siliconbruh999
Well-known member
Marketing people I guess
Panther lake delay?
- Thread starter Y.H
- Start date
Meteor Lake was actually doing pretty well for Intel - it did not cause a problem for them.
The stock doubled from ~ $25 to $50 during 2003 - that was before the foundry and
things like the Arrow Lake problems became more apparent.
So, I think it's a good testament that Panther is slightly better than Meteor.
January 27, 2023
<<Client Side
Meteor Lake is an important milestone for Intel for two reasons: it is the company's first client CPU to rely on a disaggregated multi-tile design, and it is the first to use the Intel 4 process technology (aka 7nm), which is Intel's first node to use extreme ultraviolet (EUV) lithography. The system-in-package will consist of four tiles: the compute tile (CPU cores), the graphics tile produced by TSMC (presumably using its N3 or N5 node), the SoC tile, and the I/O tile. The tiles are interconnected using Intel's Foveros 3D technology.
âOn Intel 4, we are ready today for manufacturing and we look forward to the MTL (Meteor Lake) ramp in the second half of the year,â said Gelsinger...
...âWith MTL progressing well, it is now appropriate to look forward to Lunar Lake, which is on track for production readiness in 2024, having taped-out its first silicon,â said the head of Intel. âLunar Lake is optimized for ultra-low power performance, which will enable more of our PC partners to create ultra-thin and light systems for mobile users.â>>
Intel: Lunar Lakes Tapes Out, Meteor Lake on Track for 2023 Ramp
Meteor Lake to ramp in 2023, Lunar Lake ready for production in 2024.
This shows the Intel did a good job with Meteor - particularly with the GPU:
Last edited:
From the context of the way it was phrased, it seems the yield (which may not be poor) has room to get up to MTL launch yield.Meteor Lake was actually doing pretty well for Intel - it did not cause a problem for them.
The stock doubled from ~ $25 to $50 during 2003 - that was before the foundry and
things like the Arrow Lake problems became more apparent.
So, I think it's a good testament that Panther is slightly better than Meteor.
January 27, 2023
<<Client Side
Meteor Lake is an important milestone for Intel for two reasons: it is the company's first client CPU to rely on a disaggregated multi-tile design, and it is the first to use the Intel 4 process technology (aka 7nm), which is Intel's first node to use extreme ultraviolet (EUV) lithography. The system-in-package will consist of four tiles: the compute tile (CPU cores), the graphics tile produced by TSMC (presumably using its N3 or N5 node), the SoC tile, and the I/O tile. The tiles are interconnected using Intel's Foveros 3D technology.
âOn Intel 4, we are ready today for manufacturing and we look forward to the MTL (Meteor Lake) ramp in the second half of the year,â said Gelsinger...
...âWith MTL progressing well, it is now appropriate to look forward to Lunar Lake, which is on track for production readiness in 2024, having taped-out its first silicon,â said the head of Intel. âLunar Lake is optimized for ultra-low power performance, which will enable more of our PC partners to create ultra-thin and light systems for mobile users.â>>
Intel: Lunar Lakes Tapes Out, Meteor Lake on Track for 2023 Ramp
Meteor Lake to ramp in 2023, Lunar Lake ready for production in 2024.www.tomshardware.com
This shows the Intel did a good job with Meteor - particularly with the GPU:
On the other hand, Meteor Lake even after launch may not be a totally safe reference: https://www.anandtech.com/show/2149...amp-as-meteor-lake-chips-were-in-short-supply
Fred Chen, if you read that article closely, it looks like demand was high for Meteor Lake -
so Intel decided to ramp Meteor a lot faster in Ireland than anticipated:
<<The first of Intel's high-volume manufacturing (HVM) fabs to be equipped for the Intel 4 and Intel 3 processes, Fab 34 in Ireland is a critical element to Intel's cutting-edge product plans over the next couple years. Intel was not initially planning on relying so much on Fab 34 this soon â instead using their Oregon development fabs to do more of their Intel 4 & Intel 3 fabrication â but the company opted to ramp up at a faster pace. The benefit to Intel is that they get more fab capacity sooner, but it means they're incurring around $1 billion in costs now of what would have otherwise been spread out over further quarters during a more gradual ramp-up.>>
so Intel decided to ramp Meteor a lot faster in Ireland than anticipated:
<<The first of Intel's high-volume manufacturing (HVM) fabs to be equipped for the Intel 4 and Intel 3 processes, Fab 34 in Ireland is a critical element to Intel's cutting-edge product plans over the next couple years. Intel was not initially planning on relying so much on Fab 34 this soon â instead using their Oregon development fabs to do more of their Intel 4 & Intel 3 fabrication â but the company opted to ramp up at a faster pace. The benefit to Intel is that they get more fab capacity sooner, but it means they're incurring around $1 billion in costs now of what would have otherwise been spread out over further quarters during a more gradual ramp-up.>>
It could happen again, with Panther Lake accelerated ramp in Fab 52/62.
