Intel's planned comeback: 10nm production now surpassing 14nm, 7nm remains a work in progress

[nanoguy]

Something to look forward to: It looks like Intel is making steady progress on 10nm production and 7nm development. Component and substrate shortages may lead to fewer processors being made in the coming months, but CEO Pat Gelsinger is hopeful the inflection point is coming later this year. Meanwhile, Intel has beat financial Q2 2021 estimates as revenue grew 2% to $18.5 billion.

Intel is optimistic about its 2021 outlook. The company reported second quarter net income of $5.1 billion on revenue of $19.6 billion. This is higher than Wall Street's expectations of $18.5 billion in earnings, and the company is now projecting a stronger non-GAAP revenue of $73.5 billion for 2021.

CEO Pat Gelsinger says the semiconductor industry has seen tremendous growth thanks to the accelerated "digitization of everything," and that this trend will most likely persist for the next ten years as people and companies increasingly opt for hybrid work models.

The pandemic has driven demand for desktop and laptop PCs, and Intel believes its newest CPUs sporting Evo branding have a good chance of capturing a good chunk of the 400 million people who are expected to upgrade their systems in the coming years. On the other hand, the most recent data from IDC indicates demand is already slowing as restrictions are eased.

PC processor sales were up 33 percent year-over-year, which is why Intel's Client Computing Group brought in a record $10.1 billion in revenue in the second quarter, 6 percent more than the same period last year. However, the average selling price of chips decreased as a result of a combination of factors, including high demand, chip shortages, and fierce competition from AMD in the gamer and enthusiast markets. After all, Intel reduced prices for 10th-gen Core processors to sway buyers away from AMD options, but still ended up selling more processors with lower core counts.

Gelsinger is optimistic about the semiconductor shortages and expects them to bottom out in the coming months. He also believes it will take up to two years before the industry is able to "completely catch up with demand."

Intel's Data Center Group saw revenue of $6.5 billion, which is down 9 percent YoY. The Non-volatile Memory Solutions Group brought in $1.1 billion, which is a 34 percent year-over-year decrease. Intel is currently in the process of selling off its NAND memory business to SK Hynix, but until that deal is finalized it still has the potential to stain Intel's financial reports and business projections.

The IoT and Mobileye business units are up YoY. The former is up 47 percent, while the latter has surged 124 percent to bring in a record $327 million in revenue.

Elsewhere, Intel says "7nm is progressing well" and that it will start shipping large orders of Alder Lake chips to its partners in the coming months. The company has already shipped 50 million Tiger Lake processors to date, but doesn't expect to be able to keep cranking them as easily in the third quarter due to an ongoing shortage of components and substrates.

Intel promised it will soon reveal more about its process and packaging roadmaps. Last year, the company had to admit it was falling behind schedule to develop a 7nm process node, but the company promised to fix fab issues and enlisted Pat Gelsinger to ensure it happens sooner rather than later.

Gelsinger established what he calls the "IDM 2.0" strategy, which is aimed to regain the crown of the "unquestioned" industry leader in process technology. So far, Intel has managed to increased its output of 10nm wafers to the point where it now exceeds 14nm wafers. As production of 10nm wafers ramps up, they'll cost 45% less to make than they did a year ago, which is great news for the company's bottom line.

In the coming years, Intel hopes to manufacture chips for other companies under a separate organization called Intel Foundry Services that will report directly to Gelsinger. The company is also building two 7nm fabs in Arizona and upgrading a third in New Mexico to manufacture advanced semiconductor packaging technologies, and we'll no doubt hear more about them at the "Intel Accelerated" event next week on July 26.

Permalink to story.

https://www.techspot.com/news/90539-intel-10nm-production-now-surpassing-14nm-7nm-remains.html

 

[Achaios]

QUOTE
He also believes it will take up to two years before the industry is able to "completely catch up with demand.

However, the average selling price of chips decreased as a result of a combination of factors, including high demand, chip shortages, and fierce competition
UNQUOTE

High demand & chip shortages drop prices now?

