Lumentum Holdings Inc. (LITE) — Coverage File

Data as of: Q4 FY2026 (quarter ended June 27, 2026); FY2026 10-K filed Aug 17, 2026; latest source Citi conference, Sep 9, 2026 File last updated: October 10, 2026 (initial build; Bull & Bear memo logged)

---

0. 30-second reopen

• Business: An indium phosphide laser and optical-systems maker riding the AI build-out: FY2026 revenue $3.0B (+83%), Q4 non-GAAP gross margin 50.4% and operating margin 36.6%, with two customers at 41.6% of revenue.
• Core debate: Is this a multi-year shortage of the lasers AI clusters need, or a classic optical-component upcycle priced at peak margins?
• Base thesis: Lumentum is one of very few volume makers of high-power, tight-spec InP lasers, sells them into three separate ramps (pumps between data centres, EMLs and CW lasers inside them, high-power lasers for optical scale-up) and adds OCS as a second franchise; customers are funding the capacity.
• Main bear case: New supply from Lumentum, Coherent and Chinese InP fabs meets a pause in hyperscaler builds; pricing and utilisation reverse on a much larger cost base, as in FY2024, while concentrated buyers and dilution take the rest.
• What to watch: Non-GAAP gross margin vs. revenue growth, the supply-demand language on EMLs and UHP lasers, OCS revenue and the early-2027 crossover at the largest customer, UHP laser revenue and NPO wins, Customer A's share, diluted share count.
• Valuation: ~$100B market cap at $1,103.36 (Oct 9, 2026 web close), ~28x management's $40 FY2028 EPS target and ~66x the annualised Q1 FY2027 EPS guide — priced for the plan being delivered.

---

1. Business snapshot

Lumentum designs and makes optical and photonic products, from laser chips to complete systems, in its own InP fabs (two in Japan, San Jose, the UK, and Greensboro from 2028) with back-end assembly in Thailand and at contract manufacturers [FY2026 10-K; Citi conf. Sep 2026]. Since the FY2026 reorganisation it reports one segment and two product categories [FY2026 10-K]:

• Components ($2,005.6M, +80%): EML, CW and ultra-high-power (UHP) laser chips for transceivers and co-packaged optics, narrow-linewidth and pump lasers and line subsystems for long-haul and data-centre interconnect, and VCSELs for 3D sensing. Laser chips and assemblies were 78% of Components growth.
• **Systems ($1,008.4M, +91$90M in FY2026), industrial lasers and cable access.

FY2026 revenue was $3,014.0M (+83%). Customer A took 26.6% and Customer B 15.0%; management calls Google its largest customer [FY2026 10-K; DB conf. Aug 2026]. 79.2% of revenue ships outside the U.S., including 9.4% to China [FY2026 10-K].

2. Core debate

The first question is whether the shortage is structural or cyclical. Management says lasers are "short everywhere" into 2027 and that scale-up optics will add another 3–4x of demand in 2028, so the gap will not close "within the next couple of years" [Citi conf. Sep 2026]. The 10-K, by contrast, still describes optical communications as a market that has "increasingly become subject to commoditization," and 54% of FY2026's gross-margin gain came from factory utilisation rather than structural pricing [FY2026 10-K].

The second is who captures the value of AI optics. Lumentum's next growth legs (OCS, CPO/NPO lasers, ELS modules) depend on one or two very large buyers that also design in-house, and one of them (NVIDIA) is now also an investor [8-K Mar 2, 2026].

3. Base thesis

AI clusters are moving from copper to optics, and every optical link starts with a laser. InP laser capacity takes years and customer qualifications to add, and Lumentum is among very few suppliers that can make high-power, tight-spec lasers at yield. It sells from the same fabs into three ramps — pumps and narrow-linewidth lasers for "scale-across" links between data centres (70–80% pump share), EMLs and CW lasers for scale-out transceivers, and UHP lasers for scale-up co-packaged and near-packaged optics — and adds OCS, where it is the only merchant supplier shipping at scale [Q4 FY2026 call]. Customers are paying for the capacity: NVIDIA bought $2.0B of preferred stock alongside a multi-billion-dollar commercial agreement, and the network equipment makers signed three-year, mostly take-or-pay pump-laser agreements [8-K Mar 2, 2026; OFC Mar 2026; Q4 FY2026 call].

