Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
My AMD prediction:
I think in 2H2015 they will further wind down their consumer and even corp client efforts meaning no more "biggish" cores for servers or clients and just try to make modest improvements on APUs on small cores and sell them where they can. They will continue to fare poorly in this effort due to Intel fighting ARM. Expect fewer direct headcount and more emphasis on disti served customers especially on in emerging markets. "AMD direct sales reps" to OEMs serving the client market will become largely extinct as has their server efforts already. This will be the brunt of the next round of personnel cuts both in sales and marketing as well as engineering.
They will sell off or license their x86 dense server line end products to someone (to raise cash) as they increasingly find it impossible to stay competitive against Intel based OEMs while remaining an Intel customer for the essentials inside their own products. They will likely continue to pursue their x86/ARM dense server line chips but in competition with too many of the ARMy camp chasing a too small market to ever make it profitable while OEMs selling Intel x86-based products compete in the same space.
They will continue minor incremental spins of x86 small cores to improve power efficiency and interfaces to make them easier to use in SOC and embedded designs particularly where compatibility with similar interfaces on ARM can be made. This is "B-team" stuff, and a much smaller B-team at that.
They will compete with NVidia as a licensor of graphics IP to others and as yet another ARM/graphics chip vendor.
They will continue to sell dGPU in competition with NVidia but both will find it difficult to deal with Intel's iGPU in clients and Intel's HPC chips. It will be difficult to determine AMD's profitability of those standalone products until they realign that reporting segment. Expect another reporting segment realignment in 1H2016.
They will continue to sell their x86 small cores as SOC ingredients for custom and embedded designs. I expect Atom SOCs to make this difficult for profitability.
They will be a lot smaller financially and their 2.2B debt will loom a lot larger. Expect a lot of bondholders to bail on their holdings during 2015 at well below par. They will announce a new acceptable minimum cash threshold well below their current $1B to $600M window which they plan to meet by these major cuts.
Note that selling portions of the company in whole or in part thru licensing is likely the only way to raise cash since floating more bonds won't be possible. Slashing budgets is the only other way to slow the bleeding.
gb
Now that AMD has divided their reporting into two segments which I'll term Legacy and New Stuff, I expect to see Legacy lose money perpetually while any meager profits will come from New Stuff. The open question is whether New Stuff will keep the company profitable as a whole; I'm betting not.
The other question is where will they cut future spending (almost certainly in 2H2015) when it is obvious Legacy is never going to be able to compete with Intel in CPU/iGPU and they don't have enough money to compete with NVidia either for the scraps remaining in the dGPU market.
gb
Note that in the "Enterprise" section of their earnings their main offering is dense servers which are all Intel CPU based. Su mentioned that her future offerings will be ARM. I'll bet that all of her profits for several years to come in the dense server "Enterprise" segment will be Intel CPU based.
gb
This is showing the Intel dual core as an ARM device. What am I missing??
gb
If it wasn't for licensing to Synopsis, both GAAP and non-GAAP net income would have been negative.
gb
"AMD restructuring plan, which will be largely implemented in Q4 2014, is expected to:
•Reduce global headcount by 7 percent, largely expected to be completed by the end of Q4 2014;
•Align AMD's real estate footprint with its reduced headcount;
•Result in a restructuring and impairment charge of approximately $57 million in Q4 2014, primarily related to severance, and a restructuring charge of approximately $13 million in 1H 2015, primarily related to real estate actions; ?The company expects to make cash payments related to these actions of approximately $34 million in Q4 2014 and $20 million in 1H 2015;
•Result in operational savings, primarily in operating expenses, of approximately $9 million in Q4 2014 and approximately $85 million in 2015."
So will everyone work 7% (or more given they are losing share...) harder or will they kill some programs?
gb
"AMD Q3 2014 Results
Revenue of $1.43 billion, flat sequentially and a decrease of 2 percent year-over-year
For Q4 2014, AMD expects revenue to decrease 13 percent, plus or minus 3 percent, sequentially. "
So Intel was up 7% with AMD flat in Q3 and Intel is forecasting up 1% compared to AMD's 13% drop.
I suspect this big drop and a resulting conflict with the board about how to fix is what got RR the boot.
gb
I joined Intel in 1982 and retired in 2000. I can't remember Intel trying to enter a market long after it had gotten far ahead of them. The only possible example would be the Unix/server/workstation market which took nearly two decades to overcome and that with the advantage of price. A huge chunk of that success is attributable to the software efforts of the team headed by Richard Wirt.
I don't think referring to Grove/Noyce/Moore is relevant given the disparity of the situation to anything in Intel's history.
I understand what they are doing with contra-revenue if not the accounting details. They're doing what is necessary (if not yet proven sufficient) to penetrate the market.
I think they have a greater than 50/50 shot. I wish them luck.
gb
with the repeated mention of share gains i suspect the reason for the sudden replacement of amd's ceo will become clear on thursday.
gb
while the cases look cool I suspect they will have a bit of trouble getting past fcc emissions without a metal enclosure.
gb
Lisa, Rory and Claflin (BOD chairman) all are ex-IBM.
do the math...
gb
my experience was that the only "great length" the oems went to were protracted negotiations to get the lowest price. the quality of product they received from the non-intel source was not infrequently less than they expected but they "got the lowest price."
institutional memory varied from oem to oem in this respect...some were goldfish.
gb
Before retiring I met with a lot of customers. Never met any with decision making authority who were "anyone but Intel." And of course we didn't win every deal.
