Skip to main content
PROFITVISIONLAB
Equity Deep Research

Lumentum (LITE): The Core Supplier of AI Optical Infrastructure — Worth Playing With Options?

LITE (Lumentum Holdings) holds 50-60% share of the 200G EML market and is a key supplier for 1.6T transceivers. A $2 billion strategic investment from NVIDIA, S&P 500 inclusion, and a 92% IV Rank — a full assessment of the options opportunity.

Stock Research #LITE

Ticker: LITE|Exchange: Nasdaq|Sector: AI Optical Interconnect / Photonics

📅 Published 2026/03/19 · ⏱ 20 min read · 📊 Q2 FY2026 · Includes industry map / earnings-call highlights / bull-bear debate

📋 Summary

Lumentum (LITE) holds a structural moat in the AI optical-infrastructure space — an in-house InP fab, a 50–60% share of the global 200G EML market, and it's currently the only vendor able to supply the key components for 1.6T transceivers at production scale. NVIDIA made a $2 billion strategic investment with multi-year purchase commitments attached, and LITE was added to the S&P 500 on March 23. All three of RS 99, A/D A, and IV Rank 92% clear the screen, but the technical picture is stretched, trading +33.6% above the 50MA — best to wait for a pullback and structural confirmation after 3/23 before entering via a Bull Put Spread.

⚡ Four-Layer Defensive Screen Quick View

Data as of Q2 FY2026 / IBD MarketSurge

FilterMetricDataResult
1: FlowRS Rating99✅ Go
A/D RatingA✅ Go
U/D Vol Ratio1.95✅ Buying nearly double selling
2: MoatROE29.27%✅ Go
SMR RatingA✅ Go
EPS Acceleration TrendV-shaped recovery underway✅ Conditional go
3: VolatilityIV Rank92%✅ Extremely rich
Next earnings2026-05-12⚠️ DTE must be controlled
4: TechnicalsPrice vs. 50MA+33.6% extended⚠️ Wait and watch
50MA$524.6Support reference
🎯 Overall Verdict: ⏳ Waiting for pullback confirmation
Three filters cleared; technicals are overextended. After the 3/23 S&P 500 inclusion, wait for a pullback and stabilization structure before initiating a Bull Put Spread.

INTRODUCTIONIntroduction: Plumbing and Wiring, Not a Star Stock

In the investment narrative around AI infrastructure, most investors' eyes are fixed on the flashy names — NVIDIA, Meta, Google. But deep inside the data center, a critical component is quietly becoming the bottleneck — optical interconnect.

As GPU clusters scale up, the bandwidth and power limits of traditional copper cabling are starting to show. The speed at which data moves between GPUs and servers is becoming the ceiling on the efficiency of the entire AI training process. Lumentum (LITE) is one of the core suppliers solving this bottleneck.

LITE is the "plumbing and wiring" manufacturer of AI infrastructure. The GPU is AI's brain; the optical module is AI's nervous system.— ProfitVision LAB, 2026/03/19

CHAPTER 01Industry Map: Why Must AI Use Fiber Optics?

1.1 The Physical Limit: the 2-Meter Copper Problem

This is a physics problem, not a business choice. The industry calls it the "2-Meter Limit" — at 1.6T speeds, copper cabling suffers severe signal degradation past 2 meters, and the heat it generates can't be absorbed by the data center. The 800G generation was already near this ceiling; 1.6T is the forced tipping point for conversion.

1.2 The Three Architectural Layers of Optical Interconnect

Architecture LayerTechnical TermDistanceLITE's Role
Scale-Out (between racks)Pluggable Transceiver1–100mCurrent core business
Scale-Across (between data centers)Optical Circuit Switch (OCS)100m–10kmEmerging high-growth engine
Scale-Up (within a node)Co-Packaged Optics (CPO)<1mEntering by end of 2027

LITE's strategy is to penetrate layer by layer along these three tiers, extending from its established Scale-Out position toward the higher-value OCS and CPO markets.

CHAPTER 02Anatomy of the Business Model: A Vertically Integrated Moat

2.1 The Vertically Integrated InP Fab — the Real Moat

Most optical-component companies use an "asset-light (fab-lite)" model, designing chips in-house but outsourcing manufacturing. LITE is completely different — it runs end-to-end vertical integration:

1
InP wafer fabrication: an owned indium phosphide fab in San Jose, with a 3–5 year build cycle and an extremely high barrier to entry.
2
EML chip design and fabrication: fully in-house from wafer to chip — competitors cannot quickly replicate this.
3
Module assembly: a Bangkok, Thailand plant, with a capacity upgrade completed in 2025.
4
System-level shipments: integrated transceiver, OCS, and CPO solutions.