Assumptions this rests on:

1. Laser demand stays ahead of industry InP capacity through calendar 2027, so pricing holds and laser chips keep outgrowing lower-margin systems.
2. Optical scale-up (CPO and NPO) reaches volume with the 2H 2027–2028 XPU generations and uses external high-power lasers, where only Lumentum and Coherent are credible today [Citi conf. Sep 2026].
3. OCS scales from under $100M a year to several hundred million a quarter, with Lumentum becoming the #1 supplier at its largest customer in early 2027 [Q4 FY2026 call].
4. Non-GAAP operating margin holds near 40% on the way to $2B a quarter, despite a growing share of transceivers, ELS modules and OCS [OFC Mar 2026; Q4 FY2026 call].

4. Main bear case

The bear case is that this is a cyclical at peak margins. FY2024 is the template: revenue fell 23% to $1,359.2M, GAAP gross margin fell to 18.5% from 46.0% two years earlier, and the company lost $546.5M [FY2024 10-K]. Today's cost base is far bigger: $2,354.4M of purchase obligations, $691.6M of inventory (up from $470.1M) and $451.3M of capex (up from $231.0M), with more fabs coming [FY2026 10-K]. Most customers buy without volume commitments, and contract liabilities are only $16.8M, so the long-term agreements described on calls do not yet show up as prepayments [FY2026 10-K].

How it breaks: Lumentum's own expansions, Coherent's UHP qualification (expected 2H 2027) and new Chinese InP fabs add supply into 2027–28 while hyperscalers digest their builds. Prices reset and utilisation falls just as depreciation peaks. Silicon photonics with CW lasers takes share from EMLs at 1.6T, CPO slips with new XPU silicon, and the largest OCS customer keeps more internal supply. Meanwhile the share count, already up about a third since FY2025, keeps rising as converts settle. The business could stay good and the per-share outcome still disappoint.

5. Business quality

Improving fast, but cyclical and concentrated. The moat is process know-how in InP lasers: management says its CW lasers carry a price premium because their narrow specification raises customers' transceiver yields, and the 400 mW UHP laser must meet power, noise, linewidth and mode-stability targets simultaneously at volume [Q4 FY2026 call; Citi conf. Sep 2026]. Pump lasers (70–80% share) and EMLs, where management sees no new merchant entrants from China, are the strongest positions [Q4 FY2026 call; Citi conf. Sep 2026].

The weaker businesses are systems: cloud transceivers grew on volume at lower ASPs and were "still a gross margin headwind" at OFC; Cloud Light's quality problems cost share before the 1.6T transition [FY2026 10-K; OFC Mar 2026; DB conf. Aug 2026]. Customer concentration (41.6% in two customers), capital intensity (capex 15% of revenue in FY2026, computed) and a supply chain with Chinese substrate and component exposure are the structural weak points [FY2026 10-K].

6. Key value drivers

Revenue = laser units × price/mix + systems. Laser units: EML output (>50% unit growth by the Dec 2026 quarter), pump lasers (a fourfold increase over the next several quarters), UHP lasers (~$50M a quarter by end-2026, first $100M-plus quarter in fiscal Q3 2027) [Q4 FY2026 call]. Price/mix: 200G EMLs at >25% of EML revenue, targeted at ≥50% of volume by mid-2027; CW lasers' margin gap to EMLs narrowed by a die shrink; LTA pricing "equal or higher" than escalated prices [Q4 FY2026 call; Citi conf. Sep 2026]. Systems: 1.6T transceivers ramping from fiscal Q1 2027, OCS (Q1 FY2027 first $100M-plus quarter; $8B TAM per management, with upside), ELS modules (about 2x laser revenue per unit at lower margin) [Q4 FY2026 call; OFC Mar 2026; Citi conf. Sep 2026]. Capacity: Japan, San Jose and UK expansions, then Greensboro from early 2028, which management says can enable "another $5 billion of annual revenue" [OFC Mar 2026; Q4 FY2026 call].

7. Financial profile

Sources: FY2022–FY2026 10-Ks (FY2022 from the file labelled FY2023_10-K_2022-08-24). Free cash flow was ~$300M in FY2026 and negative in FY2024–FY2025 (computed).