siliconbruh999
Well-known member
Yeah if demand is very high than they are going to be ramping faster.
Panther Lake is important proof of 18 A health and critical for product biz, but doesn't move the needle at all for IFS business success. It is the same poison pill as MTL was for intel 4 fab biz. Tiny compute tile, that even if wildly successful in sales, does not fill a single fab's capacity (not even at 3x MTL volume). It is the type of thing that would have been completely hidden from view in the old intel, were it not for Pat's brave decision to open the books on their internal "foundry" operations. We can expect dismal financial results for fabs on initial ramp of 18A Panther Lake in AZ, just like MTL in IRE. It is good decision for product yield and fast ramp, but terrible for current fab biz.
To have a successful fab biz will take 1) wildly successful data center CPU's that drive much more Si area, and 2) external customer parts with larger monolithic die and/or multiple chiplets (microsoft, Google, Amazon, +?). Wild cards: how aggressively does intel work to bring back tiles already successfully produced by tsmc, and do they revert to more SoC and less SoP in their internal parts for client to reduce performance tax from chiplet interconnections. It seems they are still betting on external customer volume and keeping tsmc capacity reserved for their existing GPU/SoC/IO tiles. We'll see soon enough if this was the right decision.
On the plus, side, Intel 3 should continue to ramp and eventually fill Fab 34 due to intel 3 base tile usage on 18 A parts - big, easy-to-yield parts that mostly drive a lot of BEOL capacity. I'm not betting on intel ever having a successful data center GPU after so many missteps and resets - they are too far behind and at least as long as product/fabs are one company, they should fully-commit to winning NVidia foundry biz rather than continuing to pour money from the product biz into this empty hole. Intel still seems to have a strong advantage in high performance space important to Nvidia that tsmc still covets, even though they lapped them in mobile/low-power space long ago.
To have a successful fab biz will take 1) wildly successful data center CPU's that drive much more Si area, and 2) external customer parts with larger monolithic die and/or multiple chiplets (microsoft, Google, Amazon, +?). Wild cards: how aggressively does intel work to bring back tiles already successfully produced by tsmc, and do they revert to more SoC and less SoP in their internal parts for client to reduce performance tax from chiplet interconnections. It seems they are still betting on external customer volume and keeping tsmc capacity reserved for their existing GPU/SoC/IO tiles. We'll see soon enough if this was the right decision.
On the plus, side, Intel 3 should continue to ramp and eventually fill Fab 34 due to intel 3 base tile usage on 18 A parts - big, easy-to-yield parts that mostly drive a lot of BEOL capacity. I'm not betting on intel ever having a successful data center GPU after so many missteps and resets - they are too far behind and at least as long as product/fabs are one company, they should fully-commit to winning NVidia foundry biz rather than continuing to pour money from the product biz into this empty hole. Intel still seems to have a strong advantage in high performance space important to Nvidia that tsmc still covets, even though they lapped them in mobile/low-power space long ago.
