Seeking Alpha

Ashraf Eassa

 
View as an RSS Feed
View Ashraf Eassa's Comments BY TICKER:
Latest  |  Highest rated
  • AMD's Conservative Guidance [View article]
    @ Fiberton

    Everybody makes bad calls, and I have certainly made my fair share of them. The important thing is to be able to learn from those mistakes and be able to apply the lessons learned to become a more successful investor in the future.

    Love it or hate it, you can learn a LOT from following AMD.
    Jul 24 04:58 PM | 4 Likes Like |Link to Comment
  • AMD's Conservative Guidance [View article]
    Fiberton

    AMD stock is not only down due to the GPU glut, but the game console guidance for 3Q being peak rather than a continued ramp into 4Q surprised people.
    Jul 24 04:46 PM | 5 Likes Like |Link to Comment
  • AMD's Conservative Guidance [View article]
    I do not think AMD's guidance was conservative in the sense that they are deliberately setting the bar low.

    AMD knew full well what sell-side estimates were, and they are not in the market to buy back stock, so the only "benefit" that they'd get from setting the bar far too low is a lot of unhappy shareholders.
    Jul 24 12:23 PM | 10 Likes Like |Link to Comment
  • Intel: Latent Profitability [View article]
    zisdead

    I think Intel's 14nm node will be extremely impressive. Bill Holt recently went around and did a tour with investors, and one of the things he apparently talked up was an innovative technique called "Reverse Optical Lithography" used to overcome traditional lithography limits for the transistor features.

    While Intel probably will have a minimum metal pitch of 56-58 nanometers (not quite "10nm" in foundry terms, but denser than the 20nm/16nm BEOL processes which use a 64nm M1 pitch), I expect Intel will have ~10nm foundry equivalent *gate* pitch, a metric where Intel has traditionally led.

    As always, time will tell.
    Jul 24 11:22 AM | 2 Likes Like |Link to Comment
  • Intel: Latent Profitability [View article]
    TechResearch

    I would like to add the following. In terms of contacted gate pitch, Intel's 22-nanometer certainly had an edge over TSMC/Samsung 28nm. I believe Intel's CGP was 90nm, while Samsung's 28nm and TSMC's 28nm were ~120nm.

    Metal pitch was closer, with Intel's 22nm M1 coming in at 90nm and M2-M6 (should the designer choose to use that many) at 80nm. This is tighter than what the foundries had at 28nm, but it's not as big a gap.

    Intel's 22nm density at the process level is theoretically a fair bit higher than what TSMC 28nm/Samsung 28nm offered, but of course, density is dependent on a lot more than your process.
    Jul 24 11:18 AM | 3 Likes Like |Link to Comment
  • Intel: Latent Profitability [View article]
    TechResearch

    TSMC detailed 16 FinFET @ IEDM in 2013 already, I believe.
    Jul 24 09:50 AM | 2 Likes Like |Link to Comment
  • Intel: Latent Profitability [View article]
    TechResearch

    "so just wait for the products to speak."

    Intel will be detailing its 14-nanometer process at IDF 2014. It will not be long now before we know exactly how Intel compares with TSMC and Samsung.
    Jul 24 09:24 AM | 6 Likes Like |Link to Comment
  • Facebook beats by $0.10, beats on revenue [View news story]
    Net income more than doubled Y/Y, which should drive down P/E.
    Jul 23 07:55 PM | 1 Like Like |Link to Comment
  • Facebook beats by $0.10, beats on revenue [View news story]
    It's a lot cheaper after this quarter, FYI.
    Jul 23 05:14 PM | 2 Likes Like |Link to Comment
  • Intel: Latent Profitability [View article]
    You're welcome, Cincinnatus.
    Jul 23 05:05 PM | Likes Like |Link to Comment
  • Intel: Latent Profitability [View article]
    Cincinnatus

    Here's the thing you need to know about Intel's processes. Even though they broadly classify them as "SoC" and "General Purpose", each broad product is built with a "customized" process.

    So, for example, Core M will be built on a slightly differently tuned process (i.e. for lower leakage, lower max clocks) than what Broadwell-DT and Broadwell-EP will be built on. Within each process "class" the designers can choose different performance/leakage transistors and the metal stack can be differently configured depending on what kind of cost/density you're looking for.

    It's pretty cool stuff, IMO.
    Jul 23 02:14 PM | 4 Likes Like |Link to Comment
  • Intel: Latent Profitability [View article]
    Cincinnatus

    " They're not trying to manufacture Xeons on a low power SoC process. In fact historically at Intel the low power process has lagged the high performance process."

    Interestingly enough, I believe Broadwell SoC (for microservers) will be out in late 2014/early 2015 -- well ahead of the traditional Xeons. This I would assume is the higher power process, too, but it does show that full SoCs can be built on it, contrary to what some would like us to believe :-)
    Jul 23 02:02 PM | 2 Likes Like |Link to Comment
  • Intel: Latent Profitability [View article]
    Cincinnatus

    "Don't you need to qualify that by whose process is being discussed?"

    Yes, that was a small lapse on my part. Cost/transistor is going DOWN for Intel, but up for the foundries according to Handel Jones as well as a number of the fabless companies (NVIDIA and Broadcom are on record in stating this).
    Jul 23 02:01 PM | 2 Likes Like |Link to Comment
  • Apple rallies post-earnings; Street focuses on margins, iPhone 6 [View news story]
    I don't doubt it!
    Jul 23 11:59 AM | Likes Like |Link to Comment
  • Intel: Latent Profitability [View article]
    Cincinnatus

    It is a common misconception that the leading edge is more expensive. On a per wafer basis, yes, but per transistor it is cheaper thanks to area scaling.

    There is a reason Intel is bringing online a bunch of 14nm factories simultaneously later this year and will transition its highest volume products first (Braswell/Cherry Trail/Broadwell ULX/ULT).
    Jul 23 11:49 AM | 5 Likes Like |Link to Comment
COMMENTS STATS
8,821 Comments
13,788 Likes