FY2026 quarters: revenue $533.8M, $665.5M, $808.4M and $1,006.3M; non-GAAP gross margin 39.4%, 42.5%, 47.9%, 50.4%; non-GAAP operating margin 18.7%, 25.2%, 32.2%, 36.6%; non-GAAP EPS $1.10, $1.67, $2.37, $3.23 (FY2026 sum $8.37, computed) [Q1–Q4 FY2026 calls]. Q1 FY2027 guidance: revenue $1.225–1.275B (>130% y/y), non-GAAP operating margin 39.5–40.5%, EPS $4.05–4.35 on ~102M shares [Q4 FY2026 call]. FY2026 GAAP net loss was $6,935.1M because exchanging $1.1B of convertibles for 10.6M shares booked a $7,756.6M non-cash loss [FY2026 10-K].

Balance sheet at June 27, 2026: cash and short-term investments $2,738.4M; convertible principal $1,554.3M (all currently convertible; $757.8M of conversion requests received by Aug 14, principal payable in cash); Japan term loans $92.8M; undrawn $400M revolver [FY2026 10-K]. Common shares were 89.7M on Aug 14, 2026, plus 2.9M NVIDIA preferred convertible one-for-one, against 69.0M weighted basic shares in FY2025 [FY2026 10-K].

8. Valuation

Management's own frame is a $1.25B quarter at ~35% non-GAAP operating margin (reached a quarter early in the Q1 FY2027 guide, at ~40%), then a $2B quarter at 40–42% operating margin, now expected to be raised by 100–200bp, and a first-ever FY2028 EPS target of $40, driven by OCS upside at the largest customer [OFC Mar 2026; Q4 FY2026 call; Citi conf. Sep 2026]. Greensboro capacity is excluded from those targets [OFC Mar 2026]. At the web price below, the shares trade at ~28x the FY2028 target, so the market is paying for the plan to be delivered, and the debate is whether earnings at that level are durable or a cycle peak. NVIDIA's March 2026 purchase at $695.31 per share is the only private-market reference in sources [8-K Mar 2, 2026].

Market data (outside sources/, web, dated) Share price $1,103.36 at the Oct 9, 2026 close; market cap ~$99.95B on 90.58M shares outstanding (stockanalysis.com, retrieved Oct 10, 2026). Derived by this file from filings: on ~102M diluted shares (Q1 FY2027 guide) equity value ~$112.5B; net cash ~$1.1B at June 27, 2026 (cash and investments less convertible principal and term loans), so EV ~$111B. Implied: ~37x FY2026 revenue, ~132x FY2026 non-GAAP EPS ($8.37), ~66x annualised Q1 FY2027 EPS guide midpoint ($4.20 × 4), ~28x the $40 FY2028 target, FY2026 FCF yield ~0.3%.

9. Management & capital allocation

Michael Hurlston has been CEO since February 2025; he previously ran Synaptics (2019–2025) and Finisar (2018–2019) [2025 proxy]. Wajid Ali is CFO and Wupen Yuen runs the global business units. Vincent Retort (EVP, Global Reliability & Quality) retires in October 2026 and stays as a consultant [8-K Jul 30, 2026]. A new Chief Accounting Officer was appointed in January 2026 and two board members with public-company CFO backgrounds joined in December 2024 and December 2025 [8-Ks Dec 12, 2024; Dec 15, 2025; Jan 5, 2026].

Pay: Hurlston's FY2025 total compensation was $27.7M, mostly new-hire equity replacing forfeited awards; $14M of his $25M initial equity is a four-year PSU that vests only if Lumentum's TSR beats the S&P 500 Information Technology index by at least 5% (50%) or 25% (100%) [2025 proxy]. FY2025 NEO PSUs vest 67% on FY2027 revenue and 33% on relative TSR; the FY2023–25 PSUs paid 24% of target [2025 proxy]. Insider transaction data is not in sources.

Capital allocation has swung with the cycle. Lumentum bought back $1,155.5M of stock in FY2020–FY2023 and spent $861.6M (NeoPhotonics, FY2023) and $700.9M (Cloud Light, FY2024) on acquisitions just before the downturn; no buybacks since [FY2022 and FY2024 10-Ks]. In the upswing it has funded capacity with outside capital: $1.265B of 2032 converts (Sep 2025), $2.0B of NVIDIA preferred (about half earmarked for strategic capex) and the $38.0M Greensboro fab purchase, while equitising older converts at a large premium to their conversion prices [FY2026 10-K; OFC Mar 2026]. The 2032 capped call stops at $268.24, so it no longer offsets dilution at today's price [FY2026 10-K].

