Comments Locked

54 Comments

Back to Article

  • shabby - Friday, September 28, 2018 - link

    Very smart, let's invest in old tech, good one intel.
  • nathanddrews - Friday, September 28, 2018 - link

    If it leads to greater profits, then yes, very smart.
  • Zingam - Sunday, September 30, 2018 - link

    How much of this increased demand is Data Centers, etc. buying new CPUs to compensate for lost performance after the Spectre/Meltdown fixes?
  • FunBunny2 - Friday, September 28, 2018 - link

    given The Asymptote of Progress re:miniaturization of chips, it's not only old but nearly the end of the line. 10s of billions of dollars for incremental improvement??? no bean-counter would approve. nor, at this point, an engineer or scientist. the periodic table has only so many entries, and there aren't any new ones on the horizon. at least any that last for more then a microsecond.
  • V900 - Friday, September 28, 2018 - link

    That 14nm node is still going to be printing money for Intel well into the 2020ies if not 2030ies.

    Look at the worldwide market for ICs. Only 40% is currently made on nodes smaller than 40nm.

    And since we're at the end of Moore's law (It was fun while it lasted!) That 14nm node is going to be one of the smallest ones available for a looong time.

    (Intel's 14nm is roughly TSMCs 10nm. What's after that? TSMSs 7nm/Intels 10nm. And then? TSMC is working on a 5nm just like Intel is working on a 7nm node. It'll probably be ready a few years from now.

    And then? Crickets.

    EUV doesn't make smaller nodes possible, it just makes smaller nodes cheaper to make.

    Some scientists are theoretically discussing nanowires and gate-all-around, but its all rather theoretical, with no clear way to massproduction.

    If the financial incentive is there, we might get a final 5nm Intel/3nm TSMC node in a decade or so.
    (Itll most likely be possible, but too expensive for practical production.)

    But unfortunately, below 10nm/7nm we run up against the forces of nature, and things like stochastic errors and phase shifting makes progress increasingly difficult and an economically viable process is almost impossible to achieve.

    TLDR: 14nm will be one of the smallest nodes available, and the smallest node thats economically viable to use for the vast majority of the market. 14nm/10nm will be in use for decades.
  • 0ldman79 - Friday, September 28, 2018 - link

    Well, it's not exactly like it's a bad process.

    We're coming to a point of diminishing returns on lithography improvements.

    That and they've apparently gotten 10% more speed and 10% less power draw by tuning the 14nm process.

    I have to wonder what the 10nm and 7nm will bring to the table. Very interesting times in the tech industry.
  • cyberguyz - Friday, September 28, 2018 - link

    Mor3 cores at higher speeds.
  • V900 - Friday, September 28, 2018 - link

    Yup. 14 nm is just a stop or two from the end of the road.

    (Maaaybe three stops. I could see Intel come up with a 5nm process roughly a decade from now.)

    And whats more important: Its the last stop on the road that's economically viable for the vast majority of the market. It'll be decades before we see the smaller nodes used for anything cheaper than medium-high end phones/laptops etc.
  • nevcairiel - Friday, September 28, 2018 - link

    If they are really smart, they build up processes that can transition from 14nm to 10nm without too much additional cost - as the article also suggests to some degree.
  • V900 - Friday, September 28, 2018 - link

    Very smart indeed!

    Those 14nm machines are going to make Intel money for years. Possibly decades.

    (To give you an idea of how long there will be a demand for the 14nm node: Currently, 60% of the worldwide market for Integrated Circuits are made on fab processes bigger than 40nm. 8% is made on nodes bigger than 180nm. And a whopping 14% of the market is made on nodes between 130nm and 180nm.)