 

[Shadowboxer]

Won’t be long now until the 5000 series from AMD faces it’s first real challenger. So far it’s only had to deal with competition using a significantly larger process node.

I just hope that when Intel inevitably decapitates AMD’s 5000 series that AMD bounce back. When Bulldozer launched and flopped 10 years ago it took AMD 7 years to produce a new architecture and it was the users who suffered because of this.

 

[Shadowboxer]

High demand & chip shortages drop prices now?

I think he’s referring to economies of scale. With more and more chips being produced, the average price to make each one will fall.

 

[Dimitriid]

It's interesting how this continued fab issues with intel have actually started to dictate their architecture decisions now: I think that the fact that they can't really squeeze in more density fast enough to compete with AMD (Well technically, TSMC but you get the point) pushed them at least in part to move to their upcoming BIG-little architecture: if you think about it that tech comes from mobile products precisely because they also had constrains but due to size of the product not problems manufacturing it smaller.

 

[wiyosaya]

Hmmm. EVO branding?? I wonder if Samsung will have anything to say about that?

 

[Kosmoz]

I just hope that when Intel inevitably decapitates AMD’s 5000 series that AMD bounce back. When Bulldozer launched and flopped 10 years ago it took AMD 7 years to produce a new architecture and it was the users who suffered because of this.

It will be no decapitation this time, Alder Lake will barely beat Zen3 (maybe not even across the board) and Zen3+(3D) will come after a couple of months only (Q4 2021) and if not outright retake the lead, it will at least equalize the situation.

In 2022 Zen4 then will "decapitate" Alder Lake and only from 2023 onward intel fanbois can expect a better Intel presence or an actual win, although AMD will not stop either.

 

[Adi6293]

Won’t be long now until the 5000 series from AMD faces it’s first real challenger. So far it’s only had to deal with competition using a significantly larger process node.

I just hope that when Intel inevitably decapitates AMD’s 5000 series that AMD bounce back. When Bulldozer launched and flopped 10 years ago it took AMD 7 years to produce a new architecture and it was the users who suffered because of this.

That's what happens when you almost going bankrupt, this time however AMD is in much better position than they were 10 years ago : -).

 

[Lew Zealand]

Won’t be long now until the 5000 series from AMD faces it’s first real challenger. So far it’s only had to deal with competition using a significantly larger process node.

I just hope that when Intel inevitably decapitates AMD’s 5000 series that AMD bounce back. When Bulldozer launched and flopped 10 years ago it took AMD 7 years to produce a new architecture and it was the users who suffered because of this.

It will be no decapitation this time, Alder Lake will barely beat Zen3 (maybe not even across the board) and Zen3+(3D) will come after a couple of months only (Q4 2021) and if not outright retake the lead, it will at least equalize the situation.

In 2022 Zen4 then will "decapitate" Alder Lake and only from 2023 onward intel fanbois can expect a better Intel presence or an actual win, although AMD will not stop either.

Personal narratives (aka: fanfic) are fun!

 

[kiwigraeme]

Won’t be long now until the 5000 series from AMD faces it’s first real challenger. So far it’s only had to deal with competition using a significantly larger process node.

I just hope that when Intel inevitably decapitates AMD’s 5000 series that AMD bounce back. When Bulldozer launched and flopped 10 years ago it took AMD 7 years to produce a new architecture and it was the users who suffered because of this.

Yes it does seem Intel is making good process - will still be interesting to see power envelops .
Plus AMD is not a sitting target - it's got huge gains locked in as well.
I was just wondering the other day - how close we are to maxing silicon - In 10 years our CPUs and SOC will blow the best now away - with better architecture . 3D stacking , using best instruction set for purpose on a chip ( ie a chip could use 3 or 4 types of processers eg arm , risk-V x86 ( or x1024 ?) - 1nm etching or 3D printing - better integration - AI aided design- that's ignoring incorporating organic, quantum, light , carbon or whatever new processers coming online .
Anyway after Intels 10 years of "stagnation",- this decade will be much more exciting - as SOCs are needed in more devices
Intel, Apple , AMD shares etc probably are a safe bet - unless a startup has deep pockets to resist buy out - add in China as a wildcard

