Intel foundry and the processor designing should have been split up in order to obtain legislative financial assistance with the push to support more manufacturing in the USA. This would force the foundry to actually focus on engineering manufacturer tooling and procedures to properly compete. Instead they are more reliant on the synergy with supporting Intel processor iterations. To be a true foundry and also improve national security, they need to be processor designer agnostic manufacturing company.
> To be a true foundry and also improve national security, they need to be processor designer agnostic manufacturing company.
This isn't obviously true now that Moore's Law is no longer in effect.
When you own the fab, your fab engineers and design engineers can help each other out. When you can gain a factor of 2 every 18 months, this is unimportant. If you can only gain 20% every 18 months, this is much more important.
Intel is about to divest themselves of their fab at a point when it is vital that they cooperate with it. Welcome to American management.
Having worked with Intel fab from inside Intel design and with TSMC at a different company, Intel fab would do very well to standardize and work on supporting their customers. If they decoupled and provided better processes, it would greatly improve the experience for those working with the fab. As it is right now, the Intel fab has always been the tail that wags the dog. They've carried Intel off and on for so many years that if design wants something and the fab pushes back, well design gets to pound sand and find some workaround to the fab's processes.
I just hope that Intel Foundries comes out better in the end than GlobalFoundries split from AMD. I do think the foundry staff at Intel have kept up on the tech side, and splitting them off will definitely allow them to take on more clients with fewer conflicts of interest.
It would be nice to see a result that isn't everybody simply trying to use TSMC for everything.
4 comments
[ 0.23 ms ] story [ 32.9 ms ] threadThis isn't obviously true now that Moore's Law is no longer in effect.
When you own the fab, your fab engineers and design engineers can help each other out. When you can gain a factor of 2 every 18 months, this is unimportant. If you can only gain 20% every 18 months, this is much more important.
Intel is about to divest themselves of their fab at a point when it is vital that they cooperate with it. Welcome to American management.
It would be nice to see a result that isn't everybody simply trying to use TSMC for everything.