Intel Corporation’s Broadwell-DE Helps Keep the ARMy at Bay

Intel (NASDAQ: INTC  ) has been moving extremely aggressively following the flurry of announcements from ARM (NASDAQ: ARMH  ) licensees claiming that they will be going aggressively after the server processor market. Intel is the clear market leader in server processors, but upstarts like Cavium (NASDAQ: CAVM  ) and server veterans of yore like Advanced Micro Devices (NYSE: AMD  ) are getting more aggressive. Can Intel hold off the assault with market-specific products?

Atom & Xeon mount a pincer attack
Take a look at Intel's public roadmap for server-oriented system-on-chip products, and you'll see that the company has two distinct product lines: one based on the lower power/lower performance Atom and another based on the higher power/higher performance Core processors.

Source: Intel

Now, this road map tells us vague launch time frames, but the point here is that Intel is not going to leave any potential market segment exposed. If the customer wants a low-power system-on-chip with a lot of small cores, then Avoton/Denverton will be the right solution. However, if fewer, more powerful cores integrated in a system-on-chip float the customer's boat, then Broadwell SoC (known as Broadwell-DE) and its follow-ons will be the right product.

What might a Broadwell-DE look like?
Unfortunately, Intel has not given any public information about Broadwell-DE, but import data on Zauba lists a "Broadwell DE 8," which seems to hint that at least one of the variants will sport eight cores. Further, given that the current generation Atom server system-on-chip integrates Ethernet, PCIe, serial ATA, USB, and in some cases a cryptography accelerator, it stands to reason that Broadwell-DE will integrate at least that level of functionality.

But given that Intel is probably building this chip in order to counter the various efforts from the likes of Advanced Micro Devices (NYSE: AMD  ) , it will likely need to include more dedicated offload engines. For example, the Seattle chip from AMD includes a number of dedicated offload engines that handle compression/decompression. From a general purpose CPU perspective, however, an eight core Broadwell should be best in class for the late 2014/2015 time frame.

How will Intel defend its share?
The obvious worry for Intel investors is just how the company will defend its share in the server market with the flurry of competitors coming online. Well, the intuitive answer is that Intel's R&D budget is much larger than that of most of its competitors and it has much more experience in the data center, which in itself should be a fairly large moat. But another major advantage is the company's manufacturing lead.

Note that Intel will be shipping 14-nanometer FinFET server parts in late 2014/early 2015, while competitors such as Cavium and even AMD will likely be stuck on 28-nanometer throughout 2015 and probably at least part-way into 2016. By the time these companies can ramp on 14/16 FinFET from the various foundries, Intel will have likely transitioned to the 10-nanometer node, once again giving it a density and power/performance advantage. This is not to mention that Intel's 14-nanometer process should offer a meaningful density edge over the foundries' processes.

The way to win is ultimately to offer the best performance per total cost of ownership, and Intel has been the clear leader for quite some time. In order for a competitor to take meaningful share from Intel, any competitor will need to demonstrate a fairly substantial edge over Intel's parts. Right now, it's unlikely that a competitor will match what Intel is doing in general purpose servers, let alone exceed.

Foolish bottom line
Intel's place in the data center is very hard won and based on years of very large R&D investments and smart business moves. While nothing is impossible and disruption by a smaller player could happen, it really is important to stand back and see that Intel has executed very well in this arena and continues to do so today. Just like the mobile system-on-chip market, where everybody thought that they could be the next Intel, many of these server chip upstarts are likely to see that trying to go head-to-head with an established, highly profitable market leader is far more difficult than initially believed. 

Leaked: Apple's next smart device (warning, it may shock you)
Apple recently recruited a secret-development "dream team" to guarantee its newest smart device was kept hidden from the public for as long as possible. But the secret is out, and some early viewers are claiming its everyday impact could trump the iPod, iPhone, and the iPad. In fact, ABI Research predicts 485 million of this type of device will be sold per year. But one small company makes Apple's gadget possible. And its stock price has nearly unlimited room to run for early in-the-know investors. To be one of them, and see Apple's newest smart gizmo, just click here!


