Friday, September 28th 2018

Intel At Least 5 Years Behind TSMC and May Never Catch Up: Analyst

Intel's in-house sub-10 nanometer silicon fabrication dreams seem more distant by the day. Raymond James analyst Chris Caso, in an interview with CNBC stated that Intel's 10 nm process development could set the company back by at least 5 years behind TSMC. In its most recent financial results call, Intel revised its 10 nm outlook to reflect that the first 10 nm processors could only come out by the end of 2019. "Intel's biggest strategic problem is their delay on 10nm production - we don't expect a 10nm server chip from Intel for two years," analyst Chris Caso said in a note to clients Tuesday. "10nm delays create a window for competitors, and the window may never again close."

By that time, Intel will have missed several competitive milestones behind TSMC, which is in final stages of quantitatively rolling out its 7 nm process. Caso predicts that by the time Intel goes sub-10 nm (7 nm or something in that nanoscopic ballpark), TSMC and Samsung could each be readying their 5 nm or 3 nm process roll-outs. A Rosenblatt Securities report that came out late-August was even more gloomy about the situation at Intel foundry. It predicted that foundry delays could set the company back "5, 6, or even 7" years behind rivals. Intel is already beginning offload some of its 14 nm manufacturing to TSMC. Meanwhile, AMD is reportedly planning to entirely rely on TSMC to make its future generations of "Zen" processors.
Sources: CNBC, MyDrivers
Add your own comment

39 Comments on Intel At Least 5 Years Behind TSMC and May Never Catch Up: Analyst

#1
hat
Enthusiast
Seems a bit unlikely. Even given that TSMC will have no issues with 7nm, that view relies on assumptions that Intel will continue to have nothing but trouble with 10nm and be stuck there forever while everyone else just blazes through not only 7nm, but 5nm and 3nm nodes with no issues...
Posted on Reply
#2
HTC
hat said:
Seems a bit unlikely. Even given that TSMC will have no issues with 7nm, that view relies on assumptions that Intel will continue to have nothing but trouble with 10nm and be stuck there forever while everyone else just blazes through not only 7nm, but 5nm and 3nm nodes with no issues...
Agreed.

That Intel is behind is quite obvious but i'd expect @ most 2 years. Unless ofc their 10nm woes continue: then it might actually be longer.

So far, we (me, @ least) haven't heard any bad things regarding the 7nm process being behind schedule so i'm assuming it's on track to it's target date.

Also, as you pointed out, they seem to think lower process nodes will be "as easy" to transition to as it is to 7nm: seriously doubt this will be the case.
Posted on Reply
#3
Vayra86
Seriously, everybody thinks 7nm is a real thing already but we know the current non-EUV 7nm process is costly and not spectacular yield wise and therefore less suitable for mass production. Any product on that node will be expensive and scarce.

The way this really works is that until you see mass produced 7nm product on the shelf and readily available, it really doesn't exist. Thus far, there is no real lead on Intel by TSMC. That only exists on powerpoint slides and early production samples.

Let's just take a look at early Intel 14nm and how that went down. It was slowly introduced to the market and delayed as well.
Posted on Reply
#4
londiste
As far as financial analyst can predict things, the guy is right. Intel is in serious trouble and that will (and should) reflect in the share price.

However:
Caso predicts that by the time Intel goes sub-10 nm (7 nm or something in that nanoscopic ballpark), TSMC and Samsung could each be readying their 5 nm or 3 nm process roll-outs.
Aaannnd... feel free to disregard anything this guy says on technical side of things or predictions.
Intel is expecting to get to 7nm around the same time TSMC/Samsung get to 5nm. That has been the plan all along including before their issues with 10nm.
Posted on Reply
#5
jabbadap
Vayra86 said:
Seriously, everybody thinks 7nm is a real thing already but we know the current non-EUV 7nm process is costly and not spectacular yield wise and therefore less suitable for mass production. Any product on that node will be expensive and scarce.

