D

Coherent (Trans): RMB 20bn LT contract secured; GPM target raised to 45%

Compiled by Dolphin Research: $Coherent Corp.(COHR.US) FY27 Q1 earnings call Trans

I. Key takeaways from the print

1. Guidance (FY27 Q2)

a. Revenue: $2.55–2.75bn.

b. Margins and opex: Non-GAAP GPM of 40%–42%, non-GAAP opex of $430–450mn, non-GAAP tax rate of 18%–20%.

c. EPS: Non-GAAP diluted EPS of $2.28–2.48.

2. Mid/long-term revenue targets

a. New targets: First $3bn revenue quarter within FY27, with FY27 pro forma revenue growth >60%, more than 2x FY26 pro forma growth. First $4bn revenue quarter in FY28.

b. Base this quarter: Q1 revenue of $2.37bn (+56% YoY, pro forma). Non-GAAP EPS of $2.09 (+80% YoY), with earnings growth continuing to outpace revenue.

3. GPM target raised

a. New target: First quarter with GPM >45% by end-CY2028, 300bps above the long-term target (>42%) set 18 months ago for a 3–4 year horizon.

b. Drivers: Ongoing pricing optimization, scaling of new growth platforms, full utilization of 6-inch InP, and other cost-downs.

c. This quarter: Non-GAAP GPM at 40.6% (+37bps QoQ, +184bps YoY), driven by pricing, lower input costs, and yield gains. Opex ratio fell to 17.4% (vs. 18.4% prior), with OPM at 23.2%.

4. Capex and balance sheet

a. Capex: $776mn in Q1 (vs. $556mn in Q4 and $104mn YoY), set to rise QoQ in Q2. Capacity can be flexed across multiple product lines.

b. Balance sheet: Cash of $2.34bn (vs. $2.59bn prior), leverage at 1.1x.

II. Earnings call details

2.1 Management highlights

1. Data Center & Comms: Segment growth accelerated for the fourth straight quarter, with orders extending into CY28.

a. Segment revenue +21% QoQ and +79% YoY, accounting for 82% of total. Within that, data center +85% YoY, with another strong QoQ and YoY increase expected this quarter.

b. CY27 coverage is ample and more orders are rolling into CY28. Several new LTAs were signed in Q1, typically running through the end of the decade; management sees no signs of demand softening.

c. Optical modules: 1.6T ramping at multiple customers; 800G growth to continue through CY27. At ECOC, COHR showcased a 3.2T OSFP based on 400G differential EML; pluggables remain the mainstream for scale-out.

2. PhotonLink: An integrated photonics platform launched in Sep, spanning the full chain from light generation to detection and opto-electronic conversion.

a. The company pegs the 2030 market for CPO, NPO, and chip-to-chip optical interconnect at ~$30bn, with the broader Data Center & Comms TAM at ~$90bn.

b. Initial shipments this quarter included 400mW ultra-high-power CW lasers and fiber components for CPO. Scale-up products for CPO/NPO to begin shipping in 2H CY27.

3. InP capacity: 6-inch expansion ahead of plan, with most cost benefits still to come.

a. Internal InP capacity doubled YoY in Q1, one quarter ahead of plan; EML (100G/200G) output hit record highs.

b. 6-inch lines in TX and Sweden continue to show higher yields than 3-inch; the Zurich 6-inch line will ramp in 1H CY27. Further expansions are planned post-CY27.

4. Comms: Revenue +34% QoQ and +84% YoY; scale-across has become the largest revenue contributor in Comms.

a. Building on pump lasers, COHR offers modules, line cards, subsystems, and multi-rail solutions to boost existing fiber utilization and connect distributed AI data centers.

5. Industrial: Q1 pro forma revenue was flat QoQ and YoY, with strong orders; growth resumes this quarter.

a. Near-term drivers: A notable recovery in front-end semi equipment orders; CO2 lasers and ZnSe optics for AI server PCB drilling; PLD systems for superconducting tapes used in fusion.

b. Long-term opportunities: Thermodite material to enhance XPU performance, with samples sent to multiple strategic customers this quarter and revenue contribution starting 2H CY27; also investing in quantum technologies.

2.2 Q&A

Q: How should we think about the PhotonLink ramp, especially considering the $2bn LTA?