    And the beauty is, that once the initial investment in R&D and machines is paid off, a chip fab is like printing money. The wafer itself costs a few hundred dollars, and each wafer produces hundreds of CPUs.
  • Wilco1 - Sunday, September 30, 2018 - link

    Tell me, which Intel CPUs are larger than 40nm? Intel isn't a foundry, so the only way they can make compete and make money is by making CPUs on the latest node. Increasing 14nm capacity means they have no confidence in 10nm producing decent volumes any time soon.
  • Calin - Thursday, November 15, 2018 - link

    To be fair, quite a bit of the market is made of chips that have so many pins/connections and so little logic that they simply can't be made any smaller (physically) - you can't link a hundred pins to a 14nm, one square millimeter chip - so you link a hundred pins to a 100 square millimeters, 130nm chip.
  • Marlin1975 - Friday, September 28, 2018 - link

    Wow, that's not a good sign in terms of when final 10nm will be coming out. I know they are getting short on 14nm but if the 10nm was coming real soon in mass-production I would think they would not spend more on 14nm.
  • Holliday75 - Friday, September 28, 2018 - link

    Wouldn't it stand to reason if Intel cannot keep up with current demand that customers would turn to other products? If they do this they develop the needed software to take advantage of AMD processors. If they put that investment in its easier for them to buy more down the road.

    I see this as not only Intel making an investment for the now to produce more chips, but also the future. If they can keep companies from going down the AMD road it might keep a lot of customers in their corner for generations to come. Seems like a billion well spent to me.
  • FunBunny2 - Friday, September 28, 2018 - link

    "develop the needed software to take advantage of AMD processors"

    what might that be?? aren't all X86 processors supposed to look the same, above the compilers?
  • cyberguyz - Friday, September 28, 2018 - link

    No. AMD has Numa vs Uma memory handling. Much the same kinds of issue you would find on mainboards with 2 or more sockets. Difference for AMD is that the single chip takes the place of 2 or 4 sockets on their sTR4 platform.
  • HStewart - Friday, September 28, 2018 - link

    I see two things out of this

    1. It is not necessary to change the node to make improvements in the processor and Intel realizes this and we includes such hardware fixes as the Spectra stuff - in fact I believe the 8705g in my Dell XPS 15 2in1 already has Spectra 2 hardware fixes in it.

    2. Intel 10nm must be extremely more dense than other 10nm or even 7nm.

    Also there was a lot of blab blab - on this stuff was going to hurt Intel financials - I think this was just BS and intel is not stupid - they know what they are doing

    In fact Intel is one of major reasons why stocks are pulled out of slide

    https://www.reuters.com/article/usa-stocks/us-stoc...
  • Cliff34 - Friday, September 28, 2018 - link

    It looks like they have to invest in 14nm bc they can't get their 7nm products to the market. This is more of a set back than an investment.

    If AMD can push out a successful 7nm chip, then Intel is in big trouble.
  • nathanddrews - Friday, September 28, 2018 - link

    Even if every AMD order could be satisfied, I don't think their production capacity would be close to the number of chips Intel can produce. Fortunately, it looks like there's plenty of need to go around, so I don't think anyone will be hurting anytime soon.
  • Dragonstongue - Friday, September 28, 2018 - link

    AMD can use TSMC or even Samsung, both of them are absolutely a match if not even more so in terms of overall production capacity for "leading edge" INTC used to be #1 overall production capacity for THEIR chips, but, these days, not so much, counteless troubles meeting their own demand etc whereas AMD who is "fabless" gets to choose whom will produce their chips, ad both TSMC, Samsung, and even GF (though not for 7nm they pulled plug) have been expanding like crazy over the past couple of years.
  • HStewart - Friday, September 28, 2018 - link

    "counteless troubles meeting their own demand"

    That not always a bad thing, this mean that demand for Intel CPU's is a lot more than what people on the web believe.
  • V900 - Friday, September 28, 2018 - link

    And prices are higher!

    Isnt it interesting, that we're seeing the dream scenario for many followers of the Church of AMD: Intels manufacturing failures mean a shortage of chips available.

    In their fevered dreams, this would be the moment where desperate customers decided to give AMDs Zen a chance. Stunned by the performance and price, they promptly swear to only use AMD CPUs from now on. And finally, AMD gets their revenge and vanquishes the Blue giants evil reignof profitable terror.

    Of course instead of that happening, most of Intels customers just shrugged when faced with a shortage.

    And decided to either wait or pay extra for a CPU now, rather than buying from AMD.
  • V900 - Friday, September 28, 2018 - link

    The fact that you think that going from TSMC to Samsung is as easy as you going from Walgreens to Kroger, tells me that you don't really know a whole lot about the topic.