 

[Mr Majestyk]

Alderlake looks promising and good to see Intel waking up, although AMD had to drag them kicking and screaming into being competitive. But I won't be rewarding Intel anytime soon. I will be buying AMD CPU's for a very long time. I couldn't less if Alderlake is a bit faster than Zen3, that does not make Zen3 all of a sudden a weak CPU. Zen4 is going to be an even larger performance jump than Zen3 and AMD has their own BIG.little patents as well. Competition is good and it would be even better to have another player. if Qualcomm could give us Apple Mx performance in a SoC things would get very interesting. Too bad Apple are so self absorbed, they should sell their Mx chips to anyone.

 

[captaincranky]

Personal narratives (aka: fanfic) are fun!

In case you haven't noticed, there seems to be an "overpopulation problem" at Techspot these days. Fanboys and SJWs are up at least 30% YoY.

I read somewhere that's Intel's 10 nm chips have a higher transistor density than those of TSMC at 7 nm. But who knows it that's accurate?. I certainly have no vested interest is researching it further.

Although, it it's true, I could be coaxed into giving Intel a little, but very loud, cheer. "Intel, Intel, what's your plan? If you can't do it no one can!"

 

[Kosmoz]

Personal narratives (aka: fanfic) are fun!

Actually all I said is based on leaks from trusted leakers (that were right in the past), but hey if you like the "head in the sand" way, enjoy your beach vacation.

Just remember this "fanfic" in 6 months and 1 year.

Again, these are not my words and my imagination, but leaks.

 

[Kosmoz]

Alderlake looks promising and good to see Intel waking up, although AMD had to drag them kicking and screaming into being competitive. But I won't be rewarding Intel anytime soon. I will be buying AMD CPU's for a very long time. I couldn't less if Alderlake is a bit faster than Zen3, that does not make Zen3 all of a sudden a weak CPU. Zen4 is going to be an even larger performance jump than Zen3 and AMD has their own BIG.little patents as well. Competition is good and it would be even better to have another player. if Qualcomm could give us Apple Mx performance in a SoC things would get very interesting. Too bad Apple are so self absorbed, they should sell their Mx chips to anyone.

Alder Lake with over 200w power draw for me is a big disappointment.
I though finally with these Big.little cores will come efficiency too, but no same old intel with a different coat.

Before Zen4, will come Zen3+ (3D) in Q4 2021. Don't dismiss that to be good also.

 

[HardReset]

I read somewhere that's Intel's 10 nm chips have a higher transistor density than those of TSMC at 7 nm. But who knows it that's accurate?. I certainly have no vested interest is researching it further.

Perhaps. But even if it's true, TSMC is still far ahead. Intel's 7nm supposedly have around same density as TSMC's 5nm has. But TSMC 7nm is already 3 year old and TSMC has been mass producing 5nm for over an year by now. By time Intel finally catches TSMC's 5nm, they already have something better.

Intel is not ahead, not even. Intel is still far behind.

 

[hahahanoobs]

Hmmm. EVO branding?? I wonder if Samsung will have anything to say about that?

Whenever the Intel® Evo™ brand appears, the following footnote must also appear: “Intel, the Intel logo, and Intel Evo are trademarks of Intel Corporation or its subsidiaries.”

 

[hahahanoobs]

It's interesting how this continued fab issues with intel have actually started to dictate their architecture decisions now: I think that the fact that they can't really squeeze in more density fast enough to compete with AMD (Well technically, TSMC but you get the point) pushed them at least in part to move to their upcoming BIG-little architecture: if you think about it that tech comes from mobile products precisely because they also had constrains but due to size of the product not problems manufacturing it smaller.

Intel 10nm has higher transistor density per area than TSMC 7nm.

 

[captaincranky]

Intel is not ahead, not even. Intel is still far behind.