A: Small shipments this quarter, >$100mn in the Mar quarter, and >$250mn per quarter by end-CY27. Customer engagements span 10+ in CPO, 10+ in NPO, and 5+ in chip-to-chip; the $2bn LTA runs 5 years with revenue starting in early CY28.

PhotonLink is a one-stop integrated photonics platform covering the full optical signal chain. The first wave ships into CPO for scale-out networks; as CPO/NPO move into scale-up networks, adoption will be larger, with shipments expected before end-CY27. Over time this could scale to a multi-billion-dollar product line.

Q: What underpins a doubling of scale-across (DCI) revenue by end-CY27 in terms of customers and products?

A: Q1 revenue topped $200mn, ~4x YoY, and will double again by end-CY27 off that base. Growth has exceeded internal expectations, with an annualized Q1 run-rate of ~$1bn. Most of the YoY growth in Comms came from DCI, driven by AI workloads spanning multiple data centers and requiring higher inter-DC optical bandwidth.

Customers include both equipment vendors and cloud operators buying direct. The portfolio covers multiple value-chain layers: components like pump lasers, ZR/ZR+ modules, line cards, amplifiers, ROADM subsystems, and some turnkey solutions. The ramp so far excludes multi-rail; multi-rail starts to scale in the Mar quarter.

Q: Progress and outlook for OCS?

A: OCS is expanding from scale-out into broader use cases, with two sites ramping in parallel. OCS is booked under the Data Center business (+85% YoY in Q1), with deployments now extending from scale-out to scale-across and scale-up; 320x320 systems are shipping, with higher densities skewed to scale-up.

The technology uses liquid crystal and no moving parts, offering better power and reliability than MEMS. COHR estimates the 2030 OCS market at at least $4bn, potentially larger, with contributions to revenue growth increasing over the next few quarters.

Q: Thoughts on Lumentum's proposed industry consolidation?

A: It validates optics as central to data center architectures, and being an independent supplier is advantageous. Links for scale-up and even chip-to-chip will migrate from copper to optics. Management believes COHR has the broadest and deepest photonics portfolio, spanning light generation, fiber transport, and detection.

Many customers value an independent optics supplier, allowing COHR to serve both cloud and chip customers without conflicts. Management sees strengths in technology, manufacturing, and strategy, already reflected in the PhotonLink revenue ramp.

Q: Why provide longer-dated revenue targets now? Is supply secured?

A: CY27 is largely booked and customers are scheduling into CY28, supported by supply agreements. Management acknowledges the outlook horizon is longer than usual. Several new LTAs include high-quality, long-term forecasts, giving strong near- and long-term visibility. The $3bn and $4bn quarterly targets are backed by supply agreements, with a highly vertically integrated model relying heavily on internal capacity.

InP was the main bottleneck over the past 18 months; InP capacity doubled a quarter ahead of plan, supporting confidence in more than another doubling by end-CY27.

Q: Breakdown of customers and products in the $2bn LTA?

A: The LTA centers on PhotonLink, anchored by the 400mW CW laser. It also includes higher-level solutions. COHR did not name the customer, only noting it is a leading, well-known AI data center customer. Revenue is expected to start in early 2028 and ramp over several years; more PhotonLink LTAs are expected over the coming quarters.

Q: Does the 45% GPM target assume artificial constraints on the optical module business?

A: No constraint assumed; upside comes from new products scaling and InP cost leverage. The model assumes reasonable growth for optical modules; PhotonLink, new DCI products, and OCS are accretive to GPM. On costs, most of the GPM benefit from InP expansion is still ahead, a key pillar for the target.

Q: Can GPM sustainably exceed 45%?

A: 45% is a staging target, with no structural ceiling to further improvement. The prior '>42%' was a 3–4 year goal that was raised after 18 months; when approaching 45%, COHR will reassess the next target. Opex ratio should also reflect further operating leverage.

Q: Micro-VCSEL opportunities in scale-up/scale-in and ability to quantify next year's revenue?

A: VCSEL will be adopted relatively quickly for NPO; chip-to-chip is around 2029–30. No quantification for next year was provided. VCSEL will be used for scale-up NPO in the nearer term, while chip-to-chip (scale-in) lands around 2029–2030.