    And no: AMD cant just say: "Welp, Ill just call Samsung and have them make my CPU instead!"

    Aside from the technical difficulty involved, its also important to remember that most of TSMCs 7nm production is already booked a long time into the future. (Apple alone is taking up 70% of it)

    And hey: Neither TSMC or Samsung have actually shipped a high density 7nm chip yet. Latest word is that theyre both delayed.
  • Wilco1 - Sunday, September 30, 2018 - link

    So the iPhone X in the shops with a density of 83 million transistors per mm^2 isn't 7nm?!? Or are you just an Intel fan-boy making stuff up?
  • NirXY - Thursday, October 4, 2018 - link

    The iPhone X uses Apple A11 CPU which is Manufactured on a 10nm
  • shabby - Friday, September 28, 2018 - link

    Intel's 7nm isn't even on the horizon, their 10nm is comparable to others 7nm but they took a gamble in some spots in it and lost.
  • HStewart - Friday, September 28, 2018 - link

    Intel is not out - it not just the node that makes a product success - maybe the gamble was making technology that was significant more advance than other technologies and they did more work. But the node size is only one part.

    Keep in mind CPU's and especially just Desktop Gaming CPU is not the only business they are into.
  • V900 - Friday, September 28, 2018 - link

    How so? Intel and TSMC are roughly neck and neck.

    Intel 10nm=TSMC 7nm. Intel has shipped a high density 10nm chip (though in limited numbers.)

    TSMC haven't even shipped a chip made by their high density 7nm process, though they are shipping a chip made in their low density/low power 7nm process. (The A12 in the iphone)

    Nobody is in the lead here yet, though we will see what the situation looks like in a year.

    And of course, the really tricky node where both TSMC, Samsung and Intel could fail, will be the 5nm node (Intels 7nm)
  • Wilco1 - Sunday, September 30, 2018 - link

    TSMC has been shipping 7nm for quite a while now given that the new iphone is in the shops. And yes this is the densest process in the world, so TSMC quite obviously has taken the lead. Next year they will do a 7nm+ process using EUV which will be even denser (far more dense than Intel 10nm).
  • FunBunny2 - Friday, September 28, 2018 - link

    may be we'll finally see 450mm lines??? that was supposed to happen, what, a decade or two ago?
  • MrSpadge - Monday, October 1, 2018 - link

    That's postponed another decade.
  • rocketbuddha - Friday, September 28, 2018 - link

    We should have the motherboard makers also correspondingly increase the motherboards. There are rumors in Taiwaneese and East Asian publications that the mobo manufacturers are going to pump more and hope that it is true.
    In addition unless the big 3 roll out more of their products for business (laptops, servers) using AMD,
    The below news shows how serious the Intel deficiencies are
    https://semiaccurate.com/2018/09/07/intel-cant-sup...

    I believe the above news would have rattled INTC that they have in public, announced 14nm expansion for Xeons to hold the stock price up...

    If for example DELL (Intel's w**re) puts some document internally similar to that, then I would say "BUY!BUY!BUY!"
  • V900 - Friday, September 28, 2018 - link

    Intels 7nm node is years away, but thats not a problem, because so is their competition: TSMC/Samsungs 5nm.

    AMD making a 7nm CPU isnt a problem, it'll be the equivalent of Intel's 10nm node, and Intel will sort out the yield issues on that node in 2019.

    (And of course, we dont even know if the 7nm Zen will be competitive with Xeons in single thread performance.)

    Its not like any of this is news to Intels customers. They know that the 10nm Xeon will be available around 2020, and AMD suddenly shipping a 7nm Zen won't change squat.

    Nobody will throw a multi year contract worth hundreds of millions of dollars, because AMD releases a 7nm chip a few months faster than expected.

    Its like when you see AMDs faithful predict that Intel is doooomed because AMD made a Zen/Epyc chip that is a few hundred dollars cheaper: NOPE, thats not how these things work.
  • jjj - Friday, September 28, 2018 - link

    The 15B CAPEX for the year was announced months ago when they reported Q2, it's not news even if Intel tries to spin it that way.
  • Mr Perfect - Friday, September 28, 2018 - link

    I wonder if Intel kept desktop core counts stagnant so long partly in an effort to avoid having to invest in greater fab capacity? Eh, they where probably just being lazy without any competition.
  • V900 - Saturday, September 29, 2018 - link

    Nah, CPU core counts were stagnant because they dont make enough of a difference.

