Seeking Alpha

nstollon

 
View as an RSS Feed
View nstollon's Comments BY TICKER:
Latest  |  Highest rated
  • Kinder Morgan: Buy The Dip, It Will Not Last [View article]
    An thoughts on the (originally 400+ million) warrants that KMI has outstanding until 2017 as part of the El Paso deal.l I own some, but don't understand all the ramifications. Even if a lot have been taken off the table with buybacks, it is a big chunk of share dilution that comes into play whenever KMI hits $40 (its strike price)
    It seems like this sets up potential for a loop where if any movement above $40 would get corrected downward by increased number of shares, from exercising warrants
    Another factor to make an already convoluted business model even more complex. Thoughts?
    Dec 11, 2014. 03:16 PM | 1 Like Like |Link to Comment
  • My Technology Dividend Portfolio [View article]
    It is always nice to see a bet pay off. Cypress, which I noted was my contrarian technology play, has seen a nice 30% jump in last week or so, based on its announced merger with Spansion. I had in the back of my mind that Cypress could get acquired, but it appears to be the acquirer here. I have rarely seen an acquiring company see this type of uniformly positive feedback.
    Dec 11, 2014. 12:37 PM | Likes Like |Link to Comment
  • Server Acceleration: Altera FPGAs Vs. Nvidia GPUs [View article]
    Yes, $1B TAM over next 5 or 6 years sounds more realistic. $200M/year is nice business, but is not going to to be world changer for either company.
    I think NVDA is better play here, both for reasons I mentioned above, and because a huge portion of the need for acceleration is graphical data becoming a majority of internet traffic. With a 5 year roll out,this is one of those to keep tabs on, rather than jump into; the landscape will change.
    All the players I mentioned above (Broadcom, Marvell, EZChip, PMC) have some products announced to address server acceleration.
    Dec 11, 2014. 12:19 PM | Likes Like |Link to Comment
  • Server Acceleration: Altera FPGAs Vs. Nvidia GPUs [View article]
    OK, I have a quiet morning, so here goes ;)

    FGPA come in different sizes with different features. As a ballpark number, higher end devices in full production might be in the $200-300 range, but they could be more.
    In comparing ASIC and FPGA, due to how the different devices work, the usual rule is 1 ASIC gate = 4 FPGA gates (ie. this the the price you pay pay for field programmability). So. all else being equal, a $300 FPGA would be a $75 ASIC.
    The equivalent ASIC would have better speed and power performance (again, due to limitations and overhead of a FPGA architecture.)

    However the ASIC has one-time costs, that are buried in the FPGA price.
    Assume that the engineering and programming work are the same for a given function in both FPGA and an equivalent ASIC. The major difference in cost is a $10M-20M cost for all the masks and fab costs to get parts built in a foundry and maybe another $10M-20M for all the infrastructure to support a new chip architecture (licenses, tools, support, etc.)
    On cost advantage, it again comes down to volume.
    For 10K devices, a FPGA solution would cost $3M, the ASIC would cost $41M
    For 100K devices, a FPGA solution would cost $30M, the ASIC would cost $48M
    For 1M devices, a FPGA solution would cost $300M, the ASIC would cost $115M
    So, The crossover point to where ASIC is better option is somewhere about 200K devices.
    All these numbers are generalized round number estimates, but you get the idea.

    Now is gets more complicated as both ALTR and XLNX have programs where they will migrate an FPGA to an ASIC for a customer, but I don't know the relative costs or performance on these, although I suspect it will be less optimal than a
    customized ASIC.
    The point on 20nm to 14nm migration is that the mask and fab costs do go up with each generation of shrink. This is the price for being at a cutting edge. It could be argued that power and performance of ASIC at 20nm might still be better than equivalent FPGA at 14 nm, but I have blathered on too much already.

    John Daane is a good guy (I worked with him at LSI Logic, before he moved to ALTR) but his job is marketing ALTR in their best light. There are a lot of applications where 100K is the sweet spot and FPGA is the right solution. I just don't think server acceleration is one of them over the long term.

