m.

Non-GAAP diluted shares of 90.2m include about 13.6m from assumed conversion of the notes, net of 0.8m from the capped call. Shares outstanding at year end were 88.6m common plus 2.9m preferred, or 91.5m as-converted, against 69.8m a year earlier — a 31.1% increase in the share count. All four note series remained convertible at holder option entering the first quarter of fiscal 2027, principal must be settled in cash, and as of 14 August 2026 the company had received early conversion requests for $757.8m of principal.

7. Critical factors driving the bull case

Six factors, ordered by how much revenue each can add and how well each is evidenced in the filings rather than in commentary.

1. Optical circuit switches — the largest disclosed and least contested opportunity. highest conviction
   Backlog was stated as “well beyond $400 million” in the February 2026 8-K, against full-year Systems revenue of $1,008.4m. Management named optical circuit switches first among growth drivers in the fourth-quarter release and described “scaling rapidly to meet extraordinary customer demand.” Neither Coherent nor Fabrinet reports a competing line.
2. Co-packaged optics laser sources. high conviction, timing risk
   An “incremental multi-hundred-million-dollar order” deliverable in the first half of calendar 2027 was disclosed in February 2026. The fourth-quarter release added “increasing demand for ultra-high-power CPO lasers” and an initial order for external laser source modules. Co-packaged optics moves optics inside the rack, a step change in units per accelerator, and the ultra-high-power laser is the component few others can build.
3. 1.6T cloud transceiver modules.
   Called out as “advancing 1.6T adoption.” This is the highest-volume and most contested part of the opportunity — Coherent, Fabrinet's customers and the Chinese module vendors all compete here — and therefore the part most likely to carry sector-average rather than premium margin.
4. Vertical integration into the chip.
   Components grew 79.7% and reached 66.5% of revenue while contract-manufacturer purchase concentration fell from 25.1% to 18.2%. Lumentum is capturing more of the value in each module. This is the mechanism behind the 1,370bp of gross-margin expansion, and it has further to run.
5. Capacity already in the ground.
   Property and equipment rose 59.6% to $1,159.1m on $451.3m of capital expenditure, and inventory rose 47.1% to $691.6m. Unlike a backlog, this is money already spent. It converts to revenue if demand holds — and it is precisely the fixed cost that breaks the downside case if demand does not.
6. The NVIDIA relationship. strategic, not contractual revenue
   $2.0bn of equity at $695.31 per share and a stated optics technology partnership. The direct financial effect is funding. The indirect effect is roadmap alignment with the company that sets the accelerator interconnect standard. Neither is a disclosed revenue commitment.

8. Three-year scenario analysis

Modelled on a non-GAAP basis, consistent with how Lumentum guides. Non-GAAP excludes stock-based compensation and related payroll taxes of $191.3m, amortisation of acquired intangibles of $135.7m, restructuring, and the extinguishment charge. Operating expense is built in dollars rather than as a percentage of revenue, so the operating-leverage assumption is explicit and testable rather than buried in a ratio.

The base year is fiscal 2026: revenue $3,014.0m, non-GAAP gross margin 46.0%, non-GAAP operating expense $489.7m, non-GAAP operating income $897.0m, adjusted EBITDA $1,025.8m, depreciation $128.8m, capital expenditure $451.3m.

Bull 57% three-year revenue CAGR

Base 41% three-year revenue CAGR

Bear 9% three-year revenue CAGR

Core assumptions

9. Sensitivity analysis

Two grids, both for fiscal 2029, crossing revenue against the non-GAAP operating margin. The rows span the bear, base and bull revenue outcomes; the columns span their margins.

Fiscal 2029 non-GAAP operating income, $m

Fiscal 2029 free cash flow, $m

The free-cash-flow grid holds depreciation at $300m, EBITDA-to-cash conversion at 78% and capital expenditure at 10% of revenue in every cell, so it isolates the revenue and margin effect alone. Each scenario table above uses its own conversion and capital-expenditure assumptions instead.