Read/Post Comments (27) | Recommend This Article (5)

Comments from our Foolish Readers

Help us keep this a respectfully Foolish area! This is a place for our readers to discuss, debate, and learn more about the Foolish investing topic you read about above. Help us keep it clean and safe. If you believe a comment is abusive or otherwise violates our Fool's Rules, please report it via the Report this Comment Report this Comment icon found on every comment.

  • Report this Comment On June 06, 2014, at 9:54 AM, raghu78 wrote:

    Ashraf

    "The way to win is ultimately to offer the best performance per total cost of ownership, and Intel has been the clear leader for quite some time. In order for a competitor to take meaningful share from Intel, any competitor will need to demonstrate a fairly substantial edge over Intel's parts."

    Wrong. In order for a competitor to take share from Intel they have to be competitive in perf and perf/watt and price lower than Intel. But thats only for standard parts.

    The other huge advantage of ARM ecosystem is customization of server silicon. You think everybody wants to buy what Intel sells. The truth is different businesses have different needs and Intel's business model does not lend to solving everyone's problems in the most efficient manner.

    https://gigaom.com/2014/04/28/amazon-joins-other-web-giants-...

    http://www.bloomberg.com/news/2013-12-12/google-said-to-mull...

    Google, Facebook and Amazon are most likely to be designing their own server silicon customized for their needs.

    btw Intel's 14nm server products are launching sometime in 2015. My guess is Q2 for Atom based and Q3 for traditional Xeon EP. ARM's partners will be a year behind with their 16/14 nm products and launch in the Q2-Q3 2016 timeframe. But the point is ARM partners can sell their chips at much lower margins than Intel is used to. ARM Cortex A57 is much more powerful than A15 and most likely Silvermont too. ARM's efficiency is better and so Intel needs the process advantage to remain competitive

    but in 2016 an army of ARMv8 server chips both based on standard ARMv8 cores like A57 and custom ARMv8 cores like AMD K12, Nvidia Denver will hit the market on 16/14 FINFET.

    This process advantage you keep pointing at is shrinking and Intel's competitors like TSMC and Samsung are closing the gap further at future nodes. With Intel's revenue base and the x86 TAM shrinking more revenue is being channeled through the ARM ecosystem to TSMC and Samsung/GF.

    Intel's server marketshare, gross margins and profits are the only casualty in this war between x86 and ARM. Anyway you can keep writing the same kind of articles till this truth is revealed in time.

  • Report this Comment On June 06, 2014, at 9:59 AM, TMFAeassa wrote:

    raghu78

    "This process advantage you keep pointing at is shrinking and Intel's competitors like TSMC and Samsung are closing the gap further at future nodes."

    Only on PowerPoint/marketing slides it is...in the real world, Intel is shipping mass quantities of 22nm FinFET silicon while Qualcomm and others will be brining 20nm planar parts to market during 2015.

  • Report this Comment On June 06, 2014, at 1:23 PM, raghu78 wrote:

    Ashraf

    you always avoid the difficult questions and like a typical fanboy choose to defend Intel. Does Intel's business model lend itself to customization and innovation like ARM's model. Today Intel is forced to allow some amount of customization like in the Rockchip deal. But Intel is doing that because they are a nobody in the mobile market.

    In the server market Intel is not going to allow that until the pressure from ARM ecosystem mounts. This will happen over the next few years but the real pressure will start from 2016.

    As for the point of Intel 22nm vs 20nm planar are you denying that TSMC is in volume production of Apple's A8 at 20nm planar.

  • Report this Comment On June 06, 2014, at 2:09 PM, TMFAeassa wrote:

    raghu78

    "you always avoid the difficult questions and like a typical fanboy choose to defend Intel. "

    Do you really think that leaving comments on each of my articles calling me a "fanboy" or a "pumper" is conducive to intelligent debate? If you would like me to continue to engage in discussion with you, I humbly ask you to cease with the name-calling.

