hist78
Well-known member
Moving money between left pocket and right pocket.
Intel Reports Third-Quarter 2025 Financial Results
- Thread starter Daniel Nenni
- Start date
Moving money between left pocket and right pocket.
siliconbruh999
Well-known member
Exactly
siliconbruh999
Well-known member
Sadly no
I understand this is just margin transfer. But with the financials reported separately now for Products & Foundry, over time both business units will develop a culture to maximize their own profits. This will push Intel Products team to push more competitive products out with ASP uplift (or cost savings), otherwise they will lose profitability. Right now I believe, they are sourcing Intel 7 wafers at 0% margin for Intel Foundry (CFO said so too) but Intel 7 is likely a costlier process to fab (CFO said that too), so Intel Products team is likely paying lot more for a Intel 7 wafer than say TSMC N7/N6 wafer (as evident by Intel Products team choosing N6 for some of the IO tiles in client products instead of Intel 7 - this could also be due to IP availability on N6 that reduces time to market). So it was easy for Intel Product team to show >60% GM in 2021 (now >50% with foundry margin negative) . When more Intel products move to Intel 3/18A, they will have margin pressure, they need to figure this out for sure. Initially it will be just margin transfer between the Products team & Foundry team but over time it should improve.
Based on some estimates I have seen, Intel's ASP is roughly 18% lower than AMD (client+DC CPUs). Intel is selling at low prices to keep market share form AMD & compensate for slightly low performance. Once that performance gap closes (it is closing every gen- desktop gaming & server is where the gap remains), this ASP difference will close, that is a tailwind for Intel Products teams Gross Margin.
Also Intel Product team already got used to wafer pricing from TSMC with 50% Gross Margin built into it for say 15-20% of their wafer needs (total outsourcing now is 30% per Intel) . So initially that 50% gross margin is going to move to Intel Foundry as they bring wafer back to Intel Foundry. So I would not be surprised both Product margins hold at same level & foundry margin improve quite a while before product margins get hit.
There are lot of moving parts, difficult to predict which way it goes. We just need to wait and see.
On last week's call:
"Capacity constraints, especially on Intel 10 and Intel 7, limited our ability to fully meet demand in Q3 for both data center and client products," said David Zinsner, chief financial officer of Intel.
I was a little surprised by this. Equipment shortages? Yield? High demand? All of the above?
"Capacity constraints, especially on Intel 10 and Intel 7, limited our ability to fully meet demand in Q3 for both data center and client products," said David Zinsner, chief financial officer of Intel.
I was a little surprised by this. Equipment shortages? Yield? High demand? All of the above?
siliconbruh999
Well-known member
I would doubt it's Yield Intel 10/7 is highly mature maybe it's due the fact that one of the Fabs is being changed to support UMC 12 in AZ and the demand being high due to windows 10 EOL so IT are upgrading .
Even though they are old, it is normal for processors from 2 to 3 years ago to be sold. AMD does the same thing.
Do you think Intel will do "full accounting / transparency" between Intel product & IFS ... as Pat said a couple of years ago?
I'd be surprised if they still would do that in the next couple of years.
has 14A PDK 0.5 already been out?I probably don't qualify as someone who knows what they are talking about, but I'll share my impression anyway.
I suspect that Intel's yield ramp is quite a bit slower than TSMC. And I believe that is a byproduct of their previous business model. When you are going to hit peak output in 3-4 years and shut the process down within 5-7 years you want to get the product out the door as soon as it is good enough to make money. That's what Intel has done for decades. And I think that issue is reflected in the 18A PDK where Intel's PDKs weren't as mature as TSMCs for a given milepost (0.5, 1.0). Intel was willing to accept the additional churn of process changes because being first to market was so important.
TSMC's business model (and Intel foundry going forward) requires them to have far process stability when they start running client silicon. So they have developed a different approach to get yield up and stabilize the process as quickly as possible. Intel is learning how to do that and 18A is the first real process they have to learn from. So I'm not surprised there are more than a few bumps in the road. 14A will be our first chance to see if Intel has learned those lessons and it isn't as far away as you might think. To the best of my knowledge PDK 05. is due out somewhere in the Dec'25 - Jan'26 time frame. If Intel has learned from 18A it should be locked down much more than 18A was.