No standard across the board and it being rather challenging to reliably package a Photonic Circuit on Chip into the rest of the required device have always been the key issues. I think the key here though is that we don'yet know the right approach so the former is very much trial and error. We are likely at this stage going to end up with a recommended standard from the company that can shout the loudest, rather than waiting to see which is obviously optimal aka best efficiency in transferring data.
TFA says that the motivation for the urgency of the standards are facts like:
> Earlier this month, Lumentum Holdings told Bloomberg that its optical components are “completely sold out” through to 2029, and it won’t be able to meet demand for some products between now and 2028.
With standardized interfaces, thus with interchangeable components from multiple suppliers, it is hoped that the shortages will be less severe.
As for memories, the shortages in optical networking components are caused by the AI datacenters and also by the fact that "China is a major producer of indium phosphide" and it might impose export restrictions as a reaction to the US "sanctions".
I read the article, but I find that the shortages may not be a bad thing as it will just end with a lot of waste otherwise due to many companies over developing a platform that is not viable long term. We are not at the stage where the demand matches the usability of these devices. It sounds a lot like Lumentum wish to influence how all it's competitors design their products for Lumentum's benefit, rather than focus on what maybe optimal.
Right now we need a lot less hype to maintain funding, and a lot more acceptance of we don't yet know what is optimal. We can still make significant progress towards a useful photonics computing platform that makes use of the differences in comparison with classic computing. That progress doesn't need to be at scale yet. Especially when that company has backing that doesn't require sales to maintain research and development.
No standard across the board and it being rather challenging to reliably package a Photonic Circuit on Chip into the rest of the required device have always been the key issues. I think the key here though is that we don'yet know the right approach so the former is very much trial and error. We are likely at this stage going to end up with a recommended standard from the company that can shout the loudest, rather than waiting to see which is obviously optimal aka best efficiency in transferring data.
TFA says that the motivation for the urgency of the standards are facts like:
> Earlier this month, Lumentum Holdings told Bloomberg that its optical components are “completely sold out” through to 2029, and it won’t be able to meet demand for some products between now and 2028.
With standardized interfaces, thus with interchangeable components from multiple suppliers, it is hoped that the shortages will be less severe.
As for memories, the shortages in optical networking components are caused by the AI datacenters and also by the fact that "China is a major producer of indium phosphide" and it might impose export restrictions as a reaction to the US "sanctions".
I read the article, but I find that the shortages may not be a bad thing as it will just end with a lot of waste otherwise due to many companies over developing a platform that is not viable long term. We are not at the stage where the demand matches the usability of these devices. It sounds a lot like Lumentum wish to influence how all it's competitors design their products for Lumentum's benefit, rather than focus on what maybe optimal.
Right now we need a lot less hype to maintain funding, and a lot more acceptance of we don't yet know what is optimal. We can still make significant progress towards a useful photonics computing platform that makes use of the differences in comparison with classic computing. That progress doesn't need to be at scale yet. Especially when that company has backing that doesn't require sales to maintain research and development.
Oh, Advanced Photonics Coalition, not American Power Conversion, the UPS people.
Most 3-letter acronyms have been used. If you're naming a new thing, check first. Use 4 or 5 letters if necessary to get uniqueness.
> the UPS people.
Oh, you mean Uninterruptible Power Supply, not United Parcel Service. Completely unforgivable to use an ambiguous acronym, death penalty for you.
I like this !!
or Agent Communication Protocol