Thanks for showing the InP tool build - been waiting for this. My main question is your yield improvement / number of InP chips required for EML vs CW, which is in the productivity improvement assumptions of MSD% each year I believe? I think tools demand is very sensitive to wafer size and whether it will be EML vs CW on the math I did, for example if assuming wafer size transition from 100 to 150 to 200mm can effectively decrease tool demand by 40-65% with each transition.
At this point, I think the opto revenues number for 2026 is pretty locked in but bigger question is can the tool demand in 2027-28 sustain strong growth rates if combination of wafer size transition + InP laser chip intensity changes. I think EML laser chip intensity is much higher than CW (which is the direction of travel as CPO uses CW)? Would welcome thoughts here.
interesting can you DM me about what you found? seems like you did good work. for CPO though I am more confident it takes much more epitaxy bc the actual cavity is much larger. EMLs are more complex but power is very low, even lower than standard SiPho CW. So a decrease in complexity vs a 10-2x increase in power you still need much more epi. unless you have a different understanding in which that case let me know.
This is a good work, man. I always appreciate the strong attempt to model. Most dont try unless they have to because it’s super easy to be wrong and hard to be right!
I’m being a little flippant here, but when Germans raise estimates things must be pretty exciting!
Great work! Look forward to the article on tool density for EML vs CPO
Thanks for showing the InP tool build - been waiting for this. My main question is your yield improvement / number of InP chips required for EML vs CW, which is in the productivity improvement assumptions of MSD% each year I believe? I think tools demand is very sensitive to wafer size and whether it will be EML vs CW on the math I did, for example if assuming wafer size transition from 100 to 150 to 200mm can effectively decrease tool demand by 40-65% with each transition.
At this point, I think the opto revenues number for 2026 is pretty locked in but bigger question is can the tool demand in 2027-28 sustain strong growth rates if combination of wafer size transition + InP laser chip intensity changes. I think EML laser chip intensity is much higher than CW (which is the direction of travel as CPO uses CW)? Would welcome thoughts here.
interesting can you DM me about what you found? seems like you did good work. for CPO though I am more confident it takes much more epitaxy bc the actual cavity is much larger. EMLs are more complex but power is very low, even lower than standard SiPho CW. So a decrease in complexity vs a 10-2x increase in power you still need much more epi. unless you have a different understanding in which that case let me know.
This is a good work, man. I always appreciate the strong attempt to model. Most dont try unless they have to because it’s super easy to be wrong and hard to be right!
I’m being a little flippant here, but when Germans raise estimates things must be pretty exciting!