Earnings Call Highlights: Lumentum at JPM's 54th Global Technology Conference Fireside Chat — From Boom-Bust to Structural Multi-Year Growth, InP Is the Real Bottleneck

2 days ago

7 min read

Executive Takeaway

Boston, May 18. At one of the fireside chats of J.P. Morgan's 54th Annual Global Technology, Media and Communications Conference, Lumentum President and CEO Michael E. Hurlston took the stage himself. The hour-long conversation produced three signals more worth remembering than any single quarter's numbers.

First, the "cyclical industry" label on optical communications can come off. The boom-bust cycles once driven by telecom carriers have been completely displaced by hyperscalers. Hurlston put a number on it: hyperscalers are sitting on roughly a $2 trillion capex backlog. This is not a business cycle — it is hard demand.
Second, a three-stage growth engine is already lined up. 1.6T transceivers are first into volume production with factories running full; OCS + optical scale-out starts contributing meaningfully in Q3/Q4 this year; optical scale-up + CPO is set to take off in 2027, aiming to eliminate copper inside the rack outright.
Third, the real bottleneck is not technology or orders — it is InP wafers. NVIDIA has already locked up a very large share of InP capacity (including options), and other hyperscalers are lining up to sign long-term "prepayment + price guarantee" contracts. This is a playbook we last saw with HBM in the 2020s; now it is optical's turn.

This article is for technology and industry trend analysis only and does not constitute investment advice.

1. Why "This Time Is Different": From Telecom Cycles to Structural Hyperscaler Demand

The JPMorgan analyst opened with the industry's most sensitive question: optics has historically been cyclical — why should we believe this time is different?

Hurlston's answer was blunt. Past optical cycles were driven by telecom carriers — once giants like AT&T, Deutsche Telekom and Verizon slowed capex, the whole industry slid into a down cycle together. That was the boom-bust pattern of the past 30 years. Demand was set by "annual capital budgets," not by "the expansion curve of the business itself."

This time, the driver has changed. Hyperscalers hold roughly a $2 trillion capex backlog. That figure is not "what they want to spend this year" but "what they have committed to spend and haven't spent yet." In other words, the demand curve for the next several years is locked in by contracts, not dictated by the economy.

The difference is much bigger than it looks. Telecom capex decisions run on a 12-month annual cycle, whereas hyperscalers follow the GPU shipment curve — as long as NVIDIA, AMD and custom ASICs keep accelerating, demand for optical links only goes up. For an optical component supplier like Lumentum that sits on the hyperscalers' critical path, this means shifting from a "cycle bet" to a "capacity bet" — as long as capacity keeps up, revenue keeps up.

2. Progress Toward the North Star: $2B Revenue, 40% Operating Margin

During OFC, Lumentum laid out an 18–24 month North Star target — $2 billion in quarterly revenue and a 40% operating margin. At this conference Hurlston said they are "roughly halfway there."

He was clearest on gross margin. Pushing it from the low-30% range to nearly 50% came down to four things: product mix optimization, pruning low-margin products, manufacturing cost and yield improvements, and repricing the backlog (especially critical for an InP supplier in a seller's market).

Note the phrase "backlog repricing." In most industries, repricing means customer pushback and lost orders, but Lumentum is clearly in a position where the seller calls the shots — hyperscalers would rather accept price increases than lose supply. This alone is the first quantitative evidence STT sees of a structural shift in the optical industry.

3. Breaking Down the Three-Stage Growth Engine: Transceivers, Scale-out, Scale-up + CPO

1.6T Transceiver: already in volume production with factories at full load; revenue to double within 4–5 quarters. OCS + Optical Scale-out: meaningful contribution starting Q3/Q4 this year, focused on cross-rack and cross-cluster connectivity. Optical Scale-up + CPO: takes off in 2027, aiming to replace in-rack copper.
1.6T transceivers are Lumentum's strongest card today. At this generational node, Lumentum moved into volume production ahead of its Chinese peers — one of the most discussed and most contested races in the industry over the past few quarters. Hurlston expects 1.6T to double transceiver revenue within the next 4–5 quarters.
OCS (Optical Circuit Switching) + Optical Scale-out is the mid-stage engine. The WSS (Wavelength Selective Switch) + MEMS technology Lumentum has built up over more than a decade is exactly the key component hyperscalers need to move optical switching from the lab into production. Hurlston noted that customer engagement has risen sharply and new use cases keep emerging, including higher port counts, lower port counts, and routing around failed GPUs within a rack.
Scale-up + CPO is the biggest long-term chip. Hurlston's timeline puts the takeoff in 2027 — the key inflection where in-rack copper gets replaced by optics. CPO (Co-Packaged Optics) has long been seen by the industry as "the boy who cried wolf," but this three-stage timeline moves it from a distant promise to "clear orders over the next three years."

4. Two Months After OFC: Three Signals Beyond Expectations

Hurlston said that in the two months since OFC, customers have sent three signals that exceeded expectations — none of this was covered on the public earnings call, so it is relatively new information.

