Intel Foundry Update from the Investor Call

[amitpatel.gt]

nghanayem said:

Unless 20A no longer exists why would you come to that conclusion? The performance isn't what makes 18A 18A or 20A 20A. The process changes and enhancements are what makes intel 3 not intel 4, intel 20A not intel 3, and 18A not 20A. By this logic 10nm icelake is 14nm++, and 4LPE is the same as 5LPE because there was no improvement to power performance characteristics. Don't get me wrong if they are weaker than they should be, that is a failure. But just because getting the performance you wanted from the node was not possible within a fixed development timeframe that doesn't mean the process doesn't exist anymore. That is simply not how process definition or development work.

If Intel didn’t meet their 20A or 18A perf per Watt improvement targets, then at least they should communicate to their investors. Especially when we are talking about a huge gap of 15%/W vs 26.5%/W goal of Intel 3 to 18A.

If you look at interview of Dr. Ann Kelleher from 2023 on more-than-Moore, he specifically asks if Intel 3 and 18A qualify as a full node jump, and she said yes because the perf per watt improvement is atleast 10%.

So by their own metric, either 18A or 20A should not be named as a full node jump if they don’t meet their targets.

nghanayem said:

If Scotten's estimates of how the processes stack up prove correct, then TSMC is starting from a lower starting point on N3E than intel on intel 3. Also A16 will start high volume production around 1.5-2 yrs after 18A if we take both manufacturers at their word. If A16 couldn't surpass 18A that would be a very bad omen for TSMC's technological competitiveness given A16 is more of a 14A competitor than an 18A competitor.

Scotten Jones’ estimate is raw performance and not performance per Watt. They are very different things. I doubt Intel 3 is better than N3E in performance per Watt.

nghanayem said:

Care to elaborate what you mean by this? 5N4Y had the goal of having "process performance per watt parity in 2024 and leadership in 2025" first and foremost. Having 18A be a 26.5% perf/W uplift over intel 3 was a goal, if you fall short of a goal I don't see how that is lying unless intel never had any intention of hitting that target. If you try to hit a target and fail, that is failure not deceitfulness. If you want to call intel out for failing to meet all of their goals I won't argue with you on that count, but I don't see how you arrived at intel "lying to themselves or investors".

The 2023 10K was filed only about 6 months ago. I doubt a failure of such a scale happens when you are less than 1 year out from volume production.

As recently as 6 months ago they told the World (including Intel products and external customers) that 18A is 26.5%/Watt better than Intel 3. If they deliver only 15%/Watt instead, and they don’t even openly communicate their shortcomings, then something is not right…

 

[prime007]

Popular (and reputable) YouTuber, Steve Burke (Gamers Nexus) just released a video criticizing Intel's past/current behavior and stating their fab(s?) haven't been running smoothly. I've linked the video where he specifically begins to talk about the following...
1. One Intel employee stated that fab is running a skeleton crew for a long time
2. Another source stated manufacturing challenges allegedly persist even in spite of fixing the problem Intel has been talking about.

Steve points out numerous other issues with Intel and honestly (after watching the entire video) in my humble opinion Intel needs to REMOVE Pat Gelsinger as CEO. He's lost investor confidence, lost costumer goodwill, mortgaged his fabs and severely hurt employee morale. But I'm not a semi or financial pro...just someone interested in following the story/history of the semi industry. Perhaps the 18A will stop the bleeding and 14A will actually turn things around or maybe the US government bails out Intel...I don't know but things seem dire for Intel/Pat right now.

It appears Gamers Nexus isn't done with Intel and that he will be releasing more videos of Intel's "slimy" behavior.

 

[hist78]

Popular (and reputable) YouTuber, Steve Burke (Gamers Nexus) just released a video criticizing Intel's past/current behavior and stating their fab(s?) haven't been running smoothly. I've linked the video where he specifically begins to talk about the following...
1. One Intel employee stated that fab is running a skeleton crew for a long time
2. Another source stated manufacturing challenges allegedly persist even in spite of fixing the problem Intel has been talking about.

Steve points out numerous other issues with Intel and honestly (after watching the entire video) in my humble opinion Intel needs to REMOVE Pat Gelsinger as CEO. He's lost investor confidence, lost costumer goodwill, mortgaged his fabs and severely hurt employee morale. But I'm not a semi or financial pro...just someone interested in following the story/history of the semi industry. Perhaps the 18A will stop the bleeding and 14A will actually turn things around or maybe the US government bails out Intel...I don't know but things seem dire for Intel/Pat right now.

