[SemiAccurate] Intel kills off the 10nm process

[Vattila]

Wow!

Here is my take: There is probably not smoke without fire, even if SemiAccurate overstates the issue by calling it a "10nm cancellation". However, process node names are marketing, and Intel is probably trying to carefully manage the narrative, not to spook investors and share price.

Intel is probably pushing ahead on 10nm to meet investor and customer commitments they have made. But I highly doubt the original specifications of the 10nm process have survived. SemiAccurate has previously claimed specifications have been relaxed and that the process is more like "10nm--" or "12nm" now. That means that Intel probably will have problems reaching the performance of 14nm++(+). The latter already has performance well beyond the target of the initial 10nm, as shown on Intel's own slides.

So what can we expect on the Intel roadmap next year? I expect mostly 14nm++(+) all year, with tiny 10nm ("12nm") chips in (somewhat) high-volume manufacture for notebooks in 2019-Q4, probably Ice Lake-Y/U. Even if the 10nm plans have survived, mainstream and HEDT desktop will likely have to wait until 10nm+ Ice Lake-S/X sometime in 2020 at best, and we will only get some minor 14nm++(+) refreshes in 2019. For server, leaked roadmaps have already shown 10nm+ Ice Lake-SP by mid-2020. Until then it is all 14nm++(+).

However — maybe SemiAccurate's claim has as much to do with products as with process. What if Intel has recently updated their analysis of Zen 2 and its performance? Maybe they've found that Ice Lake-S and Ice Lake-SP on "12nm" will be woefully inadequate? Perhaps they have been cancelled? If so, it may mean the whole 10nm roadmap has been cancelled, and all focus is now on 7nm Granite Rapids and successors, meaning their redefined 10nm ("12nm") will be a low-power process for notebooks only, i.e. Ice Lake-U/Y.

On topic: any unusual activity recorded in relation to Intel shares leading up to, and then after that SemiAccurate link was first published?

Stock price may be affected by rumour that Murthy Renduchintala has been chosen as their next CEO.

 

[catonkatonk]

What is rather likely is Intel relaxing the m2 pitch from 36 to 40 or the m1/2 from 40/36 to both 44 to avoid the quad patterning of those two metal pitches. Depending on what their double patterning can handle it will either be 40 or 44. TSMC can handle 40 on their 7nm process but it's really unclear why Intel went with 36 in the first place.

But Semiaccurate already had an article about Intel relaxing the specs, and implied that this modification had smoothed things over considerably.
He's not going to report the same news with a different headline... Right?

 

[Spartak]

Let's not forget Intel significantly relaxed the metal pitches with 14nm+ and who know with ++ and +++ yet they still call it 14nm. The same is likely happening to 10nm which will be called 10nm+ mark my words.

 

[Spartak]

But Semiaccurate already had an article about Intel relaxing the specs, and implied that this modification had smoothed things over considerably.
He's not going to report the same news with a different headline... Right?

They might have changed the cobalt gating too so maybe that's a definite cancellation of their original process in his eyes.

 

[DrMrLordX]

Option 4- Intel kills the 10nm for its own chips, but has big deal with apple on other stuff

Possible. Intel has already proven that they can fab chips on their existing 10nm node; sadly, the chips suck. But the node might be suitable for low-power stuff like iPhone modems. Salvaging their "broken" 10nm node to provide Apple with over 100 million modems for the next gen iPhone would take a lot of stress off of whatever processes they use in the future to produce CPUs.

I don't think Apple would be dumb enough to risk their product pipeline on Intel's constantly delayed processes- they've been getting progressively worse at hitting schedules since 22nm.

10nm already exists, and it (sort of) works. Just not for Core.

People, people.

If Intel's struggling immensely on the 10nm process, do you think they'll be better off skipping to a 7nm process? Technology improvements are iterative.

They've already plowed an enormous amount of time and resources into 10nm. Whatever it was that they were going to learn from 10nm, they've probably learned it. They aren't going to unlearn all that by shifting focus to 7nm on EUV.

Intel will never really "kill" 10nm. They may replace it with a less-ambitious 10nm-ish node as a stopgap before pushing 7nm (hence SemiAccurate's claim that 10nm is "dead"). Or they may relegate what's left of 10nm to low power sideline projects, such as modems and other bits.

What's clear is that 10nm, as it exists today, will not be used for Core products, and may not even be used for Atom products (Tremont).

They will probably announce the new CEO too it looks like.

Go Raja go!

(lulz)

 

[maddie]

Intel's solution to their 10nm problems.

The king is dead, long live the king.

 

[ttechf]

There was a tweet by Intel which they stated that they HAVE NOT killed off their 10nm process. BUT what someone else said that it doesn't necessarily mean they are going to 7nm right away. Perhaps they could just be jumping to 10nm PLUS [10nm+] instead?