    I simply don't buy the assumptions that you're making, so my conclusions come out differently than yours. I lay my assumptions out clearly and I detail why I am making them, complete with historical precedent. Time will tell who is ultimately right, but I feel that my views well-reasoned and grounded in fact.

    Qualcomm has made its mobile roadmap public and plain: Its 20-nanometer products ramp across 2015, which suggests that 16 FinFET isn't until some point in 2016 from them. At that point point -- if Intel stays on schedule -- it should be very well positioned to counter with 10-nanometer parts, with Atom in particular benefiting from a much larger budget and broader manufacturing focus on SoC products than prior generations have had.

  • Report this Comment On June 06, 2014, at 3:07 PM, smoohta wrote:

    raghu78,

    Intel has been building custom server chips for several big companies for quite some time now (among them are the companies you mentioned above- Facebook, Google and Amazon) and no one is "forcing" them to do it.

    (see for example http://www.wired.com/2013/05/facebook-and-intel/)

    I agree that it is not the same as the ARM model where they could do what they want and build it on their own but saying that these companies have no choice but to do it on their own is plain false.

    That is besides the point that designing/validating/shipping a CPU/SOC is a LOT of work and a huge investment - I'm not saying they couldn't do it but I'm not sure how much it would justify the actual R&D costs if they are the only customer.

    Regarding the process advantage- I guess we'll just have to wait and see.

    No one really knows how TSMC are doing and how are their yields at 20nm.

  • Report this Comment On June 06, 2014, at 3:46 PM, mtechac wrote:

    ARMy..?

    It reminds me of Custer's last stand, where the ARMy higher technology was not of any help from the thousands of incoming ARMed indians..!!

    I almost see Ashraf telling Custer about his enemy gathered INTEL-ligence and telling him that he will defeat the ARMed indians with his higher technology

    .

  • Report this Comment On June 06, 2014, at 4:53 PM, ta152h wrote:

    Ashraf,

    You conveniently, and always, leave out the one big disadvantage Intel has. x86 takes more money and time to develop, and has a cost in power use, size and performance compared to more modern instruction sets like ARM or POWER.

    AMD's K12 will not be hampered by the same limitations Intel has with x86, and I think it will be superior in some ways to what Intel has, although I doubt broadly superior. Two things make me doubt this - AMD has had no success developing higher-end products (Barcelona, Bulldozer), and they also never get the first one right (witness Piledriver's improvement over Bulldozer, on the same process).

    Plus, if you buy into the ARM infrastructure, you aren't stuck with Intel (does AMD even matter in x86 servers? Will they?)

    Another thing to factor in is, and you kind of alluded to it, is what workload are you running. AMD's Steamroller is way better than anything Intel has if your software can use the GPU and HSA. Obviously, such workloads are very, very rare, but I'm pointing out that Intel could remain a good mainstream provider, but there could be workloads where other providers offer better solutions.

    I agree broadly that Intel should be able to remain dominate looking as far into 2018, at least with the Core line. IBM will dominate the high-end, where Intel can not compete, and Intel has no worthwhile low-end design (please, Avoton? Apple can already design a better server part with their A7, if they wanted to move it there), but the middle of the road should still belong to Intel. And that's where, by far, the largest market is. But, I'd also expect Intel's hold on that to loosen a bit, and there to be more workloads where other solutions are competitive, or even better.

    By the way, in fabrication, Samsung is investing twice as much as Intel. This SHOULD close the gap, but it's not always the case. Even so, I wouldn't count on Intel always having the lead they do now, nor would I count on Samsung significantly closing it. We'll have to see how it plays out, and view long term judgments as guesswork, rather than firm.

  • Report this Comment On June 06, 2014, at 5:49 PM, PDM wrote:

    mtechac,

    Bad analogy on your part - those "ARMed Indians" eventually lost the war.

    Seriously, I see that D Bryant will be at the GigaOM Structure Conference later this month to talk about Intel's DCG plans. It will be technical marketing, but there will be information nonetheless.