2.2 Core Product Portfolio

ProductMarket PositionGrowth Driver
200G EML laser chip50–60% global market share, the only vendor shipping at production scaleMandatory adoption for the 1.6T generation
Pluggable transceiverScaling rapidly through the Cloud Light acquisitionQ2 systems revenue up 60% YoY
Optical Circuit Switch (OCS)Saves 40% of data-center power$400M backlog, customer count up from 1 to 3
Co-Packaged Optics (CPO)Already won hundreds of millions of dollars in ordersDelivery in 2027H1, a long-term opportunity

2.3 The Two-Segment Structure

SegmentShareGrowth Profile
Cloud & Networking88%The core breakout engine, up 65%+ YoY in Q2
Industrial Tech12%Cyclically weak, currently flat sequentially

CHAPTER 03Competitive Landscape: Who Are the Rivals, and Who Are the Allies?

3.1 Comparison of Major Competitors

CompanyTickerPositionvs. LITE
Coherent CorpCOHRThe leader in optical transceiver production volume, FY25 revenue of $5.8BLarger scale, but weaker in-house EML capability
FabrinetFNPrecision optics contract manufacturer (including for LITE)A manufacturing partner, not a direct competitor
Applied OptoelectronicsAAOILow-to-mid-tier transceivers, expanding capacity aggressivelyA large gap in technology tier
Mitsubishi / SumitomoJapanese manufacturersEML competitorsMore than 2 years behind on 200G production capability

3.2 The Biggest Technology-Path Competition: InP EML vs. Silicon Photonics

The InP path (where LITE sits): superior thermal stability and better signal integrity, making it the best technology for 1.6T. High manufacturing difficulty and a high barrier to entry give LITE an overwhelming advantage.

The silicon photonics path (pushed by Coherent / Broadcom): lower manufacturing cost, able to leverage existing semiconductor fabs. Production yields still lag InP today, but are expected to gradually mature after 2028–2029, at which point it will exert substitution pressure on InP.

LITE's countermeasure: it unveiled a 1060nm VCSEL solution at OFC 2026, offering a third technology path to attract customers looking to diversify risk.

3.3 The Dual Nature of the NVIDIA Relationship

NVIDIA is both LITE's most important strategic partner and a potential threat: it has also invested in Coherent, and at OFC 2026 reaffirmed that it will "keep using copper cabling." This is NVIDIA's usual multi-bet strategy — ensuring it's never locked into a single supplier. But the $2 billion NVIDIA investment comes attached with a multi-year purchase commitment and priority capacity-allocation agreement — this isn't a financial investment that can simply be dropped.

CHAPTER 04Financial Analysis: An Accelerating Recovery Off the Bottom

4.1 Quarterly EPS Acceleration Trend

QuarterEPS ($)EPS YoYRevenue (M)Revenue YoY
Jun-24$0.06-90% (the bottom)$308.3-17%
Sep-24$0.18-49%$336.9+6%
Dec-24$0.42+31%$402.2+10%
Mar-25$0.57+97%$425.2+16%
Jun-25$0.88+1,367%$480.7+56%
Sep-25$1.10+511%$533.8+58%
Dec-25$1.67+298%$665.5+65.5%

Jun-24 marked the bottom, and every quarter since has been an accelerating rebound. This kind of V-shaped recovery combined with continued acceleration is exactly the most explosive fundamental structure in the Minervini SEPA methodology.

4.2 Core Q2 FY26 Financial Data

MetricQ2 FY26Q1 FY26Q2 FY25YoY
Revenue$665.5M$533.8M$402.2M+65.5%
Gross margin42.5%39.4%32.3%+10.2pp
Operating margin25.2%18.7%7.9%+17.3pp
Non-GAAP EPS$1.67$1.10$0.42+298%
OCS backlogSurpassed $400M, customer count up from 1 to 3Beat expectations

4.3 Forward Estimates

Fiscal YearEPS EstimateYoYRevenue EstimateYoY
FY2026e$7.61+269%$2,894.5M+76%
FY2027e$14.52+91%$4,745.9M+64%

CHAPTER 05Highlights From the Last Three Earnings Calls

Q1 FY26 (Nov 2025, Sep-25 quarter)

Revenue of $533.8M (up 58% YoY), EPS of $1.10 beat by $0.07. The Thailand plant's manufacturing upgrade was completed, and management declared "the era of production volatility is over," entering a four-to-five-quarter sustained growth trajectory. Q2 guidance came in at a midpoint of $650M — two quarters ahead of the original target.

"Demand is stronger than we originally estimated."— COO Wupen Yuen, Q1 FY26 earnings call

Q2 FY26 (Feb 2026, Dec-25 quarter) — Latest Results

Revenue of $665.5M (up 65.5% YoY), EPS of $1.67 beat by $0.26. OCS backlog surpassed $400M, hitting the $10M quarterly-revenue milestone a quarter early. Won a "several-hundred-million-dollar" CPO order for delivery in 2027H1. Q3 guidance came in at a midpoint of $805M, up 85%+ YoY.