The way this really works is that until you see mass produced 7nm product on the shelf and readily available, it really doesn't exist. Thus far, there is no real lead on Intel by TSMC. That only exists on powerpoint slides and early production samples.

Let's just take a look at early Intel 14nm and how that went down. It was slowly introduced to the market and delayed as well.
Well yeah Apple A12 is on TSMC 7nm. So technically it is mass produced. Pricey it might be I agree with that, but then again it's apple so price usually have premium.

Edit. well of course I should have said you are right that we have nothing from the TSMC 7nm HPC process, apple's chip is from 7nm mobile process for low power applications.
Posted on Reply
#6
_Flare
Intel might have a solution to that even by today.
With Jim Keller they could speed up their product-pipeline-plans, that could be the 2nd or 3rd attempt for 10nm and/or plan for whats coming after that.

Personally i did think that Jim Keller would never join Intel because they whould never come in a dilemma like this, in hirering Jim Keller their must have had a problem what exactly he could help to solve, and he does solve big widespanning problems with mid- to long-term strategies and from design to manufacturing and everything involved.
Posted on Reply
#7
jabbadap
_Flare said:
Intel might have a solution to that even by today.
With Jim Keller they could speed up their product-pipeline-plans, that could be the 2nd or 3rd attempt for 10nm and/or plan for whats coming after that.

Personally i did think that Jim Keller would never join Intel because they whould never come in a dilemma like this, in hirering Jim Keller their must have had a problem what exactly he could help to solve, and he does solve big widespanning problems with mid- to long-term strategies and from design to manufacturing and everything involved.
Not sure where Keller fits on manufacturing process woes, which is problem to be solved by Physicists not by some computer architecture guru.
Posted on Reply
#8
notb
jabbadap said:

Edit. well of course I should have said you are right that we have nothing from the TSMC 7nm HPC process
Actually Intel does sell a 10nm mobile CPU as well (an i3).
And the laptop using it is half the price of the iPhones.
TSMC doesn't have to be that much in front as all stock speculators might want us to think. But Intel has to polish the process until they can make a profit on large Xeons at least.

If TSMC 7nm was usable for large chips, we would have seen the shrunk Vega already.
Posted on Reply
#9
bonehead123
Geez, why not just save everyone the frustration and hassle and go directly to 0.000001nm already, instead of constantly milking everything and everyone by these shitty incremental changes over several years....

you know that we know that you know it is out there, so lets be like Nike & "Just Do it" already...

fyi, in case you missed it, this post was created with sarcastic, jovial and unharmed 1's & 0's....
Posted on Reply
#10
laszlo
sh...t happens; maybe is the payback for all shady anti-amd practices ...
Posted on Reply
#11
Imsochobo
HTC said:
Agreed.

That Intel is behind is quite obvious but i'd expect @ most 2 years. Unless ofc their 10nm woes continue: then it might actually be longer.

So far, we (me, @ least) haven't heard any bad things regarding the 7nm process being behind schedule so i'm assuming it's on track to it's target date.

Also, as you pointed out, they seem to think lower process nodes will be "as easy" to transition to as it is to 7nm: seriously doubt this will be the case.
2 years may also be a strech.

What is more correct, Intel lost 5 years of their leadership cause they had 3 years since forever!
Posted on Reply
#12
john_
Intel abandoned the idea of invading mobiles, when TSMC and Samsung closed the gap in manufacturing. That manufacturing advantage could make x86 chips look like almost as efficient as ARM chips. With that advantage gone, Intel had no reason to keep spending billions promoting atoms in tablets and mobiles. Now with Intel dropping behind, we will really see if it's engineering teams are really good, or just mediocre that could take advantage of billions for R&D and the manufacturing advantage Intel was enjoying over AMD.