  • Report this Comment On June 06, 2014, at 7:01 PM, TMFAeassa wrote:

    ta152h

    "By the way, in fabrication, Samsung is investing twice as much as Intel. This SHOULD close the gap, but it's not always the case. Even so, I wouldn't count on Intel always having the lead they do now, nor would I count on Samsung significantly closing it. We'll have to see how it plays out, and view long term judgments as guesswork, rather than firm."

    That is unlikely. Intel is spending about $2.5B/year in process R&D alone. Samsung's entire semiconductor R&D (logic, DRAM, chip design) is under $3 billion.

  • Report this Comment On June 06, 2014, at 7:25 PM, rav55 wrote:

    Intel will lose market share in servers without a doubt. Since Intel's current market share is too high for any future growth it has nowhere to go but down. THere is no growth at 95%!

    This is especially so given the simple fact that Intel can not compete with ARM silicon in the cheap chips category now without massive rebates that are costing them $BILLIONS, the very same performance/pricing/energy efficiency issues wil also exist with ARM server silicon to make them very attractive alternaive to Intel's very expensive and power hungry x86 silicon.

    Apple is also toying with introducing ARM into it's laptop lineup. That means a further erosian of the PC market.

    Intel will lose market share. The only question that remains is how much and how low their stock will drop as a result. Actually a third question remains: how large will the rebates be to OEM's to purchase Intel sevrer silicon. LOL.

    Intel can not keep their position with the dirty tricks they have pulled on AMD here. They have to fight fair and are lined up against too many heavy hitters that want Intel's market.

    They no longer have a monopoly.

    This is going to get bloody Intel margins will suffer.

  • Report this Comment On June 06, 2014, at 7:39 PM, rav55 wrote:

    Ashraf Eassa is the ultimate snake oil salesman. He is also a fanatical believer, a religious fundamentalist bowing to his Great Omniscient Designer: Intel.

    According to Ashraf Intel will do no wrong, crush all competitors, leap tall buildings in a single bound and rescue Inspector Fenwycks daughter.

    You do not have a resoned argument here Ashraf. Your obsessive complusive writing does even Intel investors no good servicee in that your opinions are biased to the point of religious fervor.

    I especially like your point that Intel will win here because they can spend their way out!!!

    So are we going to see Intel rebating server silicon then?

    How is that working out for them in the Tablet market? RTFLOL.

  • Report this Comment On June 06, 2014, at 8:01 PM, rav55 wrote:

    20% of Intel semicinductor sales is budgeted towards R&D. That is the highest in the industry. This is against competition who's R&D budget is now between 8-14% of thier sales.

    http://www.icinsights.com/news/bulletins/Top-10-Semiconducto...

    http://blog.gsmarena.com/samsung-ranks-second-as-top-rd-spen...

    Now Samsungs total Corporate R&D budget is also higher than Intel's but they come in second behind VW. Different market, but, with a reprioritization of funds Samsung becomes a problem for Intel.

    Intel can not spend their way to success here. 22% of sales does not leave much room for fabrication, testing, packaging, margins and OEM REBATES.

  • Report this Comment On June 06, 2014, at 8:28 PM, rav55 wrote:

    "Note that Intel will be shipping 14-nanometer FinFET server parts in late 2014/early 2015, while competitors such as Cavium and even AMD will likely be stuck on 28-nanometer throughout 2015 and probably at least part-way into 2016."

    The above is wishfull thinking. Do you just make this stuff up? This is also very dangerous for Intel investors trying to base their due diligence on your writing Ashraf.

    http://www.cadence.com/cadence/newsroom/features/pages/14nmf...

    http://www.electronicsweekly.com/mannerisms/manufacturing/16...

    http://semiaccurate.com/2014/03/03/6-core-16nm-finfet-arm-co...

    http://www.electronicsweekly.com/news/research/process-rd/ar...

    http://www.marketwatch.com/story/samsungs-14nm-finfet-proces...

    http://wccftech.com/amd-tapes-20nm-14nm-finfet-chips-1h-2014...

    http://www.xbitlabs.com/news/other/display/20131017231002_AM...

