Papa Sylla

Anderson clarified no significant margin difference between EML and SiPho transceivers, with mix determined by customer application. Both require indium phosphide capacity. 200-gig VCSEL development progressing well.

Anderson confirmed healthy pricing dynamics driven by supply-demand imbalance, with 1.6T ASPs higher than 800 gig. Internal component sourcing provides buffer against external price increases.

Jim described broadening from spine/redundancy to DCI and scale-up network applications. TAM reassessment would be well above prior $2B estimate. Both customers and applications continue to expand.

Jim confirmed majority already internal, with internal share expected to grow over time as 6-inch ramps. Will continue external supply for customer and supply chain resilience reasons.

Jim explained capacity is fungible between EML and CW with allocation decisions made 4-6 months ahead based purely on customer demand mix. No significant profitability difference between EML and CW.

Jim confirmed first wave of 1.6T is silicon photonics and EML. VCSEL-based 1.6T using 200G technology expected to begin production mid-calendar 2026 with revenue in second half.

Jim confirmed DCI growing even faster than overall comms, driven by multi-datacenter workloads. Dual path to market: own modules plus component sales to other vendors.

Jim confirmed strong near-term demand for 400G and below as many customers still in transition, with market shifting to 800G and 1.6T beginning to layer on top.

Sherri noted cost reductions slightly larger contributor than pricing in Q3, emphasized yield improvement is ongoing with continued opportunity across all manufacturing processes.

Sherri confirmed still cautiously positive on traditional telecom with incremental improvement; DCI smaller but majority growth driver, telecom grew over 20% YoY.

Jim confirmed progress on 200G VCSELs, 200G differential EMLs, and CW lasers but declined to specify technology-to-generation mapping for competitive reasons.

Michael said OCS is tracking to $400M (not ahead) after early supply-chain issues, and expects Lumentum to become the #1 OCS supplier in early 2027.

CEO described ELS as a very significant opportunity and the primary currency for engaging non-primary CPO customers. Suggested wins are imminent but not yet ready to announce.

CEO assessed oversupply risk as low, citing active LTA extension discussions with customers and pricing flexibility as evidence. Noted customers would show more reticence if oversupply were expected.

Michael confirmed ~25-30% demand-supply imbalance persists. All EML capacity spoken for in LTAs through balance of calendar 2027. Wajid added LTAs have eliminated quarterly price-down negotiations, customers requesting more than LTA volumes enables incremental pricing discussions.

Michael confirmed CPO content per server roughly on par with transceivers given higher market share. New ELS module opportunity provides ~2-2.5x content gain versus standalone laser chips with decent margins. Wupen added optical scale-up is a brand-new TAM that further improves overall market position.

Michael confirmed targeted price increases in current guide with broader increases planned in 2026 LTAs. Wajid outlined margin expansion drivers: manufacturing utilization, 200G EML mix shift, plus 1.6T/OCS/CPO contributions in calendar 2026.

Michael confirmed supply up 10%+ in guide but demand-supply shortfall widened from ~20% to 25-30%. Wupen added newly announced AI projects are extending the horizon of the mismatch, enabling long-term agreements with leading customers.

Sanjay said total content is now multiple thousands of dollars per XPU with Scorpio X alone over $1,000, expecting further scaling from optical and higher signal-conditioning attach.

Management framed the Alchip collaboration as part of enabling an open, interoperable AI rack alongside GPU and ASIC partners.