Could Intel Be Delaying the Foundry Competition to 14A

[siliconbruh999]

How much of Intel Foundry business will be internal at the break even point? 75%? And what wafer price will the Intel Design groups be paying? Certainly not cost like Samsung Foundry.

Imo this depends on 18A cost structure if they actually made a good node on PPAC.
PPA is fine from what you have heard C is something only Intel knows.

 

[count]

How much of Intel Foundry business will be internal at the break even point? 75%? And what wafer price will the Intel Design groups be paying? Certainly not cost like Samsung Foundry.

I think @MKWVentures probably has better estimates for this sort of thing.

My guess is for Intel Foundry to break even they are going to need a lot of external business, I think they will need 100% of Intel's internal business and that will probably only cover 50% of the financial needs of a leading edge foundry.

 

[MKWVentures]

How much of Intel Foundry business will be internal at the break even point? 75%? And what wafer price will the Intel Design groups be paying? Certainly not cost like Samsung Foundry.

This has already been determined (as of now).
1) DZ just said today that at break even point in 2027, Intel external business will be low to mid single digits (1-5B). Total foundry business is about 20B so DZ says 10%
2) Intel Product group purchases wafers at market price, not cost. that was all set in place last year. That needs to continue. This gives Product a very positive OM and foundry a very negative OM which is about correct.
3) cost is what was paid in the past ..even when I was there. foundry was zero margin and product paid the cost not matter what. This is what drove the push to outsource since intel costs were so high. The product group only agreed to insourcing if they paid market price.

So simplistically: Since Intel currently has contracts for N3 and N2 from TSMC, they know the market price as well as anyone. my model is ~20K for N3 and about 26K for N2 ... both dropping over time.

In 2027 when Intel say "we would have been break even but unexpected low volume, unexpected start up costs, and development of 14A impacted us" remember that it was all expected.

 

[dkr1986]

My guess is for Intel Foundry to break even they are going to need a lot of external business, I think they will need 100% of Intel's internal business and that will probably only cover 50% of the financial needs of a leading edge foundry.

Intel management has said many times they don't need a lot of external foundry customer revenue for foundry break even (If you believe them). They just need to push more EUV wafers through the fabs. CFO has many times said the blended ASP/cost increases 3x by moving from Intel 7 to Intel 18A. Just for context, Intel sells about 200+ Million units of PC CPUs and 15+ Million units of Server CPUs each year (in 2024 with all the share loss to AMD etc.).

I do understand that wafer prices going up for Intel Products (remember that 3x ASP for foundry) is going to have margin implication. But my understanding is AMD will have similar wafer pricing from TSMC for N2, so as long as Intel products team is able to release competitive products, they should be able to maintain margins.