Most important: Indium phosphide laser capacity and datacom transceiver mixIn the print: Networking segment revenue, gross margin, capex, Q4 guideOn the call: 1.6T transceiver ramp timing, InP wafer capacity additionsDownstream: NVDA, AVGO, ARST, META, MSFT, GOOGL
House view
Street prices Coherent as a broad photonics recovery play. We read this print primarily as an indium phosphide laser capacity signal, tracking whether datacom transceiver demand is outrunning Coherent's ability to supply 800G/1.6T optics to NVDA, AVGO, and hyperscaler networking builds.
What is priced in
Street expects $1.78B revenue and $1.39 EPS, reflecting continued datacom strength. Not priced in: the split between datacom transceiver revenue growth and industrial/telecom, which determines whether indium phosphide laser capacity is binding. Capex running at $0.15B last quarter suggests internal capacity expansion, but whether that spend targets laser epitaxy or back-end packaging is unresolved.
What to extract from this callRanked by constraint impact
Priority 1 · primary read
Indium phosphide laser capacity vs. datacom transceiver demand
Coherent is the largest merchant supplier of InP-based transceivers for 800G and 1.6T links. If datacom networking revenue accelerates faster than capacity additions, it confirms the InP laser bottleneck tightening for NVDA GPU cluster interconnects and AVGO switch deployments.
✓Networking/datacom segment revenue grows 30%+ YoY and management flags lead times extending or capacity sold out through calendar 2026, confirming InP tightness for NVDA and hyperscaler builds.
✗Datacom revenue growth decelerates to mid-teens or below, and management notes improving lead times or excess transceiver inventory at customers, easing the constraint for AVGO and ARST competitive positioning.
Priority 2
Capex trajectory as InP capacity expansion proxy
Capex jumped to $0.15B in Q2 FY26 from $0.11B a year ago. Whether this spend targets InP epitaxial reactor additions or other segments determines how fast the laser bottleneck loosens for downstream 1.6T transceiver supply to META, MSFT, GOOGL data centers.
✓Capex at or above $0.15B with management explicitly directing incremental spend toward InP laser and transceiver capacity, signaling the bottleneck persists into 2027.
✗Capex flat or declining below $0.13B with spend redirected toward industrial or silicon carbide lines, suggesting InP demand has plateaued and downstream supply eases.
Priority 3
Working the eventRelease drop vs. Q&A
In the release · first 60 seconds
Revenue
$1.78B consensus (Q3 FY26, ending Mar 2026)
Total revenue vs. $1.78B Street. More important: segment breakdown showing networking/datacom share of total. Above 45% of mix signals accelerating InP pull.
Non-GAAP EPS
$1.39 consensus vs. $1.29 prior quarter
Beat quality matters: was margin expansion driven by datacom mix enrichment (InP-heavy, higher ASP) or cost cuts in industrial segments?
Non-GAAP gross margin
Guided 38.5%-40.5% for Q3 FY26; Q2 was 39.0%
Gross margin above 40% with datacom mix rising confirms InP pricing power intact. Below 38.5% with flat datacom mix suggests competitive pressure loosening the constraint.
Capital expenditures
$0.15B in Q2 FY26; $0.11B year-ago
Capex at or above $0.15B with forward commentary on InP-specific capacity adds. Declining capex with no InP mention weakens the tightening thesis.
Q4 FY26 revenue guidance
Q3 guided $1.70B-$1.84B; Street likely models ~$1.82B+ for Q4
Guide midpoint above $1.85B with datacom called out as the driver confirms demand outpacing supply. Midpoint below $1.80B suggests demand moderation.
Downstream readsOutcome → what it means for names we care about
InP laser constraint tightens through 2027. NVDA GPU cluster networking stays transceiver-gated. META, MSFT, GOOGL face longer lead times for 800G/1.6T optics. AVGO switch deployments pace-limited by transceiver availability.
Current 800G demand strong but 1.6T transition slower than expected. NVDA Blackwell networking adequate near-term but Rubin-era bandwidth constrained. Mixed signal: near-term supply sufficient, longer-term InP tightness deferred.
Revenue in-line + datacom mix flat + gross margin at low end of guide
InP demand plateauing at 800G. Competitive supply from Lumentum and others catching up. Constraint eases for hyperscaler networking builds. NVDA and AVGO face less transceiver-driven gating into 2027.
Revenue misses + datacom revenue declines sequentially + lead times shorten
InP laser bottleneck breaks. Transceiver supply catches up to demand. NVDA networking no longer optics-gated. META, MSFT, GOOGL can accelerate cluster deployments. ARST and smaller transceiver vendors gain share.
39 signals · 1 high-qualityResearch read-through · not a trade recommendation
1.6T transceiver qualification and ramp timing
The 800G-to-1.6T transition doubles per-port laser content. Coherent's 1.6T ramp timeline directly sets the pace at which NVDA Blackwell/Rubin rack-scale networking can deploy at full bandwidth.
✓Management confirms 1.6T revenue contribution beginning in the June or September quarter with multiple hyperscaler qualifications complete, pulling InP demand forward.
✗1.6T qualification delays push volume to calendar 2027, giving competitors like Lumentum or II-VI time to add InP capacity and loosening Coherent's grip on the constraint.
On the call · where the read moves
InP laser and EML capacity utilization rates
If Coherent says InP fabs are running at or near full utilization, the transceiver bottleneck persists for NVDA networking builds and hyperscaler 800G/1.6T deployments through 2026.
1.6T transceiver qualification status and volume ramp quarter
Datacom transceiver lead times and backlog commentary
Extending lead times beyond 20 weeks confirm InP tightness that constrains META, MSFT, GOOGL data center network buildouts. Shortening lead times ease the bottleneck.
Customer concentration in networking segment
If top 2-3 hyperscalers are absorbing most InP transceiver output, smaller cloud builders face allocation risk, widening the constraint's downstream impact.
Silicon photonics vs. InP platform commentary
Any shift toward silicon photonics for 1.6T would reduce InP laser content per port, easing the indium phosphide constraint for NVDA and AVGO over time.