    Any edge that Intel may have will be very slim at best.

    All thanks go to Samsung and their license to GloFo!!

  • Report this Comment On June 06, 2014, at 8:36 PM, TEBuddy wrote:

    I don't care if Intel has 10nm production while others are at 16/14nm FinFET. Any decent technology analyst should be able to point out that Intel will lose their biggest advantage from others having FinFET. FinFET is Intel's greatest advantage right now, even if they were two full nodes ahead. The other fabs will be happy to let Intel spend all their money being a process node ahead, which won't mean much of anything in terms of performance, just cost where Intel's will be higher for being the leader anyway.

    And the anologies used are not well thought out, and neither are the author's assumptions. So you think its super hard to break into a market dominated by Intel. Well AMD did it before in a huge way by offering a better solution, and AMD has better solutions today that are going to be entering the data center arena shortly, in X86 and ARM. And amazing how much money Intel has lost trying to break into the mobile SOC market, even though Intel is the leader in manufacturing SOCs? Odd isn't it? Intel is not that great, and they maintain a good amout of market share in areas they lag by paying for it. Intel is very anti-competitive, they would rather stagnate development to line their pockets more, as they did in the past.

  • Report this Comment On June 06, 2014, at 8:38 PM, rav55 wrote:

    @Ashraf

    Here's an example of unbiased reporting:

    http://online.wsj.com/article/PR-CO-20140530-909125.html

    The relevant quotes:

    "What: ARM and the ARM(R) Connected Community are showcasing how the ecosystem's technology is driving innovation across the industry: from sensors to servers, including smartphones, tablets, enterprise infrastructure and the Internet of Things."

    How anyone can utter or read the "Internet of Things" without rolling their eyes is beyond me but no matter. To continue....

    "ARM POP(TM) IP to accelerate SoC design cycles, along with physical IP platforms to support advanced process nodes including 16-nanometer (nm) FinFET and 28HPM"

    ARM is TARGETING the Intel server market. And why not. It's big, it's fat and ready for low capitalization and O&M.

  • Report this Comment On June 06, 2014, at 8:45 PM, rav55 wrote:

    @TEBuddy

    "And the anologies used are not well thought out, and neither are the author's assumptions. So you think its super hard to break into a market dominated by Intel. Well AMD did it before in a huge way by offering a better solution,"

    Well said.

    AMD has a market cap of $3.02 BILLION and Intel's is $137 BILLION. AMD is 2.2% of the size of Intel. Intel has 107,000 employees world wide and AMD has 10,000. Yet AMD has 20+% of the total PC and Laptop market. That's pretty dam good for a company who according to Ashraf is teetering ong the edge of failure. LOL

    Yet lowly little AMD has Intel desperately scared and it drives Asraf absolutely crazy..

    AMD exerts an influence far beyond their size.

    What is Intel going to do when Samsung with a much larger R&D budget puts them in their sights?

  • Report this Comment On June 06, 2014, at 8:51 PM, TMFAeassa wrote:

    @ rav55

    "The above is wishfull thinking. Do you just make this stuff up? This is also very dangerous for Intel investors trying to base their due diligence on your writing Ashraf."

    If Qualcomm (TSMC's biggest and most important customer) isn't getting 20nm parts out until 2015, then this timeline is quite reasonable.

    Riddle me this: if 16 FinFET *and* related designs will be ready to go in 2015, then why isn't Qualcomm going to be there? It's this kind of reasoning that constitutes due diligence.

  • Report this Comment On June 06, 2014, at 8:55 PM, TMFAeassa wrote:

    @ rav55

    "Yet lowly little AMD has Intel desperately scared and it drives Asraf absolutely crazy.."

    "Lowly little AMD" has almost entirely lost the server market and is very rapidly losing market share in PCs (thanks to Bay Trail-M which very profitably leverages the work that Intel is doing in mobile devices). It is nonexistent in tablets, and its biggest money-makers are dGPUs and game consoles, both areas in which Intel does not compete.