It appears Gamers Nexus isn't done with Intel and that he will be releasing more videos of Intel's "slimy" behavior.

Don't throw nonsense to those YouTube influencers. It can be very dangerous.

 

[siliconbruh999]

Popular (and reputable) YouTuber, Steve Burke (Gamers Nexus) just released a video criticizing Intel's past/current behavior and stating their fab(s?) haven't been running smoothly. I've linked the video where he specifically begins to talk about the following...
1. One Intel employee stated that fab is running a skeleton crew for a long time
2. Another source stated manufacturing challenges allegedly persist even in spite of fixing the problem Intel has been talking about.

Steve points out numerous other issues with Intel and honestly (after watching the entire video) in my humble opinion Intel needs to REMOVE Pat Gelsinger as CEO. He's lost investor confidence, lost costumer goodwill, mortgaged his fabs and severely hurt employee morale. But I'm not a semi or financial pro...just someone interested in following the story/history of the semi industry. Perhaps the 18A will stop the bleeding and 14A will actually turn things around or maybe the US government bails out Intel...I don't know but things seem dire for Intel/Pat right now.

It appears Gamers Nexus isn't done with Intel and that he will be releasing more videos of Intel's "slimy" behavior.

If you out gelsinger who do you bring ?

 

[Maximus]

CC Wei says tape-outs which is easily measurable. Someone on the next investor call should ask Pat how many tape-outs they have in progress for 18A thus far. If CC Wei can see N2 tape-outs Pat should be able to see 18A tape-outs.

Given Pat's track record and personality, if he didn't actively mention tape-outs, I would guess there aren't a significant number of them.

 

[nghanayem]

nghanayem said:


If Intel didn’t meet their 20A or 18A perf per Watt improvement targets, then at least they should communicate to their investors. Especially when we are talking about a huge gap of 15%/W vs 26.5%/W goal of Intel 3 to 18A.

If you look at interview of Dr. Ann Kelleher from 2023 on more-than-Moore, he specifically asks if Intel 3 and 18A qualify as a full node jump, and she said yes because the perf per watt improvement is atleast 10%.

So by their own metric, either 18A or 20A should not be named as a full node jump if they don’t meet their targets.

I think this argument is the more compelling of your arguments.

Process node naming from the big three is stupid rant:
The thing I don't like is if we go by that metric N3E isn't a full node over N4P and N5 isn't a different node than N4. Of course claiming N3E is not a different node from N4P would be a ridiculous conclusion to come to. Given all of the process differences I would even go so far as to say N3E and N3 are completely different nodes just with very similar characteristics. I say that to say this, the industry doesn't have any metric for what does or doesn't quantify as "a node" it is whatever these firms want to call "a node". If TSMC, intel, and Samsung want to call a derivative a node it is kind of hard to shame any one of them when all three firms follow the rule of there being no rule. Because there are no rules N5 and N4, intel 4/3, and intel 20/18A being different nodes might as well be personal judgement. For me personally none of these derivatives are different nodes just members of a family. My metric for what does and doesn't count is if there are significant changes to the process flow it is a new node by my book and if their aren't then it is a derivative.

Process node naming rant over:
If we put aside the superficial question of naming, Intel said 20A offers "ribbonFET and powerVIA and will debut in the Arrowlake generation of products". If there is any Arrowlake part with these two features that is 20A not intel 3 and not 18A. Intel says about 18A "ribbon optimizations for design optimization, and line width reductions debuting in CWF and PNL". If there are any FE process changes from 20A Arrowlake, or line width reductions from the minimum values found in ARL that is listed as a feature of 18A. If there are new libraries not present in ARL (as an example the 210h lib in intel 3 or TSMC 12FF vs 16FF or GF 14LP vs GF12LP) that would also be an 18A feature. TLDR unless we see none of the changes intel mentioned for 18A, then 20A and 18A aren't "just the same thing with a new name". Now will 18A be 10% better than 20A? Hard to say without intel providing more details on how the same IPs fair with the new process revision becaue intel has never mentioned an 18A ARL or 20A anything else for us to have a like for like comparison publicly available. All we can definitively confirm is if we can for line width reductions, process enhancements for yield, cap, drive etc, or new libraries.

nghanayem said:


Scotten Jones’ estimate is raw performance and not performance per Watt. They are very different things. I doubt Intel 3 is better than N3E in performance per Watt.

