Lumentum Lifts Fiscal 2028 Earnings Power Target to $40 as Optical Circuit Switching Order Book Surges Past Expectations
Deutsche Bank 20th Annual Technology Conference, August 27, 2026
Lumentum Holdings CEO Michael Hurlston used a fireside chat at Deutsche Bank's Technology Conference to disclose a materially higher earnings target for fiscal 2028, telling analyst Gianmarco Conti that the company now sees "$40 of earnings power" in that year, a number he explicitly framed as ahead of "all the sell-side analysts in terms of where we think our earnings power can be." The upgrade is driven primarily by optical circuit switching, where Hurlston said order volume from Lumentum's largest customer "has taken up their orders very significantly on us really since our last earnings call," producing what he called "a huge uptick in our order book and the order rate."
OCS Momentum Building Beyond a Single Customer
Lumentum has positioned itself as the only merchant supplier of optical circuit switching at scale, with shipments doubling sequentially and revenue from the category expected to cross $100 million next quarter. Hurlston pushed back on the market's perception that this business is a single-customer story tied to Google. "We started with customers other than Google," he said. "Google deploys a great number of OCSs in their topology, but we actually have customers that were much more substantial than Google in our early rounds of shipments." One emerging use case he highlighted: deploying an OCS per rack to route around GPUs or TPUs that fail or become overloaded, given that a stalled compute run representing $20 million to $40 million of compute time is an expensive failure mode for hyperscalers.
On total addressable market, Hurlston suggested the $8 billion OCS TAM figure the company gave at OFC last year already looks stale. "I think the TAM now looks like it's significantly undercalled," he said, adding that he expects a materially higher number to be presented at the next OFC conference in March 2027.
NPO Timeline Pulled Forward, Straining Fab Capacity Plans
Perhaps the most consequential operational disclosure was a pull-forward in the expected timing of non-co-packaged optics (NPO) adoption. Hurlston said that prior to Lumentum's last earnings call, the company expected the NPO inflection to hit in "late 2028 and early 2029." That has now moved up to "late 2027 into early 2028," as customers competing with Nvidia's co-packaged optics (CPO) roadmap have concluded they need to deploy substantially more optical content in racks, and sooner. Both the timing and magnitude of NPO demand have exceeded Lumentum's internal expectations, and Hurlston was direct about the capacity implications: "It's going to put a lot of pressure, a lot of pressure on that Greensboro fab. So we need to think now about our next leg of the stool, what can we do to add more capacity."
The Greensboro facility, acquired from Qorvo in early calendar 2026, will not reach full production until early 2028, an investment Hurlston said required conviction well ahead of visible order flow. Nvidia underwrote much of that risk through what Lumentum has characterized as a multibillion-dollar long-term agreement, alongside direct investment to equip the fab. Hurlston noted that if Nvidia wanted all of Greensboro's CPO-related capacity, "they could take all of that capacity and then we'd be in the business of looking for additional capacity," but the more pressing capacity call now stems from NPO customers, where Lumentum is trying to lock up similar long-term commitments to justify further investment.
Laser Economics Shift Across Node Transitions
Hurlston laid out granular detail on EML laser pricing and mix that investors have not previously had. Moving from 100-gig to 200-gig EMLs produces roughly a 2x uplift in average selling price, with 200-gig lasers now the dominant node as the market shifts from 800-gig to 1.6T transceivers. Competitive intensity at 200-gig remains low, with Hurlston naming Broadcom as essentially the only other meaningful supplier. The more important structural point is on silicon photonics displacement. At 800-gig, EML-based transceivers held roughly 70% to 80% share. At 1.6T, Lumentum expects that to fall to 40% to 50% as silicon photonics gains ground, even as absolute EML unit volumes rise. Critically, Hurlston said Lumentum's technical team expects silicon photonics to "run out of steam" again at 3.2T, producing what he called "a double whammy" of rising EML volumes and rising EML share simultaneously at that node.
Chinese Laser Competition Not Yet a Threat, But Watched Closely
On competitive risk from Chinese CW laser suppliers, Hurlston was notably unconcerned in the near term, disclosing that Lumentum is actually increasing its own CW laser output because it sees "a pretty big hole that we are being asked to fill," with no current impact from Chinese supply. He conceded that Chinese suppliers will likely close the gap over time, but primarily at the lower-power end of the market. "I think it's largely relegated to 70 and 100-milliwatt CW suppliers... I think it's going to take longer than people think," he said, distinguishing that segment from the high-power lasers used in CPO and NPO applications, where Lumentum's yield advantage remains the clearer moat.
Cloud Light Turnaround: From Quality Problems to $200 Million-Plus Quarterly Run Rate
Hurlston offered an unusually candid account of the struggles inside Lumentum's systems business, Cloud Light, which the company acquired at a roughly $75 million quarterly run rate before quality issues, including complaints from Google, dragged revenue down to $40 million to $50 million per quarter. After retooling engineering, the business recovered to roughly $100 million per quarter, but continued to lag competitors such as InnoLight, Coherent, and Eoptolink on time to market. That changed at the 1.6T transition, where Hurlston said Lumentum's signal integrity team allowed it to ship ahead of those same competitors on multiple SKUs, pushing the business to "well over $200 million a quarter." The unresolved problem is margin. "We've fixed the engineering problem. What we're looking at as we enter calendar '27 is how do we fix the manufacturing," Hurlston said, adding bluntly that he does not expect Cloud Light to match the gross margins of InnoLight or Eoptolink, only to improve materially from current levels.
China Policy Is a Double-Edged Sword
Asked about geopolitical risk, Hurlston acknowledged Lumentum is generally viewed as a net beneficiary of any U.S. move to restrict Chinese transceivers, but flagged the complexity given Chinese suppliers currently account for roughly 70% of transceivers sold into the U.S. market. "It would be a problem to cut that off entirely if things aren't done in a thoughtful way," he said, adding that any retaliatory restriction on Chinese-sourced substrates or raw materials, an input Lumentum has already had to secure through supplemental deals with a Japanese supplier and with AXT, "would also be a bit of a struggle." He pointed to Lumentum's manufacturing footprint in Thailand, China, and the U.K. as a partial hedge against escalation in either direction.
Heterogeneous Integration Viewed as Opportunity, Not Existential Threat
On the longer-term technology risk of heterogeneous integration, InP bonded onto silicon photonics or lasers grown directly on silicon, Hurlston framed both as enabling rather than disruptive to Lumentum's component business. "We're already working with the best silicon manufacturers to embed indium phosphide in their processes," he said, arguing that even in a fully bonded world, someone still has to grow and supply the underlying InP material. He acknowledged the shift could be more threatening to module-level competitors by pushing the industry toward commoditization, but said Lumentum's positioning as a components supplier first insulates it from that pressure.