    Single thread performance is most important, and for the vast majority of users 2-4 cores is plenty.

    In some cases, a higher core count can hurt performance: See also the i9 CPUs.
  • FunBunny2 - Monday, October 1, 2018 - link

    "Nah, CPU core counts were stagnant because they dont make enough of a difference."

    or, there're precious few embarrassingly parallel problems. running one program/core isn't an answer.
  • DanNeely - Friday, September 28, 2018 - link

    The link to Toms Hardware in the article RE the H310C chipset is broken. Should be;

    https://www.tomshardware.com/news/intel-14nm-short...
  • cyberguyz - Friday, September 28, 2018 - link

    Meanwhile AMD waltzes past them to deliver 7nm tech. I have a feeling AMD will be beating up Intel and taking their lunch money.
  • V900 - Saturday, September 29, 2018 - link

    If "taking their lunch money" = Intel's customers collectively shrugging and deciding that theyd rather wait, rather than trying a AMD/Zen solution, then yes, looks like you're right on the mark!
  • V900 - Saturday, September 29, 2018 - link

    By the time that AMD launches a 7nm Zen, Intel's supply problems will be old news and already over.

    Your comment doesnt make sense.

    Moreover, the problem with all the fevered fantasies from disciples of the Church of AMD, is that you assume that the process and decisionmaking when a corporation places a 100 million dollar order with Intel, is exactly the same as when a 16 year old decides what laptop he will spend his 700$ on.

    And thats not how it works.

    Intels top tier (and even medium tier) customers plan years out in advance. We're talking about multi year contracts for hundreds of millions of dollars, that often involve many other factors than just the CPU.

    Looking at the big picture, AMD releasing a CPU a few months ahead of time, or a brief supply problem on some SKUs means zilch. Nada.
  • HStewart - Friday, September 28, 2018 - link

    This is obvious done to shut down the rumors that Intel is loosing business because this 14nm-10nm thing.

    But these announcements had good effect for Intel

    https://www.marketwatch.com/story/intel-expects-to...
  • flgt - Friday, September 28, 2018 - link

    I love how double digit growth and maxed out fabs is interpreted as the demise of the company. The process geometry doesn’t matter, just if the products are selling.
  • Alistair - Friday, September 28, 2018 - link

    Yeah a shortage is a sign of huge profits coming in. AMD's chips have driven higher demand ironically.
  • Kakti - Friday, September 28, 2018 - link

    You forgot to mention how they're lazy and greedy too. Clearly only the dumbest companies can develop the hardware, software and manufacturing tech to build the fastest processors on the planet. If I believed most of the posters in this comment section, Intel is being run by a bunch of chimpanzees.

    Lucky for us, all of these anti-Intel posters have no idea what they're talking about and live in some fantasy land where Intel isn't constantly improving the bleeding edge of technology while also selling multiples of what AMD dreams of. Intel literally cannot manufacture processors fast enough to meet demand. It's like a restaurant has a line down the block and a month waiting list, yet some kids on a website seem to think it's doing poorly and their underdog diner is just one quarter away from taking the crown. *sigh*
  • V900 - Saturday, September 29, 2018 - link

    The kind of posters that any Intel/CPU story always attracts, remind me of some of the really annoying religious cults.

    Except instead of: "The end of the world is nigh! Repent!"

    We get: "Intel is doooomed! Rejoyce! The thousand year reign of AMD is at hand!

    Every time I click on one of Anandtechs CPU articles, I half expect a tap on the shoulder, and some guy in an ill fitting suit handing me a pamphlet and ask if I have a minute, because he'd like to share the good news of Zen with me.
  • FunBunny2 - Friday, September 28, 2018 - link

    what seems to be missing in all these discussions of Xnm fabs of cpu is that the additional transistors have not been used to improve the guts of the actual cpu, just used to import functions from what have been separate chips. yes, there may be a bit of speed improvement from that closer layout. but is there really much that computer does today that it didn't do a decade ago?? not a whole lot. that pesky Asymptote of Progress, again.
  • Kakti - Friday, September 28, 2018 - link

    Ten years ago was 2008. Technically Nehalem launched in November of 2008, but that was only three processors (MSRP $999, $562 and $284). The majority came a year later. The difference between Netburst and the later i-core architectures is so large that we'll just skip that comparison.