What the grids show

• Revenue and margin matter about equally, which is unusual. Moving from $3,900m to $8,405m of revenue at a constant 34% margin adds $1,532m of operating income. Moving from a 22% to a 46% margin at a constant $8,405m adds $2,017m. In most hardware businesses volume dominates. Here the margin range itself is 24 points wide, and that width is the real finding: Lumentum's earnings are roughly as sensitive to absorption as to volume.
• Free cash flow is resilient at the bottom. Even the worst cell — $3,900m of revenue at a 22% operating margin — produces $513m of free cash flow, because capital expenditure falls with revenue. There is no cash-burn scenario at any plausible combination of these two variables. That is the single most important defensive fact here, and it is a direct consequence of the fiscal 2026 balance-sheet reset.
• The base case is not the middle of the grid. Base-case fiscal 2029 operating income of $3,446m sits in the upper-right quadrant and requires both $8.4bn of revenue and a 41% margin. Missing either one moves the outcome by roughly $500m to $900m.
• The downside is lost growth, not losses. Every cell except the bottom-left corner exceeds fiscal 2026's $897m of non-GAAP operating income. The bear case as modelled leaves fiscal 2029 operating income approximately flat against fiscal 2026 — three years of growth forgone rather than a return to losses.
• The asymmetry favours the upside. Cumulative three-year free cash flow is $5,954m in the bull case against $2,577m in the bear — a $3.4bn spread for a business that generated $300m in fiscal 2026.

10. What would change the view

Six observable things, each of which resolves before the fiscal 2029 outcome is knowable.

• Customer concentration in the next 10-K. The top two customers went from 31.4% to 41.6% of revenue in one year, and one holds 30.4% of receivables. If that climbs further, the revenue line becomes a derivative of two capital-expenditure budgets and the downside case gets more probable rather than less.
• Whether optical circuit switch backlog converts on schedule. The “well beyond $400 million” figure is the single most checkable forward disclosure the company has made. Watch whether it is restated, extended or quantified again.
• Inventory turns against purchase obligations. Inventory of $691.6m and $2,112.5m of purchase commitments due within twelve months, against turns of only 3.0x. A quarter in which inventory grows faster than revenue is the early signal of the fiscal 2024 pattern repeating.
• Cash settlement of the converts. $757.8m of early conversion requests were in hand at 14 August 2026, against $2,738.4m of cash and short-term investments. This is affordable but consumes a large share of the balance.
• Gross margin at the next volume step. The fourth quarter reached 50.4% non-GAAP. Whether that holds as 1.6T module volume — the most competitive part of the mix — grows faster than the proprietary lines is the cleanest test of whether the margin premium is structural.
• Sector capacity landing. $2.1bn of combined peer capital expenditure in one year. If pricing normalises across the group, the bull-case gross margin does not hold regardless of Lumentum's own execution.

Method and sources

All historical figures are as reported in SEC XBRL, unsegmented facts, retrieved through the sec-api.io API. No restatements were applied. Ratios are computed here rather than taken from the filings.

• Form 10-K for fiscal 2026, filed 17 August 2026, accession 0001628280-26-057358, period ended 27 June 2026 — financial statements, MD&A, concentration and segment disclosure, debt and contractual obligations.
• Form 10-K for fiscal 2024, accession 0001628280-24-038024 — fiscal 2022 to 2024 income statement and cash flow.
• Form 10-K for fiscal 2023, accession 0001628280-23-030374 — the fiscal 2022 balance sheet.
• Thirteen Forms 8-K filed 8 September 2025 to 11 August 2026, including the Exhibit 99.1 results releases for each quarter of fiscal 2026 and the 2 March 2026 Item 3.02 and 5.03 filing covering the NVIDIA preferred issue.
• Peer Forms 10-K: Coherent 0000820318-26-000020, Fabrinet 0001408710-26-000028, Ciena 0001628280-25-056698, Applied Optoelectronics 0001437749-26-005875, together with the two preceding years for each.

Three caveats worth stating plainly. First, revenue by geography is assigned to where product first ships, which for most cloud customers is a contract manufacturer rather than the end customer — Mexico revenue grew 199% and Thailand 115% in fiscal 2026, and neither is end demand by country. Second, fiscal 2025 third-quarter revenue of $425.2m is derived as the residual of the reported full year less the three quarters disclosed in 8-K exhibits; every other figure is as reported. Third, this analysis carries no valuation, no multiple and no price target, because doing so would require market data that cannot be verified against the filings themselves.

Scenario figures are estimates built on the stated assumptions. They are not forecasts, they are not company guidance beyond the first-quarter fiscal 2027 range published on 11 August 2026, and they are not investment advice. Prepared 27 August 2026, before any fiscal 2027 Form 10-Q was filed.