The only "anyone but Intel" types that I've run into are typing in their mom's basement on these and similar forums.
gb
With windows and the pc the basic hardware was defined enough that a base OS could be loaded by the user and drivers sought and added. Not sure how android works but appears to require involvement by the OEM to adapt the release to the hardware and "drivers" for eventual download by the user if the carrier can be convinced to provide it.
The much vaunted "diversity" compared to the pc appears to be a hindrance to user satisfaction.
Messy...
gb
AMD / Synopsis deal moves 150 amd R&D engineers to synopsis
http://finance.yahoo.com/news/inplay-briefing-com-055139997.html
Each time I read another press release about AMD's SeaMicro offerings I get a nice chuckle. They continue to be based on Intel server chips and "supports the next-generation AMD Opteron(Piledriver core) processor.
gb
I just built a Haswell i7 3.5ghz system this past spring to which I have three 24" displayport monitors attached and a bunch of USB stuff.
If Broadwell can get me into a fanless NUC formfactor with a 256G or better SSD at the same performance level I'd spring for it.
The Haswell box is nothing but airspace and I'd love to get it off the floor.
gb
IMNSHO, big.little is what you do when you can't figure out how to do it right, perhaps without stepping on a patent bomb...
gb
or you can use the ignore feature on a few select individuals and be done with their crap. works well here!
gb
@mmoy:
What's your opinion on this?
http://www.tomsguide.com/us/sms-audio-biosport-headphones,news-19339.html
gb
Grove called Ken in the early 80's complaining that the vax kept crashing while taping out the 286.
Turns out the Vax had a memory refresh problem that was masked by all the swapping traffic.
The 286 tapeout was memory resident and the refresh was inadequate.
gb
Anything can be pointlessly argued.
gb
@Ideal: Fischer fantasies are far removed from reality.
gb
a billion dollar pork chop around the kid's neck and the dog still won't play with him!
gb
We have two powerful desktops of recent vintage, a pre-SB Win7 laptop, a Win8.1 IB laptop and a SB HTPC.
We also have a iPad air for my wife purchased last December.
My wife uses it rather than her phone for reading, browsing and playing games but very seldom for anything else.
Neither my daughter and I see any use for ourselves of a tablet of any flavor at this time. I doubt this one would be upgraded unless it was physically damaged or otherwise died.
gb
Apple has obviously done very well by staying away from discussions of process technology and simply talking about the features and benefits of the end product to the user.
I don't think they'll stray from that approach.
gb
Did he work for Goldman Sachs?
gb
For every seller there's a buyer. I wonder who is buying with AMD down 19% now on 2x ADV in the first 30 minutes?
gb
No gratuitous happy words this call.
gb
I listened to AMDs call and didn't hear any mention of servers.
gb
Custom silicon for Oracle.
http://finance.yahoo.com/news/intel-creates-custom-silicon-works-184746791.html?.tsrc=applewf
With this new way to lock in customers how long will Oracle continue to spend $$ on Sparc?
gb
New super computer at Los Alamos info. Note "lack of choice" paragraph. If Phi executes well then the "lack of choice" will likely become permanent IMO.
http://www.hpcwire.com/2014/07/10/los-alamos-lead-shares-trinity-feeds-speeds/
gb
re: amd cheap junk
“There is scarcely anything in the world that some man cannot make a little worse, and sell a little more cheaply. The person who buys on price alone is this man's lawful prey.”
-- John Ruskin
There's actually some question whether JR actually said this but it's a great sentiment anyway to which I subscribe.
gb
Don't forget optical interconnect and a big ease of use in terms of programming model and tools including optimization tools.
gb
I wonder what Goldman Sachs will do now.
gb
WT: So where do you see low hanging fruit?
You can only put so many GPUs in a box in a big room and that trick has played out while maxing out power delivery and cooling. No recent major GPU computation advances have occurred as the major players there have hit the process technology wall and are now spending time pasting new labels on existing product. I fully expect that process technology wall to continue despite the pronouncements of GF and TSMC which are largely relabeling tricks of their own.
No where along the way has anyone come up with a way to get the GPU code created quicker that maximizes the performance to anywhere near peak performance that is frequently quoted. No new hardware, no new software.
BTW, it's not at all clear that the monster machines make money for the GPU providers commensurate with the fanfare they generate. And neither AMD nor NVidia can afford missteps at this point in their financial life with both ARM and Intel bearing down on them in a variety of other market segments. So no monster investments in monster HPC.
So with no new GPUs and no new programming tricks/tools the only thing left is to try what Intel is planning to deliver.
Perhaps the curve will tick up again. Perhaps not.
Do you have an explanation?
gb
Perhaps most of the low hanging fruit has been plucked given the silicon tech available. If that is the case Intel should be in a good position with 14nm silicon, stacked memory, a fabric and optical interconnect.
The GPU alternatives haven't been made any easier to program and are lacking most of the above.
gb
Proof is difficult. You can prove something is broken by a single example. The opposite is not true.
gb
Although specific yield numbers are never given I would expect questions about 14nm to be asked and answered on the Q2 call.
gb
AMD vows to be in business in 2020. Or something like that....
http://m.us.wsj.com/articles/BL-DGB-35943
gb