Semiwiki on LinkedIn Semiwiki on Facebook Semiwiki on Twitter Register / Log in 
Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/threads/former-intel-chief-pat-gelsinger-%E2%80%98i%E2%80%99ve-been-called-here-for-a-purpose%E2%80%99.24109/page-2
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/TimeZoneDebug] => 1000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000970
            [ThemeHouse/XPress] => 1010570
            [XF] => 2030770
            [XFI] => 1060170
        )

    [wordpress] => /var/www/html
)

Former Intel chief Pat Gelsinger: ‘I’ve been called here for a purpose’

I am just telling you they will not break even in 2027. Fab 52 isnt even fully loaded at the end of 2026. Fab62 is not tooled out in 2026.

Intel does not have plans to run anything in Ohio at this point.

yields may or may not be an issue. wafer cost on 18A and 14A is believed to be an issue
 
Got the CHIPS act grant money for Intel (Although brain dead Biden admin was prioritizing DEI & helping out foreign companies more than helping out their own😲😲😲😲

dkr1986 said:
Come on, this is not true at all.

Things that have changed for Intel Foundry since 2021,
- Successfully caught up to TSMC on node process tech or at least land in the same ballpark as TSMC compared to stuck on 10nm in 2021. 18A>=N3P with 3-4 months lag on actual products in shelf, 18A-P is withing ballpark of N2. 14A will be ~=A14. IMO, the best thing LBT can do for Intel Foundry is to accelerate 14A to be manufacturing ready before A14.
- Intel has now shells built in USA and ready for onboarding customers on advanced nodes in USA to service external customers. Strategically aligned with US government's initiative to onshore\re-shore semiconductor production back to USA.
- Successfully leveraged advanced packaging tech & supported Intel Product's tile based products during those years supporting HVM of many such products.
- Executed cost reductions all the while their top line was under pressure by low wafer volume due to Bob Swan's plan to use TSMC for PC products (Meteor Lake supporting tiles, Arrow Lake & Lunar Lake). [Outsourcing to TSMC was a good stop gap plan that helped Intel Products keep market share (especially Lunar Lake) - so it was a necessary step I don't blame BS for - but it is undeniable that led to pressure on Intel Foundry topline). All those wafer volume would have offset the fixed cost of a foundry business like in 2021 at least on a Gross margin basis.
- Successfully stood up an external foundry framework (node process, capacity etc) & established foundry ecosystem (still WIP). TSMC was not built in couple of years, IFS will not be built in couple of years, imo it will take at least 5+ years to stand up a foundry that is profitable just like any other fixed cost businesses that rely on economies of scale. At least Intel has the product division to satisfy a portion of that wafer volume demand from Intel Foundry before others sign up.
- Got the CHIPS act grant money for Intel (Although brain dead Biden admin was prioritizing DEI & helping out foreign companies more than helping out their own - I think they didn't want to be seen as being discriminatory to Taiwan since virtue signaling seems to be one of their goal- Most Asian countries would throw a foreign business under a Bus to favor their own companies if the roles were reversed).
- Signed up customers like Microsoft, Amazon and DoD on 18A, Mediatek on Intel 16. Colloborations with UMC & Tower (China made a major blow to Intel Foundry by not letting them acquire Tower Semi)

All this while their primary CPU business was being disrupted due to myriad of reasons. All this heavy lifting was done under a brain dead BoD and US admin. Intel Foundry is setup for success especially with Trump admin in power, they just need to execute the strategy forward. With LBT's push for efficient operation, I am confident they will become a reliable 2nd source Foundry by 2030.
Click to expand...
 
dkr1986 said:
IIRC, John Pitzer again reiterated that they expect to break even exiting 2027 at an investor conference recently on operating profit basis, only caveat being they will see expenses raise if they land a 14A customer (meaning they need to re-hire a bunch of people they fired to run the Ohio fabs leading to increase in OpEx again)

View attachment 3925

If you are referring to the 18A yield comments, he said yield is at what they expect it to be for a node on ramp up process. It will be high enough to drive margins end of 2026 and industry standard level margins in 2027. I don't see a smoking gun in that comment without knowing appropriate level of margin he is talking about here. Intel target operating profit for IFS was 30-40% back in 2023, may be he is referring to that for 18A. Even if we take a pessimistic pov, he thinks they can drive "appropriate level of margin" in 2027. Industry standard is probably referring to TSMC's OM of 50% recently.
View attachment 3926
And everyone conveniently forget about this comment about 18A yields in the same ER call. 18A yield is where they want them to be at this point of time.
View attachment 3927

I agree with you on this. That is why I said by 2030. Global Foundry had negative gross margin between 2018 to 2020 and as they have increased topline, economies of scale kicked in and now they are profitable by Gross Profit & Operating Profit basis. There is no reason to think Intel can't achieve that too. With USG onboard and LBT driving operating efficiency, slowly and surely, more customers will signup for IFS and they will become profitable by Gross profit basis first, then operating profit & free cash flow positive (these usually correlates) and then GAAP net income basis.
Click to expand...
The upcoming event:
www.intc.com

UBS Global Technology and AI Conference (Updated)

www.intc.com www.intc.com
 
MKWVentures said:
I am just telling you they will not break even in 2027. Fab 52 isnt even fully loaded at the end of 2026. Fab62 is not tooled out in 2026.