 

[catonkatonk]

There was a tweet by Intel which they stated that they HAVE NOT killed off their 10nm process. BUT what someone else said that it doesn't necessarily mean they are going to 7nm right away. Perhaps they could just be jumping to 10nm PLUS [10nm+] instead?

It was assumed that they would do that, though. Canonlake would be limited to this weird limited almost-paper launch, and Icelake would be on 10nm+.

 

[Vattila]

Perhaps they could just be jumping to 10nm PLUS [10nm+] instead?

Would they respond to difficulty by making things more difficult, by piling on features and iterative improvement they were planning to do on top of a working initial 10nm process? That sounds unlikely to me. Intel has reiterated claims to fulfil promises about 10nm product on shelves by 2019-Q4. If 10+ and 10++ are still alive, there would no cancellation or bad news of any sort to report. For SemiAccurate's big headline to make sense, I suspect 10+ and 10++ have been cancelled, along with the respective products (Ice Lake-S/X/SP). They probably will fall flat against Zen 2/3, and Intel has decided to accelerate 7nm next-generation products instead.

 

[Spartak]

What's really weird to me is how up to as close as a year ago, well maybe one and a half, they were still touting their 3yr process advantage in a way they still believed in it themselves . It really seems they got both blindsided about the competition and overconfident in their own process engineering.

 

[Maxima1]

Would they respond to difficulty by making things more difficult, by piling on features and iterative improvement they were planning to do on top of a working initial 10nm process? That sounds unlikely to me. Intel has reiterated claims to fulfil promises about 10nm product on shelves by 2019-Q4. If 10+ and 10++ are still alive, there would no cancellation or bad news of any sort to report. For SemiAccurate's big headline to make sense, I suspect 10+ and 10++ have been cancelled, along with the respective products (Ice Lake). They probably will fall flat against Zen 2/3, and Intel has decided to accelerate 7nm next-generation products instead.

Ice Lake was always suppose to be 10nm+. We already knew the 10nm CNL process was screwed but 10nm+ or 10nm++ may have not been. This is common knowledge.

What's really weird to me is how up to as close as a year ago, well maybe one and a half, they were still touting their 3yr process advantage in a way they still believed in it themselves . It really seems they got both blindsided about the competition and overconfident in their own process engineering.

It's called lying.

 

[Spartak]

Would they respond to difficulty by making things more difficult, by piling on features and iterative improvement they were planning to do on top of a working initial 10nm process? That sounds unlikely to me. Intel has reiterated claims to fulfil promises about 10nm product on shelves by 2019-Q4. If 10+ and 10++ are still alive, there would no cancellation or bad news of any sort to report. For SemiAccurate's big headline to make sense, I suspect 10+ and 10++ have been cancelled, along with the respective products (Ice Lake). They probably will fall flat against Zen 2/3, and Intel has decided to accelerate 7nm next-generation products instead.

14nm+ was a relaxing of the original 14nm node, since quad patterning seems the main issue here, a similar 10nm+ that relaxes the lowest metal pitches would be an exact repeat of 14nm.
10nm+ =! more dense by definition .

 

[jpiniero]

I was thinking more along the lines that Intel created a special 14 nm version that's optimized for low power and high density and ported Icelake Client and perhaps the modem to it. They are now calling that 10 nm. And then they are essentially cancelling everything else.

 

[deathBOB]

What's really weird to me is how up to as close as a year ago, well maybe one and a half, they were still touting their 3yr process advantage in a way they still believed in it themselves . It really seems they got both blindsided about the competition and overconfident in their own process engineering.

It’s easy to fool yourself when you really need to fool yourself.

I could see this. Why polish this turd when EUV is apparently actually happening and you’re still selling every 14nm chip you can make?

 

[Vattila]

Ice Lake was always suppose to be 10nm+. We already knew the 10nm CNL process was screwed but 10nm+ or 10nm++ may have not been. This is common knowledge.

Still doesn't make sense though. We have to presume Intel isn't lying. 10nm products will be on shelves for 2019-Q4. If, as you suggest, Intel has a leapfrogging team achieving success on 10nm+, and this now replaces the initial 10nm, with products taped out and presumed ready to go for high-volume manufacture by mid-2019 —well, that would be just excellent news, wouldn't it? SemiAccurate's tone is different. It clearly communicates further trouble and limited future scope for 10nm.

 

[Zucker2k]

Semi-accurate has never been fully accurate hence the name. Burying it behind such a massive paywall enables them to stir up the hype without the need to deliver the meat. Those silly 5 investors won't share that 'knowledge' or make a dent in the stock market.