On the other hand we all remember what a joke 20nm from TSMC was and look where TSMC is today.
Posted on Reply
#13
epialthes
By that time, Intel will have missed several competitive milestones behind TSMC, which is in final stages of quantitatively rolling out its 7 nm process. Caso predicts that by the time Intel goes sub-10 nm (7 nm or something in that nanoscopic ballpark), TSMC and Samsung could each be readying their 5 nm or 3 nm process roll-outs
Clearly, this guy has no idea what he's talking about. TSMC's 7FF process is only an equivalent of Intel's 10nm process in almost every sense. Every manufacturers started to fudge their process number several years ago, but TSMC went too far this time. Intel has been doing the same but they're not as aggressive on fudging as others.
Posted on Reply
#14
Indra18
john_ said:
Intel abandoned the idea of invading mobiles, when TSMC and Samsung closed the gap in manufacturing. That manufacturing advantage could make x86 chips look like almost as efficient as ARM chips. With that advantage gone, Intel had no reason to keep spending billions promoting atoms in tablets and mobiles. Now with Intel dropping behind, we will really see if it's engineering teams are really good, or just mediocre that could take advantage of billions for R&D and the manufacturing advantage Intel was enjoying over AMD.

On the other hand we all remember what a joke 20nm from TSMC was and look where TSMC is today.
Nailed.. yes intel in own decade of "supremacy" lost contact with real world tech and life.. now not only tsmc is leader , dont forget chinese build own and this is now nearly 2 B people asean,market . Samsung and whole s.korea detto with stable incomes from selling in 500 million europe and investment there.. or in australia , not only IT tech but whole samsung division. And they are now reach N.koreans soo there is forming nation what will be stronger than japan..japan with lower birthrate from 1949..falling nation. Even in oozys land is not better..
Posted on Reply
#15
FordGT90Concept
"I go fast!1!11!1!"
It's only a matter of time until physics stops TSMC and Samsung too. There needs to be a revolution in process technology and Intel should be knee deep in that.
Posted on Reply
#16
Alduin
technically intel's 10nm and tsmc _ globalfoundries 7nm are almost the same. in density,power...
even intel 10 nm is better in some aspects
maybe intel lags behind competitors
but not for 5 years.
Posted on Reply
#17
zelnep
just remember - intel do not have problems with 10nm. intel has problems with 14nm! and it does not looks like intel will sort it (14nm yields) out any time soon and only then we will hear about 10nm problems - so it is multiple years behind for sure. 7nm TMSC (aka 10nm by intel standard) - have not heard problems so far - the newest iPhones use TMSC 7nm process and it is out it is real and chip or it yields are not the problem.
Posted on Reply
#18
R0H1T
zelnep said:
just remember - intel do not have problems with 10nm. intel has problems with 14nm! and it does not looks like intel will sort it (14nm yields) out any time soon and only then we will hear about 10nm problems - so it is multiple years behind for sure. 7nm TMSC (aka 10nm by intel standard) - have not heard problems so far - the newest iPhones use TMSC 7nm process and it is out it is real and chip or it yields are not the problem.
What, you have it backwards! Intel have problems with 10nm, 14nm fabs were being transitioned to 10nm & since the latter's basically MIA they have a capacity crunch.
FordGT90Concept said:
It's only a matter of time until physics stops TSMC and Samsung too. There needs to be a revolution in process technology and Intel should be knee deep in that.
I dunno, the last revolution from Intel, if you can call it that, was 22nm FinFET otherwise they've been meh since the days of conroe.
Posted on Reply
#19
londiste
zelnep said:
just remember - intel do not have problems with 10nm. intel has problems with 14nm! and it does not looks like intel will sort it (14nm yields) out any time soon and only then we will hear about 10nm problems - so it is multiple years behind for sure. 7nm TMSC (aka 10nm by intel standard) - have not heard problems so far - the newest iPhones use TMSC 7nm process and it is out it is real and chip or it yields are not the problem.
Intel has problems with both processes, the problems are just different:
- 14nm is fine and awesome, there are no yield issues, it works well, but they simply do not have manufacturing capacity (enough fabs). Most likely because they are moving fabs over to 10nm.
- 10nm, the problems are supposedly yields or reliability of the product.