    At Computex 2014, I saw a video of a 14nm Broadwell in a tablet reference design slimmer and lighter than an iPad Air. From AMD, I have only seen a bulky Discovery Tablet.

  • Report this Comment On June 06, 2014, at 9:09 PM, TMFAeassa wrote:

    @ rav55

    To be perfectly clear, I actually do agree with you that AMD has been able to do surprisingly well given that it is operating at a pretty substantial R&D disadvantage + process disadvantage.

    If AMD can pull one out of the hat and actually start gaining back some market segment share from Intel, there would be a real investment opportunity there. However, it's just getting tougher and tougher to see this happening in light of the widening gap between the product offerings of the two companies.

    Of course, time will tell and maybe your bullishness on AMD/bearishness on Intel will play out (and if you have a position, then you will gain handsomely). We'll just have to see.

  • Report this Comment On June 06, 2014, at 9:10 PM, TMFAeassa wrote:

    That said, I do hope that AMD can win some designs with Mullins so that it can be tested in the real-world for power/performance :-)

  • Report this Comment On June 06, 2014, at 9:16 PM, rav55 wrote:

    Without a doubt the Intel process timeline will see FinFET litho sometime in 2015.

    But they will not be alone. To think or postulate that the rest of the world and their dog is standing still is a ludicrous dismissal that is typical of the danger your so-called writing presents to the Intel investor trying to glean ANYTHING useful to add to their knowledge base.

    "Riddle me this: if 16 FinFET *and* related designs will be ready to go in 2015, then why isn't Qualcomm going to be there?"

    Somehow it's appropriate that you quote the comic book character the Riddler. His grip on reality was just as tenuous as yours.

    I am not here to answer your questions Ashraf. I don't care about Qualcomm, however if they are not there with FinFet litho in 2015 then others will be. Dare I even mention AMD without you collapsing in a paroxysm of spastic yellow journalism?

    Your are a tedious bore to AMD investors and a bible thumping, proselytising danger to Intel investors.

    Everythng that you write must be fact checked as it is all distorted.

    "reasoning that constitutes due diligence. "

    Surely you jest?

  • Report this Comment On June 06, 2014, at 10:55 PM, raghu78 wrote:

    Ashraf

    "Qualcomm has made its mobile roadmap public and plain: Its 20-nanometer products ramp across 2015, which suggests that 16 FinFET isn't until some point in 2016 from them. At that point point -- if Intel stays on schedule -- it should be very well positioned to counter with 10-nanometer parts, with Atom in particular benefiting from a much larger budget and broader manufacturing focus on SoC products than prior generations have had."

    Qualcomm goes into high volume production at TSMC 20nm for the S810 and S808 in Q4 2014 for a H1 2015 launch. Given TSMC 16 FINFET is a year behind 20nm planar expect the 16 FINFET products to launch in early 2016.

    Intel 10nm is not going to be out anytime before Q2 2017. Intel 22nm -> Intel 14nm has taken 2.5 years. The future nodes are not getting any easier. So 2.5 years from Q4 2014 puts you in Q2 2017 timeframe.

    btw Intel needed 22nm FINFET to compete against ARM at 28nm planar. what happens in 2016 when that advantage is gone. TSMC 16FF+ matches Intel 14nm in performance and TSMC clearly stated that of the 2 FINFET nodes the vast majority of chips ramping at 16 FINFET in 2015 will be at TSMC 16FF+. So Broxton vs Apple A9 vs Qualcomm custom ARMv8 core based Snapdragons. Its not looking good for Intel in 2016.

  • Report this Comment On June 06, 2014, at 10:59 PM, TMFAeassa wrote:

    @ rav55

    "I am not here to answer your questions Ashraf. I don't care about Qualcomm, however if they are not there with FinFet litho in 2015 then others will be. Dare I even mention AMD without you collapsing in a paroxysm of spastic yellow journalism?"

    AMD is planning 20nm APUs in 2015 and FinFETs debut at some point in 2016. This was confirmed by AMD itself at its Core Innovation Update presentation.