Nobody measures raw frequency. If they did even A3 class nodes will be absolutely spanked by 180nm HEMTs. Advanced logic performance is ALWAYS measured at iso power, iso voltage, or iso leakage otherwise you aren't making a useful comparison. With that said I highly doubt intel 3 can deliver lower power mobile APs than N3E (something intel indirectly admitted), but I think it is plausible for an intel 3 chip to use less power than the same design on N3E in higher power HPC scenarios. For its part, Intel claims that intel 3 is "only" "about equal" to presumably N3E rather than way ahead in Scotten's models. They are not necessarily measuring from the same point with the same methodology so some degree of delta is to be expected.

nghanayem said:


The 2023 10K was filed only about 6 months ago. I doubt a failure of such a scale happens when you are less than 1 year out from volume production.

As recently as 6 months ago they told the World (including Intel products and external customers) that 18A is 26.5%/Watt better than Intel 3. If they deliver only 15%/Watt instead, and they don’t even openly communicate their shortcomings, then something is not right…

At that accounting change investor meeting all they said is where they think 18A stacks up versus the best node from competitors (whatever that means) with vague "+", "-", and "~". If 18A products come out before N2 products and the process leads N3P perf per watt with similar density, then I don't see how that would be inconsistent with what was presented in April or the prior claim of "perf/watt leadership in 2025".

 

[amitpatel.gt]

I think this argument is the more compelling of your arguments.

Process node naming from the big three is stupid rant:
The thing I don't like is if we go by that metric N3E isn't a full node over N4P and N5 isn't a different node than N4. Of course claiming N3E is not a different node from N4P would be a ridiculous conclusion to come to. Given all of the process differences I would even go so far as to say N3E and N3 are completely different nodes just with very similar characteristics. I say that to say this, the industry doesn't have any metric for what does or doesn't quantify as "a node" it is whatever these firms want to call "a node". If TSMC, intel, and Samsung want to call a derivative a node it is kind of hard to shame any one of them when all three firms follow the rule of there being no rule. Because there are no rules N5 and N4, intel 4/3, and intel 20/18A being different nodes might as well be personal judgement. For me personally none of these derivatives are different nodes just members of a family. My metric for what does and doesn't count is if there are significant changes to the process flow it is a new node by my book and if their aren't then it is a derivative.

Process node naming rant over:
If we put aside the superficial question of naming, Intel said 20A offers "ribbonFET and powerVIA and will debut in the Arrowlake generation of products". If there is any Arrowlake part with these two features that is 20A not intel 3 and not 18A. Intel says about 18A "ribbon optimizations for design optimization, and line width reductions debuting in CWF and PNL". If there are any FE process changes from 20A Arrowlake, or line width reductions from the minimum values found in ARL that is listed as a feature of 18A. If there are new libraries not present in ARL (as an example the 210h lib in intel 3 or TSMC 12FF vs 16FF or GF 14LP vs GF12LP) that would also be an 18A feature. TLDR unless we see none of the changes intel mentioned for 18A, then 20A and 18A aren't "just the same thing with a new name". Now will 18A be 10% better than 20A? Hard to say without intel providing more details on how the same IPs fair with the new process revision becaue intel has never mentioned an 18A ARL or 20A anything else for us to have a like for like comparison publicly available. All we can definitively confirm is if we can for line width reductions, process enhancements for yield, cap, drive etc, or new libraries.

Nobody measures raw frequency. If they did even A3 class nodes will be absolutely spanked by 180nm HEMTs. Advanced logic performance is ALWAYS measured at iso power, iso voltage, or iso leakage otherwise you aren't making a useful comparison. With that said I highly doubt intel 3 can deliver lower power mobile APs than N3E (something intel indirectly admitted), but I think it is plausible for an intel 3 chip to use less power than the same design on N3E in higher power HPC scenarios. For its part, Intel claims that intel 3 is "only" "about equal" to presumably N3E rather than way ahead in Scotten's models. They are not necessarily measuring from the same point with the same methodology so some degree of delta is to be expected.

At that accounting change investor meeting all they said is where they think 18A stacks up versus the best node from competitors (whatever that means) with vague "+", "-", and "~". If 18A products come out before N2 products and the process leads N3P perf per watt with similar density, then I don't see how that would be inconsistent with what was presented in April or the prior claim of "perf/watt leadership in 2025".

You’re basically defending the behavior that the 2023 10K filing to SEC (6 months ago) is irrelevant… and even if there are major setbacks vs what they told SEC, they don’t need to be communicated to investors.

This is just further eroding Intel’s credibility and inviting litigation... if true, that is.

PS: my intent behind raising this issue is to have someone confirm if there was a mistake in their Q2 earnings slide # 7 or 18A really is under performing. Can someone please verify?

Thanks!