Marvell Maybe Buy Celestial Rumor Note
Finally circulated widely. Can write about it.
Irrational Analysis is heavily invested in the semiconductor industry.
Positions will change over time and are regularly updated.
Opinions are authors own and do not represent past, present, and/or future employers.
All content published on this newsletter is based on public information and independent research conducted since 2011.
This newsletter is not financial advice, and readers should always do their own research before investing in any security.
Feel free to contact me via email at: irrational_analysis@proton.me
Ok so The Information (I hate them… tabloid with very poor technical understanding) finally leaked Marvell/Celestial rumor.
Thank you for publishing this. Still think you suck.
Anyway, I heard this rumor from one source around a month ago.
Have a self-imposed rule of not writing about a rumor unless there are two good sources or three weak sources.
This is super interesting so lets go!!!!!!!
Marvell really needs this. Their current DR-MZI based CPO appeared to be vaporware at OCP. See old post.
Not a single shred of evidence all of the lanes turn on and exhibit acceptable post-FEC performance.
The timeline is kinda hilarious if you think about it.
Marvell CPO marketing lead has been (effectively) working at the same place for 13 years.
Dude shows up to OCP conference, presents not a shred of data while his direct competitor (Broadcom CPO) shows tons of performance, reliability, and power data.
<resigns from Marvell 1 month after conference>
Announces he joined Nvidia same day Marvell/Celestial rumor leaks widely.
Anyway, Celestial is easy to explain. They are the Germainium-EAM SiPho modulator CPO company.
Very very good technical intro into EAM (electroabsorption modulators:
https://people.engr.tamu.edu/spalermo/ecen689_oi/lecture10_ee689_eam_tx.pdf
Here is the very fast summary:
EAM is very popular in traditional EML style transceivers.
This means the laser and EAM are on the same monolithic Indium Phosphide substrate.
Celestial use EAMs that are Germainium based on Silicon Photonics.
All silicon photonics platforms have Germainium-based photodiodes.
Germanium photodiodes are a necessary evil.
You need photodiodes for optical-to-electrical conversion and for monitoring control signals.
Every Germanium photodiode is a point of failure.
Germanium photodiodes are sensitive to ESD damage.
Germanium photodiodes are a yield/reliability risk.
By using Germanium-based modulators, Celestial is exposed to similar risks that are intrinsic in SiPho systems.
More Germanium content is not a good thing.
More points of failure (defect risk, reliability risk) is not a good thing.
In general, Germanium EAMs are at risk of degrading over time (“burning out”).
The reason I am covering Germanium and reliability in such detail is this is the one outstanding issue.
If they have solved/mitigated this or have a plan to solve/mitigate this risk then the solution is quite compelling.
Solvable Problems // Minor Risks
There are a few other minor (nuisance) risks with their solution but its these all have clear lines of sight to “solved”.
The point of engineering is to solve problems. Celestial either has solved all four of the above minor problems or has a clear path to solving them.
To be clear, I am NOT COMMENTING on if Germanium EAM reliability is solvable or unsolvable.
All I am saying is this is the biggest risk. If they deal with this, solution is quite good and exciting.
Good move by Marvell.
Nice payday for Celestial shareholders/employees/investors.
Very exciting from engineering perspective.