Multiple PhotonLink programs involve multi-core VCSEL paired with COHR's own multi-core fiber; over the past year the company shipped 700mn+ VCSELs into data centers. At ECOC, the PhotonLink launch showcased micro VCSELs, VCSEL arrays, and industry-leading 1 pJ/bit energy efficiency.

Q: Will capacity expansion by Chinese laser makers increase competition in ultra-high-power CW lasers?

A: No competition seen from Chinese vendors in 400mW ultra-high-power CW lasers. Management currently sees only one competitor in this space.

Q: For scale-up optics, does adoption start with CPO or NPO, and with VCSEL or InP?

A: Scale-up will start with CPO, followed by NPO, landing between late-CY27 and early-CY28. If NPO is not in place by late-CY27, it will be deployed in early-CY28. Early deployments use InP high-power CW lasers for CPO, moving from scale-out to scale-up; VCSEL adoption will come with NPO in early-CY28.

Q: Mix of internal vs. external InP sourcing and the timing of GPM impact?

A: The mix is similar to a year ago, shifting toward internal over the next few quarters. No specific ratio disclosed. Internal InP is primarily for data center optical modules, with module growth tracking InP capacity (lagging by about one quarter: lasers produced this quarter, modules shipped next), so the structure has not changed.

As capacity more than doubles again by end-CY27, the mix will tilt toward internal, though some external sourcing remains. With a major cost advantage on 6-inch internal lasers, 'most of the 6-inch InP benefit is ahead' refers to this mix shift.

Q: Of the $15k per chip full-solution content, how much can COHR capture near term?

A: $15k is the upper bound assuming full-stack adoption; actual share varies by customer. No near-term split was provided. The figure refers to a single chip requiring 100Tbps I/O using the full PhotonLink stack, including lasers, isolators, TEC, laser modules, fiber components, and connectors.

Over the next 1–2 years, some customers will adopt most of the stack while others take subsets, and COHR supports both approaches. Regardless of share, PhotonLink is driven mainly by scale-up, representing net-new demand from copper-to-optics migration.

Q: Rank the drivers for GPM >45%, and thoughts on China competition and price sensitivity?

A: Biggest contributions come from new product ramps and 6-inch InP scale leverage, then pricing, then cost-downs. The timing differs by driver. New products are GPM-accretive; shifting supply from external to internal opens significant headroom.

Pricing optimization is most evident in Data Center & Comms and has been a key contributor for several quarters, with room in Industrial as well; ongoing cost reductions are a constant. Management did not directly address China competition and price sensitivity.

Q: How do you align laser capacity with large customers? Any prepayments?

A: Nearly all LTAs include prepayments or NRE and other upfront payments to finance expansion. The only disclosed LTA is the one signed with NVIDIA in Mar. Upfront payments give customers skin in the game and help share capex risk, with strong customer appetite.

Demand planning is tightly coordinated with customers; InP demand is driven by both data center growth and the secular copper-to-optics shift independent of DC demand, as scale-up and chip-to-chip need optics to meet required bandwidth and link rates.

Q: Will 6-inch and 3-inch substrate supply constrain expansion?

A: Substrates to more than double capacity by end-CY27 are secured, with further expansions being locked in. COHR still buys 3-inch wafers and keeps 3-inch lines, but expansion is centered on 6-inch; there are 5+ substrate vendors, with several under LTAs.

Q: Can InP and GaAs capacity be repurposed to address a scenario where VCSEL leads?

A: Some interchangeability exists but switching is nontrivial; GaAs annual capacity exceeds 1bn VCSEL arrays. This was disclosed at the PhotonLink launch. If VCSEL adoption accelerates, management believes no one in the industry is better positioned to supply; InP capacity is also scaling fast.

Q: Risk of industry InP EML overcapacity in CY28?

A: No overcapacity risk seen before CY28. Demand comes from pluggable modules, integrated photonics, and some DCI use cases. Currently, all internal InP is used in-house and still insufficient for COHR's own needs; the company does not sell InP lasers externally, so only when external sales occur would outside supply rise, which is not a key concern. Internal sourcing will continue to increase.

<End of text>

Risk disclosure and statements:Dolphin Research disclaimer and general disclosures

Login to unlock13,063characters for free

This content is only available to signed-in users. Sign in to your Longbridge account to read the full post.