FWIW people speculating about skipping 10nm for 7nm clearly have no clue about the consequences and the ambitious specifications of 7nm. It's more dense then TSMC's 3nm process based on what is known right now about both processes.

If you can't jump 5ft you won't solve that by jumping 10ft.

What is likely at play here is Intel relaxing the m2 pitch from 36 to 40 or the m1/2 from 40/36 to both 44 to avoid the quad patterning of those two metal pitches. Depending on what their double patterning can handle it will either be 40 or 44nm. TSMC can handle 40 on their 7nm process but it's really unclear why Intel went with 36 in the first place.

Charlie sells that as 'dropping 10nm' which is simply hyperbole from an attention whore.

This. Charlie wants Intel to deliver what we now know as impossible, and because Intel cannot deliver the impossible, they have failed to deliver. Charlie is the biggest anti-Intel troll on the internet.

 

[railroadmaster]

Intel should just buy 7nm technologies from IBM, TSMC, GF, Samsung or whoever else has the technology and use that tech in their fabs. They could buy it for pocket change.

 

[VirtualLarry]

Intel is probably pushing ahead on 10nm to meet investor and customer commitments they have made. But I highly doubt the original specifications of the 10nm process have survived. SemiAccurate has previously claimed specifications have been relaxed and that the process is more like "10nm--" or "12nm" now. That means that Intel probably will have problems reaching the performance of 14nm++(+). The latter already has performance well beyond the target of the initial 10nm, as shown on Intel's own slides.

The thing is, Intel's Core microprocessors, on whatever advanced 14nm (+,++, +++,etc) node, aren't even taking advantage of all of the density improvements of those processes, due to performance reasons, transistor sizes for switching speeds, dark silicon area for cooling needs, etc.

So, it makes sense that if they were focusing on density for 10nm, that it might not reach performance goals for their Core microprocessors, if they stuck to those density metrics.

But by relaxing them...

 

[Vattila]

FWIW people speculating about skipping 10nm for 7nm

Here's a thought — It might not be their own 7nm. Perhaps they won't put all their eggs in the same basket, as they did for 10nm. If they see that 10nm+(+) product plans will not be competitive, it may be better to cancel them and ride it out with 14nm++(+) until they have something better. It would give them time, and they can hedge their bets by designing for foundry processes as an insurance (like Lisa Su had parallel engagements with GF and TSMC, and probably flirted with Samsung, as well).

 

[NostaSeronx]

Intel's major issue is that they bought into ASML. Planned obsolescence in manufacturing is a massive hurdle for Intel.

Intel however quickly released their ASML stake. So, there is only two options for Intel further 193i with Nikon Precision or J-FIL from Canon Nanotechnologies.

NP's NSR-631/NSR-635 with iAS attached and CNT's NZ3C seem to be the tools of choice for Intel for post-10nm. Intel and GlobalFoundries are both leaving EUV and thus are differentiating from TSMC and Samsung.

10-nm is still coming out, but it is remaining LVM till Intel restructures its tool roadmap.

 

[jpiniero]

FWIW, Charlie updated the story with Intel's denial and is standing by the story.

 

[frozentundra123456]

This. Charlie wants Intel to deliver what we now know as impossible, and because Intel cannot deliver the impossible, they have failed to deliver. Charlie is the biggest anti-Intel troll on the internet.

Yea, I love the way Charlie uses carefully nuanced language, is always sure of his facts, and never goes off on a rant. Some people want to pay 1000.00 to see what is behind the paywall. I would pay 1000.00 if it was needed to not read his overblown rhetoric.

 

[mikk]

Charlie also said there will be another 14nm+++ which is wrong, Coffeelake-R is on 14nm++. Charlie isn't reliable.

 

[IntelUser2000]

If so, by getting rid of the quad patterning steps 7nm would be free of what broke 10nm.

Some claim its due to cobalt, some claim its the difficulty due to going above quad patterning.

My verdict?: It's all speculation until its actual.

People like techinsights could quickly take one of these hypothetical chips apart and do their measurements only to find parameters that match a 14nm derivative, not a 10nm class process.

Why speculate when you have hard data?

https://fuse.wikichip.org/news/1371...ghts-reports-on-the-i3-8121u-finds-ruthenium/

Tech Insights already opened the chip and analyzed it. And yes, its using 10nm technology.

"As far as cell size is concerned, there are no surprises. TechInsights silicon measurements matches Intel’s own IEDM paper."
"Techinsights has confirmed Intel is using contact-on-active gate (COAG)."
"For the three lowest layers (poly, metal 0, and metal 1), Intel introduced Cobalt and Tungsten."

 

[jpiniero]

The 8121U is definitely the real 10 nm. The question becomes if any other product also called 10 nm will actually use the 10nm tooling and/or be anywhere near as dense.