Seeking Alpha
View as an RSS Feed

Justin Jaynes  

View Justin Jaynes' Comments BY TICKER:
Latest  |  Highest rated
  • AMD's Second Half Guidance Seems Unrealistic: Zen Will Not Topple Skylake [View article]
    @Gofx -

    Precisely. Intel does this with their Haswell-E processors now (socket LGA-2011v3), but the socket whatever for i7-4790k chips. Haswell-E is about 315 mm^2 or so, without any iGPU.
    May 17, 2015. 09:44 PM | Likes Like |Link to Comment
  • AMD's Second Half Guidance Seems Unrealistic: Zen Will Not Topple Skylake [View article]
    Binartech - good catch - media outlets are reporting that it's coming to both - hopefully there were no wires crossed -

    "It’s assumed that Zen will also power the 7th-generation of desktop and mobile APUs that are shown here, though AMD only explicitly stated that Zen was coming to the desktop."
    May 17, 2015. 09:31 PM | Likes Like |Link to Comment
  • AMD's Second Half Guidance Seems Unrealistic: Zen Will Not Topple Skylake [View article]

    Don't disagree with any of those points, but I will probably short Intel for a quick swing trade if they get into a price war with AMD, as I would imagine there would be a temporary downswing during earnings compression during the price competition.

    Shrinking earnings would probably lead to some negative headlines and a contraction of the PE for Intel, just imho

    Regarding the HEDT market (core extreme processors) the real sweet spot of the market right now is the non extreme i5/i7 series.

    Intel's LGA-2011v3 socket brought some more mainstream products to market at lower ASPs, which I found surprising how low they were. You could get a bottom end LGA-2011v3 processor (i7-5820k, 6 core/12 thread) processor for ~$400 now, which is way cheaper compared to older prices.

    Regarding the size of the DIY market, last time I saw (these were very old numbers that look like they were posted from a mercury research data report - IDC/Gartner only track OEM shipments as far as I am aware - mercury research tracks OEM + retail channel sales, based on your position in Intel I might look at trying to find those numbers if I were you, but the reports are a bit pricey), the DIY market was around 15M chips per quarter

    IDC had ~75M units, Mercury had ~90M units, meaning about 15M units/quarter via retail channels

    In AMD's hay day, they had almost equal market share with Intel concerning the DIY market, at least according to Passmark (probably not the best source, but it's a good free source as best I can tell)

    If the company can muster a competent solution in Zen, and GloFo can actually build it, even if AMD could move an additional 5M chips a year (which would equate to an additional 10% market share if we assume about 50M chips sold per year in the retail channel), and assuming an ASP of around $200 each, that would be about $1B in added revenue, at likely margin accretive levels compared to the corporate average.
    May 17, 2015. 09:20 PM | Likes Like |Link to Comment
  • AMD's Second Half Guidance Seems Unrealistic: Zen Will Not Topple Skylake [View article]
    @Randsec - when I say smaller, I mean smaller compared to a hypothetical 8 core zen with an added iGPU.

    If you look at those that are buying say a 4790k or 4590k (whatever the i5 is) - most of the time those are getting paired with an discrete GPU, so for the most part (outside of quick synch which I view as more of a fringe case than the mainstream case of gaming), that added iGPU is going largely unused.

    So rather than using the die space for an iGPU, AMD is adding the extra 4 zen cores in place of the iGPU, giving AMD an 8 core, 16 thread CPU (i7-4790k is 4 core/8 thread), so moar corez!! (internet joke/reference, insert doge dog here) without tacking on the iGPU that would probably go unused anyway, at least as far as the HEDT market is concerned.

    I'm saying, specifically, that AMD will save the die area of the iGPU to double the core count, rather than simply disabling the iGPU and selling it as a CPU only, as the athlons are doing now.
    May 17, 2015. 09:01 PM | 1 Like Like |Link to Comment
  • AMD's Second Half Guidance Seems Unrealistic: Zen Will Not Topple Skylake [View article]
    Couldn't edit the post, but to add - in the desktop the performance should be around that of a gtx 960 (maybe a little lower), and in notebooks it would really shine. AMD has no real market share in notebooks to speak of, so that will hit Nvidia (regarding GPUs specifically) the most.
    May 17, 2015. 03:03 PM | Likes Like |Link to Comment
  • AMD's Second Half Guidance Seems Unrealistic: Zen Will Not Topple Skylake [View article]
    @Fib -

    Now imagine that scenario with Intel or AMD having an APU with the performance roughly in line with a GTX 960 or so.
    May 17, 2015. 02:58 PM | Likes Like |Link to Comment
  • AMD's Second Half Guidance Seems Unrealistic: Zen Will Not Topple Skylake [View article]
    GOFX, RSA -

    Dr. Su said Zen would be a desktop product first, so I'm looking more there than Servers.

    If you look at Intel's operating dollars from DCG (or whatever it's called now), those margins are incredibly high, but a lot of that is because the development of the cores is (was, not sure if this changed with intel's restructurings) in the client business.