"Virtually every AI network is powered by Lumentum technology, whether through a direct hyperscaler partnership or as a critical component supplier. The vast majority of the growth is still ahead of us."— CEO Michael Hurlston, Q2 FY26 earnings call

Progress on the Three Growth Catalysts

CatalystCurrent StatusNotes
Cloud transceivers🟢 ExecutingQ2 transceiver systems revenue up $50M sequentially, continuing to accelerate into Q3
OCS🟢 Beating expectationsBacklog above $400M+, customer count expanded from 1 to 3, hit its target ahead of schedule
CPO🟡 Just getting startedSeveral hundred million dollars in orders already secured; the bulk of revenue arrives 2027H1

Q3 FY26 Outlook (to be reported 2026/05/12)

ItemGuidanceNotes
Revenue$780–$830MMidpoint of $805M, up 89%+ YoY
Non-GAAP EPS$2.15–$2.35+29–41% sequentially
Non-GAAP operating margin30.0%–31.0%Another substantial improvement

CHAPTER 06The Full Bull and Bear Case

The Bull Case

1
An EML monopoly position (2026–2027): 50–60% global market share in 200G EML, and the only vendor able to ship at production scale. 1.6T transceivers require this technology, and competitors can't catch up until 2027–2028 at the earliest.
2
Three waves of growth still haven't fully unfolded: transceivers have already contributed; the $400M OCS backlog has barely shipped; and large-scale CPO revenue doesn't start until 2027.
3
NVIDIA's $2 billion is supply-chain lock-in, not just money: a multi-year purchase commitment, priority capacity access, and support for building a domestic US plant. Order visibility extends beyond 2027.
4
Signed demand, not expectations: AWS, Google, Meta, and Microsoft's capex plans are all already publicly confirmed, and LITE's OCS backlog is written in black and white.

The Bear Case

1
NVIDIA's copper-cabling statement is a real threat: Jensen Huang reaffirmed at OFC 2026 that NVIDIA will keep using copper cabling, and NVLink's shift to optical is mainly slated for the Vera/Rubin generation after 2027 — the near-term timeline may be overestimated by the market.
2
A complex capital structure with double dilution pressure: NVIDIA's $2B convertible preferred stock (+2.88M shares) plus convertible bond conversion (+5.7M shares), for a combined potential dilution of over 8.5M shares. GAAP EPS remains under pressure.
3
Priced for perfection, with no margin for error: a forward P/E of 82x means any minor stumble in a single quarter — a supply-chain delay, ASP compression — could trigger a 20–30% multiple compression.
4
The long-term substitution threat from silicon photonics: Coherent is the designated silicon-photonics supplier for NVIDIA Spectrum-X. Once silicon photonics costs fall substantially around 2028–2029, it's expected to pressure InP EML as a substitute.