    Like I said in the beginning, I am long both ALTR an NVDA, both of which I think will be good investments.
    Dec 10, 2014. 11:57 AM | 2 Likes Like |Link to Comment
  • Server Acceleration: Altera FPGAs Vs. Nvidia GPUs [View article]
    I own roughly equal amounts of ALTR and NVDA. so I really wish them both well.

    It is important to understand some key difference between FPGA and GPU and why neither of them may make their expected killing on Server acceleration.

    First off - they are not adversaries, they address different part of the puzzle.
    FPGA are a general purpose programmable device, FPGA are good option for control types of applications to perform a specific function (say packet assembly or header processing in server acceleration). However they are expensive, and have some power and speed issues for complex functions.
    GPU are highly parallel processor architecture, basically hundreds of little processors, they excel on simple and repetitive operations (like compressing and reformating data, which is a lot of what graphics is about). They are a poor choice for packet operations that a FPGA would do well. An FPGA would not be a very cost or power effective way to large graphics operations.

    So why would not one of both winning in server acceleration? Because they are not (cost/power/performance) optimal enough in these applications compared to custom chips that can emerge if the market demands. So Altera and Nvidia are good solutions for 1000 or 10000 servers needing acceleration, but if the numbers are in the millions of servers (and I expect they will be), someone(s) will
    release custom chips to address that specific market with at least equivalent performance and better cost. This scenario has played out several times in past network build-outs, with FPGA being in 1st gen products, but being replaced by NPU (Network processor units) chips soon after. NPU sold by the like of Broadcom, Marvell, EZChip, PMC, etc. (may be better) candidates to claim this server acceleration space (assuming Intel does not just add the features to its next server processor, not unlikely given their recent Axxia acquisition).

    I think one lesson here is you have to be careful on what technical marketing you take to heart. They will tell you their chips can handle all problems and domains, but the evolution of silicon dictates that each major applications spaces will always have their own optimized chips to address it.

    Oh, and by the way, ANN are another example of the above. Prototypes may use FPGA and GPU, but if they ever get to any volume, they will be a different class of chip altogether.
    Dec 10, 2014. 01:21 AM | 1 Like Like |Link to Comment
  • 6 Oil Stocks to Buy in the Next Correction [View article]
    Well, here we are with the next correction.
    Just curious on what you think now, with 3 years of hindsight.
    Dec 9, 2014. 10:16 PM | Likes Like |Link to Comment
  • New Source Energy Partners 23% Yield With Coverage To Spare [View article]
    Thanks for the article. NSLP does seems to be terribly beaten up for a stock that no one seems to have ever heard of. It has given me too much fodder for tax losses, but still seems to be a worthwhile bet.

    Where did you get your hedging and coverage info? I did some investigation in NSLP last month (see my SA article on Nov. 5) and got hedging of 68% in 2015 and 76% for 2016. No surprise for 2015 hedging to increase, but surprised to see the 2016 hedging decrease.