Intel does not have plans to run anything in Ohio at this point.

yields may or may not be an issue. wafer cost on 18A and 14A is believed to be an issue
Click to expand...
The text said they would break even 'exiting 2027' 'by run rate'. So even at the very end of 2027 they won't be operating profitably (say on a weekly or monthly basis)?

(Challenging lightly because I recall analysts saying Intel 10nm would never be profitable, and Intel 7 has carried the company for years now)
 
Xebec said:
The text said they would break even 'exiting 2027' 'by run rate'. So even at the very end of 2027 they won't be operating profitably (say on a weekly or monthly basis)?

(Challenging lightly because I recall analysts saying Intel 10nm would never be profitable, and Intel 7 has carried the company for years now)
Click to expand...
Intel 7 is still carrying the company but it's a bad node in terms of Intel as a whole it was not a cost effective node why do you think IFS has such a negative gross margin
 
siliconbruh999 said:
Intel 7 is still carrying the company but it's a bad node in terms of Intel as a whole it was not a cost effective node why do you think IFS has such a negative gross margin
Click to expand...
In 2022, when most of Intel's wafer volume was Intel 7, Intel Foundry had ~0% Gross Margin indicating foundry was selling Intel 7 wafers at cost to Products team. Only in 2023 and onwards foundry margin turned negative. I think that is due to early ramp of Intel 4/3 nodes (startup costs) and outsourcing to TSMC for MTL, ARL & LNL that pushed the margins negative. So it's a combination of those things.

May be I should have said the most important thing LBT could do for IFS is to accelerate 14A and make Intel products use IFS, keep outsourcing to minimum and use architectural improvements to compete with AMD+TSMC combo.

I am really disappointed by MJH's move to change PG's plan of keeping >80% of NVL on 18A/A-P. I think that is going to have negative effect on IFS in 2027 & 2028 (similar to 2023 & 2024 but less in magnitude - although 14A startup cost around that time may amplify that too ).
 
MKWVentures said:
Fab 52 isnt even fully loaded at the end of 2026. Fab62 is not tooled out in 2026.
Click to expand...
By my napkin math estimation is Fab 52 will be utilized 30-40% only in 2026 (assuming Fab 52 = 20kwspm). So, I agree with you on that (my assumption is 20Mu of Panther Lake, 400ku of CWF, 100ku of DMR being shipped in 2026 & no nova lake shipments - very rough math only). But also, my rough math shows, Fab 52 utilization should be around 70-80% in 2027 (60Mu of PTL, 600ku of CWF & 1.5Mu of DMR alone) I think another 30Mu of 18A based Nova lake (both notebook & desktop representing the SoC & iGPU tile) should be added to that. That should fill Fab 52 and spillover to Fab 62 by end of 2027 (Capex would be raised in 2027 to tool Fab 62 too).

Fab 62 will not be required until end of 2027 per my expectation unless external customer orders ramp up in early 2027. But that is okay, the Fab 62 shell costs about $8B? if we assume a straight line 30yr amortization schedule, impact on P&L is $270 million per year (that's peanut to make any impact on Intel's P&L for 2026).

Just my rough math, so margin of error will be very high (or I could be plainly wrong too...take it with boat load of salt).
 
MKWVentures said:
As IBM and Samsung and Global found out, the goal is not be just be a leader in process node... you have to be profitable. 202618A finances will give us a peek at the future of IFS.
Click to expand...

I agree but times have changed and US based manufacturing is a top priority. Intel 18A is the only leading edge node in production in the US today and TSMC N2 and Samsung 2nm are a couple of years away. That counts for a lot in my book.
 
dkr1986 said:
By my napkin math estimation is Fab 52 will be utilized 30-40% only in 2026 (assuming Fab 52 = 20kwspm). So, I agree with you on that (my assumption is 20Mu of Panther Lake, 400ku of CWF, 100ku of DMR being shipped in 2026 & no nova lake shipments - very rough math only). But also, my rough math shows, Fab 52 utilization should be around 70-80% in 2027 (60Mu of PTL, 600ku of CWF & 1.5Mu of DMR alone) I think another 30Mu of 18A based Nova lake (both notebook & desktop representing the SoC & iGPU tile) should be added to that. That should fill Fab 52 and spillover to Fab 62 by end of 2027 (Capex would be raised in 2027 to tool Fab 62 too).

Fab 62 will not be required until end of 2027 per my expectation unless external customer orders ramp up in early 2027. But that is okay, the Fab 62 shell costs about $8B? if we assume a straight line 30yr amortization schedule, impact on P&L is $270 million per year (that's peanut to make any impact on Intel's P&L for 2026).

Just my rough math, so margin of error will be very high (or I could be plainly wrong too...take it with boat load of salt).
Click to expand...
great numbers thanks!!
 
You must log in or register to reply here.
Back
Top