Given IFS will be a standalone company, their metrics will be much easier to compare vs TSMC. With all this staffing benchmarking going around, we know IFS is extremely obese and bloated, given their manufacturing product cost structure is about ~60%-70% or more than TSMC. I expect there will be multiple rounds of cuts until they get within maybe ~20% more expensive than TSMC to be considered in the ballpark). Given cost is exponential figure (50% more expensive is 2x, 66% is 3x more expensive), the cuts will have to be deep... very very deep. How many HC will IFS cut next year (2025)? My guess, 20% to start (after this year's cuts). Your guess or thoughts?
11 replies (most recent on top)
@zoq I agree that for current production, having a full fab will reduce the product cost, as deprecation and fixed OH are spread over more units. However, Intel's cost problem has always been there even when the fabs were full. If you look back on financial statements, Intel product margins were prob in the high 60%... which sounds ok. However, if you look at TSMC's overall margins, they were even greater than Intel's product margins... and that's TSMC charging their customers product costs, R&D costs, and Marketing/Operations costs. If we just think about that... TSMC overall is making more margin % after including all those costs than Intel's product margin % alone. That's crazy.
@xxa While I agree that fab construction, and new equipment costs are high, which will create huge depreciation expenses. However, new construction or early technology costs do not really factor into current production product costs since those costs are held in another expense bucket (typically mfg R&D... they move into product cost once actual production units for sale are mfg/made). Don't be fooled, Pat has made several statements about how un-cost competitive TMG/IFS is vs TMSC... it starts with HC, and then moves on to how effective they use the equipment being purchased (are leaky gates due to mfg parameters designers have to follow, or is it poor silicon design that creates leaky gates?).
You need to have the fabs at full capacity to get the wafer cost down. This requires you have a good product customers want manufactured on a good process at the right price/wafer. Currently Intel getting 30% of its chips manufactured at TSMC says it all.
Oh man...naming specific teams..care to show more why you think this team is a joke?
Can start by getting rid of VSE.
Validation as a Service? Biggest joke
Correction 2024 -
Intel is on track to lose $10 billion (with a B) in IFS / TD in 2014. There is not enough headcount to get anywhere near that at IFS / TD. IFS cant simply get a note to cover the cost / bleed for the next 6 yrs (expected turn around point) Good luck Pat...
While demand was dropping, Intel was adding 30,000 heads ... Now revenue Is below the starting point and asp pressure is huge. So, back of napkin calculation says we gotta cut 30,000 - 45,000 heads. Still that won't make us as efficient as AMD TSMC but might just pass the red face test.
Most obvious is tmsc vs foundry headcount disparity. Ultimately wafer cost is the decider and is going to brutal in the near term.
Benefits will have to be reduced as part of the new cost structure.
The current financials, as they have already been virtually separated via IDM2.0, are easy enough to compare with TSMC. But having completely separate reporting with a separate business entity with, much better accounting and focused finance personnel should help.
That said, most of the losses that Intel Foundry are racking up right now are due to the exorbitant construction costs. Once that capital intensive spending abates, then so will the magnitude of the deficits we see on the foundry's finance sheet.
I'm sure there will need to be additional HC reductions based on volume and demand at the same time.