    And yes, NG export in 2016 is one of the upside X-factors no one seems to be considering. There are at least 4 liquification facilities that should be ready in 2016. This could be a game changer in US NG pricing.
    Dec 9, 2014. 11:27 AM | Likes Like |Link to Comment
  • AT&T Vs. Exxon: Dividend Yield Vs. Dividend Growth [View article]
    I have owned both companies for many years.
    I think your comparison has a flaw, that penalizes ATT, in that you are looking at historical comparison of Exxon, which to my immediate recollection has not had any spinoffs to ATT, who in the last 19 years (your historical breakeven compounding period), has spin off Lucent, NCR, MediaOne, ATT Wireless and Comcast, many of whom have in turn either spinoff or merged themselves. I believe I get dividends today from Centurylink and Comcast, at a minimum, that can be traced back to my original ATT shares.
    A reasonable comparison needs to factor these spinoffs into historical comparison of the two, otherwise you are just stacking the deck.
    As a second point, is a 9%+ dividend growth rate really sustainable for a decade or more for a company as large as XOM.
    Dec 5, 2014. 02:01 AM | Likes Like |Link to Comment
  • Linn Energy: Is The Market Pricing In A Distribution Cut? [View article]
    I will
    Dec 3, 2014. 01:33 PM | Likes Like |Link to Comment
  • Linn Energy: Is The Market Pricing In A Distribution Cut? [View article]
    Hopefully we don't find out.
    I have not invested in MEMP, so i am not familiar with it.
    Do you really feel it is that well hedged?
    Dec 3, 2014. 11:26 AM | Likes Like |Link to Comment
  • Linn Energy: Is The Market Pricing In A Distribution Cut? [View article]
    My analysis on the upstream MLP hedges, published as "An Upstream MLP Forward-Looking Model"on SA on 11/5 is that LINE is 79% hedged for 2015 and 74% hedged for 2016. This is 10% higher level of hedging than BBEP, VNR, etc. So all else being equal, LINE will be last upstream MLP standing if low oil pricing drags out.

    If VNR's coverage estimate's are typical (they may not be), sustained $70 is not a disaster as long NG prices can be maintained at $3/80 /mfc. See my SA article "How Vanguard Natural Resources (Almost) Restored My Faith In Upstream MLPs" for specifics. NG prices seem to be holding up so far, so I am sitting tight.
    Dec 3, 2014. 02:21 AM | 1 Like Like |Link to Comment
  • Why Now Is The Time To Sell Texas Instruments [View article]
    Perhaps TI trades at a premium to Qualcomm because it does not have a variety of of international investigations and lawsuits acting as a headwind. (I do own some Qualcomm, so I have faith in them)
    Regards other parts of your argument, can you really say a diversified customer base is a negative. This too far the stretch in logic to say this precludes them from growth.
    Regards insider selling, yes TI has done well, management is cashing in their options. Good for them. If there is negative sentiment, where is the executive exodus that would be expected? With exception of Gregg Lowe moving to be Freescale CEO (a year ago), management seems to be rather content.
    TI does trade at a premium, and it has for most of the last decade. That is the burden to being the best in class company in its space.
    I am long TI and an ex-TIer. No plans to sell any time soon.
    Dec 2, 2014. 11:13 AM | Likes Like |Link to Comment
  • Income Investor Tips For Tax-Loss Harvesting [View article]
    Sadly the energy market this week, has given too many opportunities for harvesting.
    A question - Can I sell stock from my account and buy it in wife's account within 30 days without it being a wash sale?
    Nov 29, 2014. 10:44 AM | Likes Like |Link to Comment
  • Making A Case For CenturyLink [View article]
    Me too...another dividend coming next week (so can't complain ;)
    Its just not the screaming value play that it was at $32 with a 8% dividend. Heck, I can get a better dividend from ATT today. Unless CTL breaks itself up like WIN is doing ( I tend to doubt it), further upside looks to be slow and steady.
    Nov 27, 2014. 12:08 AM | Likes Like |Link to Comment
  • Is Managing A Large Dividend Growth Stock Portfolio Time-Consuming? [View article]
    While there is always the risk to how much fun it is when your avocation becomes your vocation, I have managed an increasing majority my own investment over last 10 years, typically with the intent of (re)investing and never selling, except for tax issues. This provides two benefits, important to me.
    1. I, rather than a fund, can control what my capital gains and losses are for any given year.
    2. Saving on fund investing costs, which are a typical 1% in fees, not even counting the hidden ones.
    For a reasonable asset base, maintaining this control can get yield 5 figures pretty quickly.
    When I get tired of looking at the reports and numbers, the knowledge that I am paying myself a reasonable rate, either in saved fees or reduced taxes, to do this is a reason to stay the course.
    Nov 24, 2014. 11:58 AM | 4 Likes Like |Link to Comment
COMMENTS STATS
231 Comments
183 Likes