I had to disagree with that. To command for a higher multiple, having a meaningful share in the AI market is a must. Intel is already gearing towards it. They just need to continue, the same way as AMD does it to compete with Nvidia.
hist78
Well-known member
I think Intel's new strategy right now is to do everything to improve Intel Product Division's product offerings and profitability, using both Intel Foundry and TSMC. Profitability and Intel product competiveness are first priority, even if that means using TSMC's fabs or cut back Intel Foundry's spending on getting more external foundry customers.
Finding more external customers is important for Intel Foundry but it's second priority or third priority now.
It's official: Intel will continue outsourcing up to 30% of its wafer production to TSMC
"Q - Joe Moore (Semiconductor Industry Analyst at Morgan Stanley) So, you went through a CEO transition, lots of transformational stories coming out every day. But it seems like from the way you're talking about on the earnings call and the meetings you've been doing, strategy is still...
semiwiki.com
It must be an equal priority as it needs IFS to break even asap to recover its finance.
This appears to be true.
I personally believe it would take slightly more than a miracle for IFS to actually be in the green in the next 3 years.
The idea that Intel will become profitable by gaining a significant performance competitive position is also questionable IMO (regardless of which fab they use to achieve it). At best, Intel will hold a small lead over AMD. Currently, AMD is managing a performance lead while keeping costs lower (and fabbing on an older, less dense node).
It seems like the market cap for pricing will not change for desktop/laptop chips .... and may actually be driven downward somewhat.
I wonder if the "regain the performance crown at all costs" strategy might just be a little too rich for Intel's current economic and political position.
Maybe they can break even if Intel Product puts in extra 15 billion revenue?
If Intelâs fab processes are not competitive with TSMCâs processes, whatâs the point of keeping them around. Might as well shut them down.
Why maybe? They said the break-even for IFS is largely due to transition away from Intel 7. Not due to revenue.
We can see from Intel's last quarterly conference call that their foundry losses have been improving -
they're the least amount of losses since late 2023 - check slide 8 below. If there can be a Windows
11 refresh like several analysts are predicting, Intel could be making additional traction toward their
goal of breakeven in foundry in late 2027.
Break-even in Foundry is due to shifting away from Intel 7 - not an if.
"""
Can you just remind us of the walkthrough to breakeven in foundry? And how much of that is reliant upon external foundry wins?
John Pitzer, Intel: Yes. To answer the last part of your question first, not very much. I mean, we talked about sort of a target of giving Intel foundry to op profit breakeven by the end of twenty twenty seven. Thatâs mostly on the back of Intel products and actually mostly on the transition from moving our mix from Intel seven wafers to Intel 18A. As weâve talked about in the past, as we make that transition, you will see ASP per wafer go up three times faster than cost per wafer.
And itâs really whatâs driving us to a breakeven op profit point as we exit 2027.
"""
Thanks - that could put Intel's total foundry revenue at roughly $40 billion
by the end of this decade. John Pitzer recently said (at the Morgan Stanley
Conference) that the company's projections of roughly $15 billion in outside
foundry by late 2027 (I think) is realistic and conservative.
siliconbruh999
Well-known member
Hard disagree here PTL tile is not small it is as big as a Mobile SoC around114mm2 (almost 62% bigger than MTL 69mm2) and 18A is not like Intel 4 we have like 5-6 products on 18A:
Panther Lake , Clearwater Forest, Nova Lake(mobile compute tile and iGPU), Diamon Rappids,Future Client GPU,Wildcat lake vs Intel 4 Meteor Lake CPU, DMR/GNR SoC IO tile and Horse Creek and Lohi 2.
Fab are all about tech and volume Intel domination for 20+ years were due to their fabs.
siliconbruh999
Well-known member
he said $15 billion annual external sales by 2030 and if we include internal and increased ASP that should be like easily $40 billion annual revenue if they don't fumble