I have extensively written up both as longs but yes this is very good for Aixtron. The more powerful the laser the more difficult the epitaxy. Very non-linear.
Going to translate all this. My take. Correct me if anyone spots mistakes. Lite has near monopoly in 1.6T EML and narrow line width. VCSEL will be out of scale out and scale up. That means Lite will have a huge share of 1.6T EML transceivers. Ramp in that is stronger than expected currently but the real story is in 2027/8. CW in Sipho CPO letter salad is the other replacement for VCSEL. Lite has competition from Cohr. Narrow line width is for scale across to connect datacenters. It was used by telecom in the past but now for hyperscalers. Cien, Nok, Csco are the customers for the lasers. That market just accelerated big time in the past quarter. That's my quick translation. Oh, one more, IA is basing this on current tech. If Cohr etc also ramp in 200gb/lane or someone else comes up with narrow linewidth tech, this may change. But if it does not, the 2027/8 1.6T ramp is huge and so is the scale across opportunity depending on what the politics say about centralized hugge DCs. Estimates don't take this into account. Maybe I will write something more thorough about this later or maybe someone else will do the work instead. I much rather read.
2. Just to circle back to Intel: IMHO the big moment of truth will be when (as in by which date) Clearwater Forest Xeons are widely available for purchase. If Intel can deliver here - compute dies in 18A, all on Intel 3 base dies, that would indeed be a game changer. Not least because those base dies (in an EUV node!) are "smart substrates", and include Intel's answer to X3D Cache in big, plus I/O, plus....
So, for now, I wait. And I am also holding on to my INTC shares, although my holdings are small compared to yours.
I have no idea what any of this means and have no intention of doing my own research.
But i like star wars and IA likes $LITE so I am full aping my bank account in Lumentum. YOLO lol
shut the fuck up
Wonder if it’s worth playing the InP angle here through Aixtron for tooling. Depends on how high demand drives CapEx for tooling.
I have extensively written up both as longs but yes this is very good for Aixtron. The more powerful the laser the more difficult the epitaxy. Very non-linear.
Just. Awesome.
shut the fuck up scammer
This dude below is impersonating the real Irrational Analysis. I reported him to substack already
I spend 7m reading this and 30m thinking about it.
You should really charge for this stack.
Going to translate all this. My take. Correct me if anyone spots mistakes. Lite has near monopoly in 1.6T EML and narrow line width. VCSEL will be out of scale out and scale up. That means Lite will have a huge share of 1.6T EML transceivers. Ramp in that is stronger than expected currently but the real story is in 2027/8. CW in Sipho CPO letter salad is the other replacement for VCSEL. Lite has competition from Cohr. Narrow line width is for scale across to connect datacenters. It was used by telecom in the past but now for hyperscalers. Cien, Nok, Csco are the customers for the lasers. That market just accelerated big time in the past quarter. That's my quick translation. Oh, one more, IA is basing this on current tech. If Cohr etc also ramp in 200gb/lane or someone else comes up with narrow linewidth tech, this may change. But if it does not, the 2027/8 1.6T ramp is huge and so is the scale across opportunity depending on what the politics say about centralized hugge DCs. Estimates don't take this into account. Maybe I will write something more thorough about this later or maybe someone else will do the work instead. I much rather read.
1. Another great post!
2. Just to circle back to Intel: IMHO the big moment of truth will be when (as in by which date) Clearwater Forest Xeons are widely available for purchase. If Intel can deliver here - compute dies in 18A, all on Intel 3 base dies, that would indeed be a game changer. Not least because those base dies (in an EUV node!) are "smart substrates", and include Intel's answer to X3D Cache in big, plus I/O, plus....
So, for now, I wait. And I am also holding on to my INTC shares, although my holdings are small compared to yours.
scam
Dislike the margin rate below 50%
I dislike when margins are above 50% lol. Low margins means earnings explode
What about cohr? How far are they behind
scam
Isn’t Aeluma intending to disrupt this market? They have contracts with NASA and DARPA and it will take time but they could be a future challenger?
If the datacenter bubble pops, how will this affect $LITE?
Vapourized 💨
scam
Everyone is short LITE now on valuation + Coherent ramping up their Indium Phosphide plants. Seems scary to get long right now, but love the company.
scam
jeez they make cables big deal
scam
Please do share your latest thoughts on Cerebras. We’d all like to see an NVIDIA competitor.
scam
I don't like to average down, but I'd be willing to on this one. Otherwise I'll just hold my bag
scam