Wednesday, December 23rd 2020
Intel Expands 10nm Manufacturing Capacity
In response to incredible customer demand, Intel has doubled its combined 14 nm and 10 nm manufacturing capacity over the past few years. To do this, the company found innovative ways to deliver more output within existing capacity through yield improvement projects and significant investments in capacity expansion. This video recounts that journey, which even included re-purposing existing lab and office space for manufacturing.
"Over the last three years, we have doubled our wafer volume capacity, and that was a significant investment. Moving forward, we're not stopping… We are continuing to invest into factory capacity to ensure we can keep up with the growing needs of our customers," says Keyvan Esfarjani, senior vice president and general manager of Manufacturing and Operations at Intel. The company also ramped its new 10 nm process this year. Intel currently manufactures 10 nm products in high volumes at its Oregon and Arizona sites in the U.S. and its site in Israel.
In 2020, Intel introduced an expanding lineup of 10 nm products including 11th Gen Intel Core processors and the Intel Atom P5900, a system-on-chip for wireless base stations. In addition, the company introduced 10 nm SuperFin technology, which enables the largest single intranode enhancement in Intel's history and delivers performance improvements comparable to a full-node transition.
Esfarjani explains: "10 nm progress is coming along quite well. We have three high-volume manufacturing operations that are going full steam ahead to see how we can do more, better and faster, and continue to support our customers."
NOTE: Intel's capacity expansion program has been a multiyear journey. The factory and office footage in this video was captured prior to Covid-19 safety measures. Intel workers currently working on-site observe appropriate social distancing and mask measures in accordance with internal policies and local requirements.
"Over the last three years, we have doubled our wafer volume capacity, and that was a significant investment. Moving forward, we're not stopping… We are continuing to invest into factory capacity to ensure we can keep up with the growing needs of our customers," says Keyvan Esfarjani, senior vice president and general manager of Manufacturing and Operations at Intel. The company also ramped its new 10 nm process this year. Intel currently manufactures 10 nm products in high volumes at its Oregon and Arizona sites in the U.S. and its site in Israel.
Esfarjani explains: "10 nm progress is coming along quite well. We have three high-volume manufacturing operations that are going full steam ahead to see how we can do more, better and faster, and continue to support our customers."
NOTE: Intel's capacity expansion program has been a multiyear journey. The factory and office footage in this video was captured prior to Covid-19 safety measures. Intel workers currently working on-site observe appropriate social distancing and mask measures in accordance with internal policies and local requirements.
66 Comments on Intel Expands 10nm Manufacturing Capacity
The point is that Intel is no longer in hegemonic position, and the industry dependency on them is decreasing rather quickly.
There're viable (some even say superior) processor alternative with ARM and competitive manufacturing with TSMC / Samsung.
edit: forgot Qualcomm/Microsoft push for Windows on ARM.
edit 2: in this case by mobile i ment laptops, of course x86 and phones is a nonsubject.
.
I focus on desktop, because that's where Intel has been the dominant leader in the past decade. Even that position seems to be shaking with the fast evolution of the Ryzen generations. That's why I think that expanding manufacturing capacity on their 10/14 nm nodes is much rather an attempt to gain market presence short term than a solution to a problem, which is lack of innovation.
The mobile segment is a totally different story.
For myself as example, there's no choice. The tools I use simply aren't there in the Mac ecosystem. In my case I'm talking about things you probably haven't heard of, FactoryTalk Studio, RSLinx, RSLogix. For others, MatLab, IIS, SQL Server, and so on. The software I run on my PC is worth literally 30X what my PC cost. That's not abnormal.
I am not saying there isn't a hardware performance delta where people would start switching, there is. I hyar don't think there is anywhere near enough of that disparity at this point. I think it would need at least 50% and really 100%+.
If you look at the example below, a 4+4 M1 vs the latest 4C TGL-H (35W) which will be the closest x86 competitor, the single core scores are not that much different. Yes the M1 is much better on power usage, but it is also on 5nm TSMC vs the 10nm Intel (Equivalent to 7nm TSMC).
It would also need to be more than just thin-and-light laptop space with max 16GB RAM. That memory number is pretty much useless to me, there's a reason I have 32GB and it's called VMWare and Hyper-V.
Theoretically a 6 core 35-45W TGL-H should hit north of 8000 on multi-core, which would best the M1. The single core scores are less than 7% apart. That would be a more fair comparison 6C vs 4+4.
This is not the type of thing that would drive people to switch operating systems, software, and so on.
There's also an entire ecosystem built around x86. AMD and Intel sell chips. Apple sells computers.
Point being, you probably won't see this <-- link from an Apple partner anytime soon. You're not going to see CAT scanners or MRI scanners using Apple chips, you won't see self-driving cars or trucks (Intel is big into this), nor industrial controllers, nor 5G routing controllers (again, Intel is there).
You'll just see Apple MacOS laptops and desktops. Apple would need to significantly expand the use cases for the M1 line and allow new types of integrations and uses from 3rd parties.
I don't see that happening.
I already said this but let me reiterate it : AMD is worth almost 80 billion more compared to the beginning of the year, Intel is worth roughly 50 billion less. You're opinion that they are still bigger than AMD will ever dream of being is actually laughable.
Intel has shown not only that they can't get their shit together with their new foundries but they didn't even have the incentive to do so. They kept actively prioritizing 14nm capacity more and more when they clearly should have slowly phase it out and introduce some 10nm volume. They chose short term profit at the expense of everything to do with the future and investors finally woke up, hence the 50 billon dollars less bit I mentioned above.
But anyway, it’s irrelevant that they lost 50 billion but the fact that they are still worth twice as much somehow isn’t ? :roll:
Yeah, right. You’re delusional dude. Only keyboard warriors such as yourself think Intel has a chance to ever take the node advantage again, everyone else doesn't and their drop in market value shows that. And their track record isn't great either, their success always seems to have relied one way or another on making their competitors fail rather than making better products and people have caught onto that as well.
There's no need to deny the fact that markets and their participants change. Just look at Via, S3, Matrox and 3DFX for example. No one ever thought they would leave the gaming desktop market when they were big. I for one, was dreaming about being able to afford a 3DFX graphics card one day. Then came nvidia...
For the next two years 14nm intels are still going to be good unless you need 30+ cores/threads from AMD to generate all the polygons for your hentai tits. I might need it.
And this isnt the first time Intel has been behind. On both the 386 and 486 platforms AMD chips offered better gaming performance at a lower price, and on the P5 socket 7 AMD's OG K6 offered 300mhz chips to intels 233 mhz. Both times Intel responded with the P5 and PII, respectively.Erm, not so sure on that one. The Williamette pentium IV launched in november of 2000. It would take until july of 2006 before Intel finally moved away from Sukburst. That's five and a half years. The first Zen chips launched in march of 2017, so we're still not at the 4 year mark for all of zen, and arguably ryzen 2000 was the first that had a distinct advantage over intel, as ryzen 1000 had many performance pitfalls and some stability issues regarding memory speed. Intel still has, what, another 15-18 months before hitting Failburst's epic of woe. If (STRONG IF) rocketlake delivers on the 10-12% IPC gain over comet lake, intel would take back the gaming crown, and we'd be back to the dynamic we had between ryzen 3000 and intel 9000, where intel was for high refresh rate gaming and zen for much everything else.Again, netburst. Intel was way behind AMD in both power use and performance. PPC 970 was coming along nicely, and the apple G4s slammed netburst to the curb in perf/w. There were thoughts that PPC might continue to make enough headroom to take more from intel and x86 in general. We all know how that ended.