    I think AMD is more going for the desktop DIY and OEM channel - specifically the desktop DIY community - as this is less heavily influenced by Intel marketing dollars

    Every thing I've seen points to AMD always having considerably higher market share among DIYers - I don't think Passmark is completely accurate, but it just gives an idea.

    Now, AMD also said Zen would be available as a CPU only in desktop - this will decrease die size dramatically compared to Intel to let AMD better compete on price.

    I then think this is where servers come into play - the cores have been developed, largely paid for in client sales, now AMD doesn't really even have market share in servers - like 1% to 2%. AMD's competing on a 32nm process with Intel's 22nm/14nm. A move from 32nm to 14nm is huge - implying a shrink factor of (14/32)^2 = around 20% - meaning a new 14nm opteron would be around 20% the size of the old opterons - as an example, assuming a 32nm opteron is 600 mm^2, one on 14nm would be around 200 mm^2 conservatively (actual number would be 120 mm^2, but I fudged the numbers heavily to be more in line with what would actually be feasible).

    Not to mention the power savings from the move to 14nm. Data center - power consumption is king, and AMD isn't competitive here in any large scale deployments.

    So I look at it like this - AMD will (hopefully) have a product that's much better suited for the desktop and DIYers - and this will also make it a little more competitive among OEMs as well, so client sales will be better - but targeting the DIY desktop market will bypass a lot of Intel's pull since they're sold directly to end consumers.

    This chip will also happen to be better for servers, and since the chips have largely been paid for, there's not really any harm in trying to grow from ~1% server share to get back to 5%+ server share.

    Personally, I think by about 2017/2018 Nvidia will be the casualty of the happenings between Intel and AMD, and specifically if Intel doesn't have to continue to pay Nvidia money. As soon as AMD or Intel put HBM/HMC on a CPU, that will make probably 70% of the consumer dGPU market obsolete. If Intel is no longer required to pay Nvidia any money, Nvidia is going to see much of it's primary market gone and no 100% margin dollars (Intel payments) going to the bottom line.
    May 17, 2015. 12:31 PM | 1 Like Like |Link to Comment
  • AMD's Second Half Guidance Seems Unrealistic: Zen Will Not Topple Skylake [View article]
    I doubt few on here know what you're talking about Matthias
    May 8, 2015. 10:26 PM | Likes Like |Link to Comment
  • Apple: Becoming A Semiconductor Powerhouse [View article]

    Although I don't disagree with your premise, your article seems to certain of itself, and I think you've made similar arguments before.
    Mar 31, 2015. 10:41 AM | Likes Like |Link to Comment
  • BP - Buying This 6% Yielder Is Really A No-Brainer [View article]
    I own both - would like to add to $COP next though as right now I have a little more $BP
    Mar 18, 2015. 08:39 PM | Likes Like |Link to Comment
  • Samsung S6 Will Keep Up Competition With Apple [View article]

    I owned a feature phone until about 3 years ago when my wife washed my pants, which included my old LG filp phone. Since then, my wife and I have both had iPhones, my wife even slightly before mine, and we've never had ours fail. Her iPhone 4S still "feels" faster than my mother's Samsung last generation huge notepad phone (don't know Samsung naming conventions).

    But to your point, I've never seen an iPhone fail - and know *tons* of people with iPhones, unless the phone was dropped or abused. In fairness, I've never seen a samsung fail either, but know far fewer people that use them.

    Youre statistic is ridiculously misleading it seems. 2 failures out of every 5 phones implies a 40% failure rate for an iPhone. Maybe the 30 or 40 people I know that use iPhones are just lucky...
    Mar 17, 2015. 10:37 AM | 27 Likes Like |Link to Comment
  • If history's a guide, AT&T shareholders should celebrate Dow exit [View news story]
    Thinking about adding more $T as well
    Mar 8, 2015. 07:24 PM | Likes Like |Link to Comment
  • AMD's Su Stakes Future On Low Power Carrizo [View article]
    Title seems heavy handed and off. Future is staked on diversification and new chips, not bulldozers final stand
    Feb 27, 2015. 10:29 AM | 10 Likes Like |Link to Comment
  • Bloomberg: Apple's first car could arrive by 2020 [View news story]
    Am I the only one here that finds it weird that we've heard for years about an Apple TV we've never seen? Now we're looking at cars? I'll not hold my breath for a car until I've seen a TV from Apple first.
    Feb 19, 2015. 11:23 PM | 3 Likes Like |Link to Comment
  • AMD Is Poorly Positioned In Tablets And Smartphones [View article]

    "He cites this specific source in his argument. He believes AMD's (NASDAQ:AMD) tablet lineup of CPUs are superior to last-generation Intel (NASDAQ:INTC) parts, using flawed logic. However, his arguments are no longer relevant, because AMD no longer competes with Haswell parts. The Broadwell-Y series is the direct competitor to AMD's Mullins family of SoCs."

    Mullins doesn't compete with Broadwell, or Haswell for that matter. It competes with Bay Trail, and soon cherry trail and the other lower cost Intel chips. Broadwell-Y is $100's more expensive than mullins.
    Feb 19, 2015. 01:55 PM | 5 Likes Like |Link to Comment