CHAPTER 07Five Tough Questions: A Stress Test

Q1|NVIDIA has invested in Coherent AND keeps using copper — what makes you so sure you won't be dropped in two years?
The copper NVIDIA uses is for NVLink (GPU-to-GPU Scale-Up connections); what uses fiber is Scale-Out connections (long-distance, between racks) — the two needs aren't mutually exclusive. More importantly, the $2 billion investment comes attached with a multi-year purchase commitment and priority capacity-allocation agreement, spelled out in black and white in an SEC 8-K. NVIDIA needs LITE to build a domestic US plant to hedge geopolitical risk — this level of entanglement isn't something you can just walk away from.
Residual risk: the contract is "non-exclusive," so the possibility of substitution remains once silicon photonics matures.
Q2|Your GAAP net income has been negative for a long time, and free cash flow is negative too — is the company actually making money?
The GAAP loss comes mainly from three non-cash items: amortization of convertible bond interest (about $40M/quarter), stock-based compensation (about $30–40M/quarter), and goodwill amortization. Free cash flow turned negative because CapEx surged (Q1 $76M, Q2 $84M), spent on expanding the InP fab and upgrading the Thailand plant — to fulfill orders that are already confirmed. The key metric to watch is that Non-GAAP operating margin has already reached 25.2%, with Q3 guidance calling for a further rise to 30–31%.
Residual risk: the interest burden on $324M of net debt will continue to erode GAAP EPS through FY2027.
Q3|FY2024 EPS was only $1.01, worse than 2020. Isn't this just a legacy optical-networking company stuck in a downturn?
The FY2022–FY2024 decline had clear causes: a trough in telecom-equipment capex spending combined with falling demand for Apple's 3D sensing (Face ID), together wiping out about $600M in annual revenue. But the company's structural position didn't collapse during that downturn — the core InP fab kept operating, and the Cloud Light acquisition closed in 2023, laying the foundation for the cloud business. The 2025–2026 breakout is "AI activating a technology moat that already existed and was simply waiting for its moment," not a flash-in-the-pan bubble.
Residual risk: if AI capex cools in 2027–2028, the valuation multiple could face severe compression.
Q4|Coherent is three times your size and also endorsed by NVIDIA — why wouldn't customers just choose it instead?
Scale and technology leadership are two different dimensions. Coherent is indeed larger, but in 200G EML chips — the single most critical technology node right now — Coherent doesn't fabricate its own InP chips and relies on third-party supply. NVIDIA invests in both because it uses different technology on different platforms: Coherent serves Spectrum-X's silicon-photonics approach, while LITE serves Blackwell/Rubin's InP EML approach. For a customer, the question is "which interconnect architecture does your AI cluster need," not "who's bigger."
Residual risk: once silicon photonics matures, Coherent's scale advantage will translate into a real competitive threat.
Q5|The stock went from $45 to $783 — a full 17x. How much of that is real fundamentals, and how much is sentiment-driven froth?
What's real: EPS grew a genuine 27x from $0.06 to $1.67; gross margin genuinely improved structurally from 32.3% to 42.5%; the $400M OCS backlog and NVIDIA's purchase commitments are both in black and white with the SEC. What's sentiment premium: the 17x move does include a "biggest AI beneficiary" narrative premium, and the mechanical buying from S&P 500 inclusion isn't fundamentals-driven either. Measured on a PEG of 0.82, the current price is still reasonable if the growth materializes — but there's no longer any margin for error. That's exactly why now isn't the time to chase the stock: it's not that the story is fake — it's that the story is already fully priced in.
Residual risk: if there's any gap in execution, the market will punish it extremely quickly. Waiting for a pullback and structural confirmation is the only way to protect yourself.

CHAPTER 08Options Strategy Design

Recommended Strategy: Bull Put Spread

Sell a put below a technical support level to collect premium, while buying a lower-strike put as protection. An IV Rank of 92% is a gift — the premium collected from selling options here is far richer than in normal conditions.

Entry conditions (all required):

A pullback appears after 3/23 A volume-light stabilization candle confirms it Short strike sits below support DTE expires before the 5/12 earnings report
ItemSettingNotes
StrategyBull Put Spread
Short put$570Below strong support, roughly 50MA + 9%
Long put$550Protective leg, $20 wide
ExpirationThird week of AprilEnsures expiry before earnings (5/12)
Target premium$0.80–$1.20Achievable in a 92% IV environment
Max loss$20 - net premium⚠️ Must be sized down to fit within the 5% Risk Unit limit

CHAPTER 09Action Framework: Watchlist

TimingAction
Now (3/19)Add to watchlist, take no action
3/23 (inclusion day)Watch price/volume reaction closely. Light-volume green close → uptrend continues; heavy-volume red close → wait for a pullback
3/24–3/28Watch for whether a pullback appears and confirm a stabilization signal
First week of AprilIf the structure confirms, check the IBKR options chain and design the Bull Put Spread
By end of AprilThe position must expire before earnings (5/12)

📋 Final Verdict: ⏳ Waiting for pullback confirmation

LITE is a rare, strong combination this year — top-tier RS, aggressive institutional accumulation, an accelerating EPS recovery, a strategic NVIDIA endorsement, and a 92% IV Rank. Within the long-term AI infrastructure narrative, optical interconnect is a genuinely existing structural need, not a speculative concept play.

Short term: after the 3/23 inclusion day, wait for pullback confirmation — don't chase the price.
Once confirmed: Bull Put Spread, short put at $570, expiring the third week of April.
Long term: the OCS + CPO growth wave doesn't fully unfold until 2027–2028.

"Wait for the structure, wait with discipline, wait for the right moment."
A true edge never comes from jumping in first.

📌 Tracking Record
DateEventVerdictFollow-Up Result
2026/03/19Initial publication, Q2 FY2026 + IBD data⏳ Waiting

▸ Next update: after observing price/volume on the 3/23 S&P 500 inclusion day
▸ Conditions for an earlier update:
  ① The stock meaningfully retests the 50MA and stabilizes ② A significant change in IV Rank ③ A new NVIDIA statement related to optics ④ The Q3 FY26 earnings report (5/12)

⚠️ Disclaimer|This article is part of ProfitVision LAB's Stock Research series #LITE, with data current as of 2026/03/19. Content is for educational and research purposes only and does not constitute investment advice. Options trading involves significant risk, and losses can exceed the initial investment. The author does not manage accounts on others' behalf, does not run tip-sharing groups, and makes no guarantee of profit.
LITELumentumAIOpticsEMLLaserOCSOptionsFourLayerDefensiveScreenStockResearch