Claims to verify over time:

• Laser supply will not catch up with demand "within the next couple of years" [Citi conf. Sep 2026].
• Chinese InP fabs will have no impact on Lumentum's numbers [Q4 FY2026 call].
• InP will regain share from silicon photonics at 3.2T [Citi conf. Sep 2026].
• The pump-laser and NVIDIA agreements underwrite the capex (terms not in sources).

10. Risks ranked by damage

1. Cycle turn: shortage ends into a bigger fixed-cost base Damage rank: High · Permanence: High · Fixability: Low Mechanism: Industry InP capacity (Lumentum, Coherent, China) catches up while hyperscaler capex pauses; prices reset and utilisation, the largest margin driver, falls as depreciation peaks. Evidence today: 54% of the FY2026 gross-margin gain came from utilisation; purchase obligations $2,354.4M; inventory up 47%; most customers have no volume commitments [FY2026 10-K]. What would confirm it: non-GAAP gross margin falling while revenue still grows, or management dropping the "supply-constrained" language on EMLs. Impact: FY2024 showed the downside: a 23% revenue fall took GAAP gross margin to 18.5%.

2. Customer concentration and in-house design Damage rank: High · Permanence: High · Fixability: Low Mechanism: Two customers (41.6% of revenue) can dual-source, design in-house or squeeze price; OCS and CPO, which carry the $40 target, depend on one or two buyers. Evidence today: Customer A rose to 26.6% from 15.4%; the largest OCS customer still has internal supply; large customers "often request and receive more favorable terms" [FY2026 10-K; Q4 FY2026 call]. What would confirm it: no OCS crossover in early 2027, or a second OCS supplier at the largest customer. Impact: a lost socket strands dedicated capacity for a product generation.

3. Technology shift away from the highest-margin chips Damage rank: Medium · Permanence: Medium · Fixability: Medium Mechanism: Silicon photonics with CW lasers displaces EMLs at 1.6T; thin-film lithium niobate challenges InP at 3.2T; CPO/NPO timing slips with new XPU silicon; Coherent becomes a second UHP source. Evidence today: CW lasers taking "a significant portion of transceiver output"; Coherent expected in 2H 2027 [Q4 FY2026 call; Citi conf. Sep 2026]. What would confirm it: 200G EMLs short of 50% of volume by mid-2027, or scale-up ramps slipping into 2029. Impact: cuts the mix gain behind 29% of FY2026's gross-margin improvement [FY2026 10-K].

4. Dilution and capital structure Damage rank: Medium · Permanence: High · Fixability: Low Mechanism: Converts, the NVIDIA preferred and equity pay grow the share count faster than intrinsic value per share. Evidence today: ~92.6M common-plus-preferred shares vs. 69.0M weighted basic in FY2025; $757.8M of conversions pending; RSU withholding-tax cash $281.0M and stock comp $170.2M in FY2026 [FY2026 10-K]. What would confirm it: diluted shares above ~102M or more equity-funded capacity deals. Impact: per-share value lags the business.

5. Supply chain and export controls Damage rank: Medium · Permanence: Low · Fixability: Medium Mechanism: InP substrate and critical-mineral supply depends on Chinese policy; past Huawei shipments are under BIS and DOJ subpoena; Greensboro must convert from GaAs to InP and requalify. Evidence today: China's curbs on exports to Japan hit the substrate chain; Lumentum added AXT supply and may need more [FY2026 10-K; Q4 FY2026 call]. What would confirm it: shipments capped by substrate shortages, a qualification delay, or a BIS penalty. Impact: missed revenue during the best-priced window and share lost to Coherent.

11. Metrics to monitor

12. Thesis tripwires

13. Open questions

1. What are the terms of the NVIDIA commercial agreement and the March 2026 OCS agreement (volumes, pricing, take-or-pay, duration)? Only the preferred stock terms are filed.
2. How much of FY2026's pricing gain is contractual LTA pricing versus spot and expedite fees that disappear when supply loosens?
3. What do UHP lasers, OCS and transceivers each earn at the gross-margin line? Revenue and margin by product line are not disclosed.
4. How will the $757.8M of pending conversions and the remaining converts be settled, and what is the full diluted share count after they are?
5. What is the source document for the $40 FY2028 EPS target, and what revenue and share count does it assume (see §9 claims)?

14. Update log

15. Source map