    Nehalem didn't even have an iGPU, something that only came two years later in Westmere. The iGPU has improved tremendously since then, from something only able to render the OS and applications to now being a semi-respectable GPU. There's a reason NVidia doesn't even make xx20 and xx30 cards anymore, and likely won't be making x40 or possibly x50 cards soon.

    Some additional CPU improvements (we're not counting chipsets) include hardware encryption acceleration such as AES-NI, the CLMUL instruction set that improves the speed of encryption, CRC checks, compression algorithms, etc.

    We now have AVX and AVX2, and soon will have AVX512. AVX is used by a huge variety of software, everything from video rendering to web servers, software RAID to emulators.

    Modern processors have TXT and related TPM for secure hardware authentication (useful primarily in business settings).

    TSX is a recently introduced x86 instruction set that provides hardware transactional memory support, speeding up multi-threaded tasks.

    SGX is another instruction set designed to create secure "enclaves" restricting the software to only have access to code in its security box.

    There's plenty of other advancements such as the DMI, support for DDR3 and now DDR4 RAM, PCIe 1.0 to PCIe 3.0, etc. Overall, there's been a ton of improvements to Intel CPU's (and chipsets dramatically expand that) besides just cramming more transistors into a smaller space. Intel has not been resting on their laurels, they haven't been lazy. They invented and brought to market Optane non-volatile RAM. They pioneered regular SSD's when they first came out. I could keep going but the picture should be painted pretty well by now. Sure AMD has the occasional win and I applaud them finding niches that Intel hasn't addressed. But to act like Intel hasn't invented or improved anything in ten years is as laughable as most of the other posts in this comments section.
  • FunBunny2 - Monday, October 1, 2018 - link

    "The iGPU has improved tremendously since then, from something only able to render the OS and applications to now being a semi-respectable GPU. There's a reason NVidia doesn't even make xx20 and xx30 cards anymore, and likely won't be making x40 or possibly x50 cards soon."

    this is just the most obvious support for what I said: using all those billions of transistors to move off-chip functions to on-chip. the user sees little difference, except further monopoly of cpu and may be a tad improvement in performance. if one defines a cpu as whatever function a vendor wants to embed in that die, then I suppose one can say there's been lots of progress. I, for one, just view these changes as wider and deeper moating to protect Intel.
  • Dragonstongue - Friday, September 28, 2018 - link

    IMO is Intel knew all along that they could not get 10nm out in due time, they should have doubled down on making sure their best chipset processors as well as 14nm (whatever version) would be absolutely "up to snuff" the moment they found out how good Ryzen (and the various products based around it) started to launch in 2017, seemed they sat on their a%% instead of making sure the cupboard was full of cookies that were missing 1/2 their icing

    NOW they "race" to meet demand after all the crud they have "messed up" in the past 3 or so years, they need to go back to the drawing board and not make fairy tales, did they not just lose an important person on their team because of this ^.^
  • HStewart - Friday, September 28, 2018 - link

    Where does Intel state they are Prioritizing on High End Core and Xeon - I think this is general and include laptop production.
  • Lord of the Bored - Saturday, September 29, 2018 - link

    I initially read that headline as "Intel invests $1B in Ponzi scheme"
    Such disappointment when I double-checked that read.
  • iwod - Saturday, September 29, 2018 - link

    I wish some one Could do an analysis on this. So Demand is outstripping supply for BOTH Intel and AMD? I often question whether AMD is doing good enough to justify its current stock price.
  • thesavvymage - Sunday, September 30, 2018 - link

    If you question whether AMD is doing enough to justify their stock price, why not put your money on that thought and short the stock?

    I dont personally have an opinion either way, its just hilarious seeing people in this comment section talking about stocks as if 95% of the readers here understand stock pricing.

Log in

Don't have an account? Sign up now