    So, if the world's leading fabless chip-maker isn't going to be implementing FinFETs until 2016 (Qualcomm), Broadcom's FinFET networking chip (Vulcan) won't arrive until 2016, and AMD won't be on FinFETs until 2016, then what does this suggest to you?

    Perhaps we will see Xilinx on 16 FinFET since the FPGA players are the first to adopt bleeding edge nodes, and possibly Apple, but really any direct competition to Intel probably won't transition to 16 FinFET until 2016 at which point Intel will transition to 10nm.

    You can hurl insults all you would like and accuse me of "yellow journalism" until you are blue in the face, but you have yet to actually find fault with the reasoning here and instead are choosing to attack the messenger.

    This weakens your argument fairly substantially.

  • Report this Comment On June 06, 2014, at 11:04 PM, TMFAeassa wrote:

    raghu78

    "Qualcomm goes into high volume production at TSMC 20nm for the S810 and S808 in Q4 2014 for a H1 2015 launch. Given TSMC 16 FINFET is a year behind 20nm planar expect the 16 FINFET products to launch in early 2016."

    HVM in Q4 2014 implies availability in Q2 2015, probably on the later side if Snapdragon 805's launch timeframe is anything to go by. I'll buy that.

    Even if TSMC is ready at 16 FinFET from a process perspective to go into HVM in Q4 2015 (which would imply again Q2 2016 availability if you want to get aggressive), there is absolutely no guarantee that Qualcomm will actually get the product out on that cadence.

    It took roughly 15 months to go from 28nm LP -> 28nm HPm for Qualcomm, and 20nm SoC is looking like May 2013 (availability in devices) to April/May 2015 (23-24 months).

    Do you truly believe that Qualcomm will have a FinFET product out (given the yield issues and extreme design complexity on this process) just 12 months after 20 SoC? It's possible, but it really is highly unlikely.

    "Intel 10nm is not going to be out anytime before Q2 2017. Intel 22nm -> Intel 14nm has taken 2.5 years. The future nodes are not getting any easier. So 2.5 years from Q4 2014 puts you in Q2 2017 timeframe."

    Why do you not apply these arguments to Intel's competition?

    "TSMC 16FF+ matches Intel 14nm in performance "

    This is TSMC's claim. How would they know?

  • Report this Comment On June 06, 2014, at 11:07 PM, bugsy1339 wrote:

    @AE

    "and its biggest money-makers are dGPUs and game consoles, both areas in which Intel does not compete."

    exactly - its the pure definition of AMD's strategy - dont deal with the contra's - diversification - ohh and execution (amd getting products out in the market in time with not bull)

    this makes clear how AMD's management is clearly on the right track.

  • Report this Comment On June 07, 2014, at 2:32 AM, HughJeinus wrote:

    You realize that Intel processors have 5X (or more) the number of transistors - right? My point is that Intel MUST move to 14nm aggressively to remain price competitive at the silicon level. What you are touting as a "advantage" is really a disability of the MicroSoft-optimized Intel processor architecture.

  • Report this Comment On June 07, 2014, at 5:49 PM, SSchlesinger wrote:

    What I do know for certain is that Nvidia exposed TSMC a couple of years ago for the pricing levels of 16rm. Even if it was to come out at the same time as Intel it would be according to TSMC at twice the price.

    I am skeptical that 20nm planar can be manufactured in significant yields at a price that the market is willing to pay. I don't hold out much hope that 16nm finfet will see the light of day for ARM. Even if they could make it, who's willing to pay the ARM tax?

Add your comment.

Sponsored Links

Leaked: Apple's Next Smart Device
(Warning, it may shock you)
The secret is out... experts are predicting 458 million of these types of devices will be sold per year. 1 hyper-growth company stands to rake in maximum profit - and it's NOT Apple. Show me Apple's new smart gizmo!

DocumentId: 2983824, ~/Articles/ArticleHandler.aspx, 10/24/2014 3:53:29 AM

Report This Comment

Use this area to report a comment that you believe is in violation of the community guidelines. Our team will review the entry and take any appropriate action.

Sending report...


Advertisement