First, optical scale-out demand is far above original expectations. This is the hottest inflection right now, yet still under-discussed by the market — as AI training clusters expand from a single rack to multiple racks and multiple data centers, every link that used to run on cable has to switch to optics.
Second, 1.6T transceivers are being pulled in at full scale earlier than expected. Lumentum's factories are currently at full load, meaning Lumentum has won the first round of the market's debate over "when 1.6T hits volume."
Third, demand for CPO + scale-up has clearly stepped up. Hurlston's words were that "customer engagement has deepened significantly" — moving from "research-phase conversations" to "concrete order conversations." This maps to two applications, in-rack connectivity and cross-cluster connectivity, rather than the single scenario CPO pitches have typically focused on.

5. The Real Bottleneck: InP Supply Short by More Than 30%

The thing most worth remembering from the whole fireside chat is not 1.6T, not CPO — it is InP (Indium Phosphide).

Hurlston gave a hard number: the InP supply-demand gap already exceeds 30%. In other words, even stacking up all the capacity of the two major U.S. players, Lumentum and Coherent, can only meet part of the demand from customers other than NVIDIA. This is no longer a matter of "selling a bit more in a good market" — it is a structural supply shortage.

Lumentum's own InP expansion is aggressive:

Capacity is already up 8x versus 2023
The 5th InP fab is in production
Another 50% capacity to be added in the coming months
Evaluating a move from 4-inch to 6-inch wafers (though edge effects pose risks)

That already sounds like blistering expansion. But Hurlston followed up with a line that silenced the room: once scale-up ramps in 2027, InP demand will grow another 10x or more.

Do the math: this year's capacity is already 8x 2023, and 2027 needs another 10x on top — meaning total InP capacity demand balloons to 80x the 2023 level within four years. Expansion of this magnitude cannot possibly be absorbed by Lumentum and Coherent alone — it is a challenge for the entire industry.

Even more telling is NVIDIA's move. NVIDIA has already locked up a very large share of Lumentum's InP capacity — including options. For other hyperscalers, that means if they don't act now, they won't even get InP. Hurlston said plainly that Lumentum is "actively negotiating long-term contracts with prepayments and price guarantees with other hyperscalers," aiming to lock in more customers before next quarter to diversify risk.

This script is all too familiar — it is the HBM capacity-lock playbook of the 2020s, and now it is optical's turn.

6. Risks and Moats: Execution, Chinese Competition, and the Ceiling of Vertical Integration

Hurlston didn't just share good news; he was willing to confront the risks. STT organizes this into three dimensions, because they directly determine whether the multi-year growth story can actually play out.

The biggest risk is execution, not demand. Moving from deliveries in the "thousands of units" to the "hundreds of millions of units" is a huge challenge in yield, manufacturing and supply chain management alone. The demand curve is already drawn; keeping up is purely about execution. That is why Hurlston went into such detail on InP expansion — for him, capacity is the single biggest variable for the company over the next two years.
Chinese competition will struggle to catch up on core technology in the near term. Product lines like OCS and high-power lasers have 10–15 years of development history, and the reliability issues of WSS + MEMS took a long time to solve. Hurlston is very confident in the core-technology moat. But he acknowledged cloud transceivers are a different story — Lumentum's gross margin in this line is clearly below Coherent (mid-30%) and Chinese peers (mid-40%). Vertically integrating its own lasers and PICs (Photonic Integrated Circuits) into transceivers is the long-term fix, but it takes time.
Will hyperscalers build optical components themselves? Hurlston's answer was nuanced: they have "assembly capability," but core components — lasers, drivers, TIAs (Transimpedance Amplifiers) — still depend heavily on external suppliers. Lumentum's strategy is to stay at the component level and avoid competing directly with customers at the system level, so this risk is manageable. It's a smart position: sell the shovels, not the gold mine.

7. Optical Communications Officially Enters the "Capacity-Lock War" Era

From STT's perspective, the real signal from this fireside chat is not the gross margin improvement, not 1.6T progress, and not even the CPO timeline.

The real signal is in this sentence — "We are actively negotiating long-term contracts with prepayments and price guarantees with other hyperscalers".

It means the optical industry has officially entered the "capacity-lock war" era. In the past, this playbook only applied to severely supply-constrained links with very deep technology moats — HBM, leading-edge foundry, AI ASICs. Now, InP has joined the club.

The implications for the supply chain play out on three levels:

Upstream InP material suppliers and InP epitaxy vendors will be the most direct beneficiaries of this wave
Transceiver makers that can keep pace with the 1.6T volume ramp will pull away from the laggards
The OCS and CPO component supply chain will see order visibility improve sharply from 2H26 into 2027

Three metrics to track next quarter:

Progress on Lumentum's prepaid contracts with other hyperscalers (who's next after NVIDIA)
The 1.6T transceiver competitive landscape (the gap in volume-production timing among Lumentum, Coherent and Chinese vendors)
Progress on 6-inch InP wafer adoption (the key variable for whether expansion can accelerate further)

The story of optical communications leaving the cycle behind for structural multi-year growth has been told countless times over the past two years. This fireside chat didn't just repeat it — it moved the evidence from "demand expectations" to "capacity-lock contracts" — the hardest, most concrete signal yet.

This article is for technology and industry trend analysis only and does not constitute investment advice. Source: J.P. Morgan 54th Annual Global Technology, Media and Communications Conference Fireside Chat, May 18, 2026, Boston.