TMSC 7nm we do not know much about.
- First, SOC and GPU/CPU processes are somewhat different.
- Apple A12 chips have been in manufacturing for over half a year, we do not know how much capacity TSMC has for 7nm nor how the yields are.
- A11 was 83 mm² and considering 7nm should give about twice the density (probably less that that is reality) A12 is more likely smaller rather than larger. This is the same size if not smaller than Intel's 10nm CPUs (that are 70-72 mm²).
Posted on Reply
#20
R0H1T
londiste said:
Intel has problems with both processes, the problems are just different:
- 14nm is fine and awesome, there are no yield issues, it works well, but they simply do not have manufacturing capacity (enough fabs). Most likely because they are moving fabs over to 10nm.
- 10nm, the problems are supposedly yields or reliability of the product.

TMSC 7nm we do not know much about.
- First, SOC and GPU/CPU processes are somewhat different.
- Apple A12 chips have been in manufacturing for over half a year, we do not know how much capacity TSMC has for 7nm nor how the yields are.
- A11 was 83 mm² and considering 7nm should give about twice the density (probably less that that is reality) A12 is more likely smaller rather than larger. This is the same size if not smaller than Intel's 10nm CPUs (that are 70-72 mm²).
Well we know that Kirin 980 & A12 are coming out this year IIRC, between them I bet the volumes are enough to outpace & outsell any other semiconductor behemoth, excluding Intel.
Posted on Reply
#21
First Strike
Well, that a dog bite a man is no news, but that a man bite a dog will make it all the way to the headlines. Now we see our journalist friend and analyst friend set out to bite.
Posted on Reply
#22
DeathtoGnomes
Intel spread itself to thin with so many products. This is a result of mis-management.
Posted on Reply
#23
notb
DeathtoGnomes said:
Intel spread itself to thin with so many products. This is a result of mis-management.
Exactly the opposite. It's a great decision that will help them keep the revenue and grow even further. And it's much safer as well. What you call "spreading to thin" serious people call "diversification".

You may think you only care about CPUs and GPUs, so AMD is perfectly focused and Intel just makes countless pointless products.
But AMD has a huge potential of growth in the CPU/GPU markets, which Intel lacks. So Intel has to try other things to grow.
They had some fails (GPU) and some wins (SSD, network chips, car AI). Now they're investing into drones and it looks very promising.

You know... those network chips also go into AMD-powered OEM PCs and on many AM4/TR4 motherboards. But don't worry - yours is Intel-free. I can't vouch for your car, though. ;-)
Posted on Reply
#24
WikiFM
It has been agreed that Intel's 10 nm is similar or even more advanced than TSMC's 7 nm, Samsung's 7 nm has been delayed too and they are releasing an 8 nm (10 nm+) in the meanwhile. Also Intel's 7 nm is expected to be more advanced than others foundries 5 nm or 3 nm, and none of these processes are ready, so expect issues yet to come in all of them. So yes Intel is behind TSMC, but in process node naming marketing not in technological advantage.
Posted on Reply
#25
londiste
R0H1T said:
Well we know that Kirin 980 & A12 are coming out this year IIRC, between them I bet the volumes are enough to outpace & outsell any other semiconductor behemoth, excluding Intel.
Both are small chips on the SOC version of the node. A12 is 83mm² and Kirin 980 is said to be comparable size. Samsung might also want in on the same action as their 7nm is late. Phones with these are expensive pieces of hardware. They can absorb a high cost of the SOC if they want/need to.

On the high performance variant, there is Vega 20 - samples of which were ready in late spring but we do not know much about - and Zen2 has to be already manufacturing if they really will release in 2019H1. I would suspect Nvidia is already manufacturing something next as well or is preparing to.

There is a lot of contention and I honestly suspect the prices might be high whether yields are OK or not.
Posted on Reply
Add your own comment