Elevating a comment...

This was buried under an earlier post but merits more visibility...

The majority of semiconductor products being manufactured across all product lines in the industry don't need to be manufactured on the latest process technology. Intel's foundry business should be going after the non-leading edge customers if for nothing more than gaining experience in working with external fab customers. Intel has excess fab capacity with the factories it already has so why is it building more? Even if 18A exceeds TSMC at some point why would leading edge customers want to switch to Intel? TSMC won't be behind for long and they have an incredible amount of experience with working with external customers...that is all they do.

Intel needs to stop being obsessed with being the industry leader and focus on just being an industry survivor. Go after the lower end product manufacturing and build experience and work back to being an industry leader. Do some self reflection and figure out why businesses don't want to work with you. Make changes in how customers are treated. You really aren't the smartest people in the room and haven't been for quite a while.

Comment! It's anonymous!

11 replies (most recent on top)

IFS does offer Intel16 (1222) as a mature old node for customers and does have MediaTek as a customer for Intel16. HVM is in Ireland Fab24.

@ovn
Both Micron and TI are doing huge expansions right now (Boise & Utah), others are finding ways to make money at much less margin on US soil, Intel needs to adapt or die.

Operating US Foundries is a money losing proposition. End of story

Intel has been aggrogant over the years. Pat has finally admitted this and this is a significant step forward. Build products that customers want not what Intel thinks is good. The emerging memory effort is a prime example. Customers were not ready and willing to adapt their software architecture to make use of this. Intel kept spending billions of dollars on it, only to see it stop when Micron pulled out. Micron knew the customers did not want this new memory.

Intel has a minimum of one customer on a minimum of one legacy node

I’ve said too much.

Intel’s legacy nodes were built for Intel. That is why Intel is partnering with United Microelectronics to create foundry process nodes using legacy processes/equipment/facilities. However, it takes quite some time to convert what was an internal process to a competitive foundry process.

There prob are some good reasons not to focus on legacy nodes. Do they actually have a team in place to design products for them? Or is the team moved on to design products for the newer nodes?

Is it even price competitive to build up a team for such purposes when the ASP is going to be lower?

@dkj

As usual, the value is in the comments.

@OP argument doesn't make sense given Intel took the backseat for legacy nodes.

Not to mention the tooling just isn't there IE: a lot of design processes are closed source only for Intel since legacy nodes were not designed with contract manufacturing in mind.

Sorry @OP, swing and a miss. Try again next inning.

Pretty much that is where Intel is heading towards. By the time 18A becomes production ready, it won’t be the leading edge but good enough for second tier customers. The question is, how fast could intel earn their trust?

The ship has sailed for legacy nodes. There’s no incentive for customers to redesign for Intel.

Yes. If Pat actually understood how this business works, he would have prioritized intel 22nm or intel 3 over the ridiculously complex 18A. The issue is intel 3 won’t command as much revenue nor will it strengthen Pat’s public narrative of “leadership”.

Thread regarding Intel Corp. layoffs

Elevating a comment...

11 replies (most recent on top)

Post a reply