Soitec's SOI wafers remain necessary no matter who builds optical engines if co-packaged optics standards take hold. Around 95% of these photonics wafers come from Soitec, so consolidation among module makers changes the buyer, not the need for wafers. Large buyers such as TSMC or Broadcom could pressure pricing and seek a second source, though none is apparent before 2030. That pricing risk is by far smaller than the module makers' architecture risk. x.com 

Co-packaged optics standards could make finished transceiver modules smaller and more interchangeable as chip designers build optical engines inside their packages. That is another reason to avoid Lumentum, Coherent and Applied Optoelectronics and prefer Soitec. Lumentum and Coherent now earn a lot from finished modules, leaving them exposed if the architecture changes. x.com 

Rising 1.6T switching demand favors optical suppliers, though their execution risks vary. Bank of America says the market could grow ~101% annually through 2030. Marvell's leading DSPs and Lumentum's supply-short 200G EML lasers are lowest risk; Applied Optoelectronics' capacity expansion and Fabrinet's customer concentration make them medium risk. Aeluma remains pre-commercial, while Lightwave Logic's polymer modulators are a high-risk moonshot despite its TSM partnership. x.com 

Lumentum is the only volume supplier of EML lasers for 1.6T optics, capacity is sold out, and indium phosphide fabs do not scale overnight. Qualcomm and Corning partnered with it on 1,060nm VCSEL chip-to-chip links targeting about 10 Tb/s. Revenue grew 109% and it trades at 30.6x forward earnings with $1.07B of net cash; a one-time $7.8B non-cash convert charge hid profitable operations. Risks are convert dilution and Coherent and Broadcom adding InP lasers before the shortage eases. x.com 

Broadcom's integrated optics, electronics and packaging could lower custom-chip system costs because separate suppliers defend their margins. At similar valuations, it is preferred to Nvidia and has more room to beat expectations. Google TPU v8 shipments are estimated at 4.0M in 2026 and 9.1M in 2027, assuming Broadcom supplies the remainder after MediaTek's allocation. Anthropic financing adds value only if chip profit exceeds likely support payments and capacity costs. x.com  x.com 

Optics and photonics should be among AI's biggest winners as faster GPU networks need optical connections that suppliers cannot expand quickly enough. Lumentum says some components are sold out through early 2029; Sivers is investing $30M to target more than 100M lasers a year, with operations expected in Q4 2027. Potential US restrictions on Chinese transceivers could reshape supply in 2027. Applied Optoelectronics and Sivers are the favorites. x.com  x.com 

Soitec wins from Google's TPUv8i regardless of which foundry builds the chips, because it supplies the photonics-SOI wafers upstream. TPUv8i's silicon-photonics optical circuit switch adds a new layer of large photonic dies, and its signal loss forces transceivers to go coherent, using about 2-2.5x more silicon per port and roughly 3-4x the wafer starts. Risks include coherent modulators moving to indium phosphide or thin-film lithium niobate, and Marvell building its own photonics. x.com 

Lumentum, the largest indium phosphide laser maker and supplier of the ultra-high-power laser in NVIDIA's Spectrum-X CPO switch, has a lot to gain from the silicon photonics roadmap. Its indium phosphide is sold out for years, which explains why its SEMICON Taiwan talk promoted gallium arsenide VCSELs for scale-up, rating them strong on 5 of 7 axes. VCSELs are its only incremental revenue in that period, even though they could displace its InP business. x.com  x.com 

Pluggable optics have a new scaling path alongside CPO and NPO, supporting demand for light from Lumentum, Coherent and Applied Optoelectronics. Nvidia and Cisco's planned USFP packs 16 full-duplex 400G/lane connections into an OSFP-sized footprint, targeting four times the bandwidth density of 200G/lane modules while retaining replaceability. Volume-production timing and adoption remain unknown. x.com  x.com 

Lumentum, Coherent and AXT benefit because every 800G/1.6T pluggable and CPO engine still needs an InP laser, which silicon photonics cannot generate. Lumentum's CEO says the InP gap is now worse than DRAM and NAND, with EML demand over 30% above supply even after output rose 50%+ this year, and Nvidia put $2B each into Lumentum and Coherent to reserve that layer. NPO is fully additive, getting scale-up lasers paid before CPO ramps, and Lumentum is first into that socket. x.com  x.com