Some days, when I'm tired of hot asphalt, and rap music, I like to imagine myself out in the country, on a farm, maybe there's a barn on the farm... Perhaps it's a barn where Equus asinus are kept.

I haven't quite figured out whether the noise they make is soothing or annoying. Nonetheless, I enjoy provoking it. So I walk into the barn anyway. It's my nature.

 

[itgerald]

It will be good to see Intel bounce back to being the process leader. TSMC has too much on their hands ATM. We need the competition. I had been routing for AMD pre Ryzen days and switched from an i7 6700 to an 8 core Ryzen just because it was a good competitor. I've being staying with AM4 ever since.
Now with Alder Lake up to 16 cores it looks Intel is finally waking back up. Am hoping it will be good enough to compel me to switch back to Intel

 

[HardReset]

Now with Alder Lake up to 16 cores it looks Intel is finally waking back up. Am hoping it will be good enough to compel me to switch back to Intel

Only 8 "strong" and 8 "weak" cores, to be precise. Ryzen offers 16 "strong" cores.

 

[Dimitrios]

Won’t be long now until the 5000 series from AMD faces it’s first real challenger. So far it’s only had to deal with competition using a significantly larger process node.

I just hope that when Intel inevitably decapitates AMD’s 5000 series that AMD bounce back. When Bulldozer launched and flopped 10 years ago it took AMD 7 years to produce a new architecture and it was the users who suffered because of this.

That was because AMD's previous CEO had his head in the sand.

1.) AMD has Lisa Su
2.) AMD has money.
3.) Use this new equation for your answer. Remember this isn't the old AMD. Stop using the old playbook.

 

[Lounds]

9% drop in data center revenue... That's so bad. Honestly watch this space AMD is going to be the data centers number one choice over the next few years due to cost per core.

 

[Mr Majestyk]

Alder Lake with over 200w power draw for me is a big disappointment.
I though finally with these Big.little cores will come efficiency too, but no same old intel with a different coat.

Before Zen4, will come Zen3+ (3D) in Q4 2021. Don't dismiss that to be good also.

It depends in Zen3+ comes to all the range. It may only come to 5900/5950. But I would still wait for Zen4, I'm in no rush to update my Zen1 1700X and would rather wait for AM5 MB's for more future proofing.

 

[pcnthuziast]

No fanboy, but Intel isn't and doesn't need to make any 'comeback'. They are still a world leader in semiconductors and have the profit reports to prove it. If anything they continue to solidify their dominance even when the competition's technology outpaces them at times. When their progress goes stagnant, their marketing and installed base maintain high margins. If their foray into the discrete graphics market is even moderately successful, it will put much needed pressure on both amd and nvidia.

 

[Watzupken]

Won’t be long now until the 5000 series from AMD faces it’s first real challenger. So far it’s only had to deal with competition using a significantly larger process node.

I just hope that when Intel inevitably decapitates AMD’s 5000 series that AMD bounce back. When Bulldozer launched and flopped 10 years ago it took AMD 7 years to produce a new architecture and it was the users who suffered because of this.

In my opinion, it is not just a fab disadvantage that Intel is facing, but also from their exising CPU architecture as well. Evident when you compare an 8 core Rocket Lake vs 8 core Zen 3 CPU. While Rocket Lake would have been more potent with a comparable 10nm, but early 10nm don't clock well (consider the low clockspeed on Ice Lake U), and Intel obviously are not as competitive in multicore performance. It is not until Tiger Lake that managed to close that gap, but with a higher power requirement to push clockspeed.

It is true that AMD will start to feel the heat from Intel this year with Alder Lake, but that is after a year, and it is not like AMD is sitting there waiting. They have already fired a warning shot with the 3D cache earlier as an indication of what is to come with newer AMD processors. Objectively, competition is good because it pushes the companies to work hard. For far too long have Intel been sleeping on the job because people will still buy whatever unexciting products they release year on year. Now both AMD have to work 2x harder, while Intel have to work 3x harder because Intel lost the fab advantage that they enjoy so much in the past. So much so that Intel have to utilize TSMC to try and fend AMD off.