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.
Kate
Hmm. It all depends from the viewpoint. How about looking at it from this stance?
http://www.siliconinvestor.com/readmsg.aspx?msgid=21458046
K.
Mike, thanks.
6. Sempron demand slowing down.
In addition to last week's comments, this clip (from a PR in the mailbox couple hours ago) might also be a reason.
Based on Intel’s industry leading 90nm process technology, available in the LGA775 package, the Intel Celeron D processor 351 features a 256KB Level 2 cache, a 533 MHz system bus, a processor speed of 3.20 GHz, and support for the Execute Disable Bit.
Intel also announced that it is shipping the Intel Celeron D processors 346, 341, 336, 331 and 326 in the LGA775 package with support for Intel EM64T and the Execute Disable Bit
What a difference a year makes... Who'd thought Intel will have the cheapest available 64-bit part in the market a year ago? :)
K.
doug
Could be. Or because of these considerations http://www.investorshub.com/boards/read_msg.asp?message_id=6730225
K.
Herb
It appears that AMD is announcing chips it can not make in numbers just to create an illusion of having superior technology.
Sure. And to skim the cream. Everybody does (not only) in this industry. Why shouldn't AMD do it when it can?
K.
Keith
In addition, there´s the monthly slip of the "Galaxy" line, now to the end of the year.
I can't remember Andy and others from Sun getting more specific than "2005". Although I definitely have seen any half and probably any quarter of the year as a likely launch timeframe.
K.
Dan
Uh oh - if you're holding Intel based upon that misapprehension, you'd better dump it, quick.
Done. Many thanks for setting this one straight.
K.
Rink - OT
Your moderation-efforts are heartly appreciated, many thanks.
Thing is, I do not think in dimensions of right or wrong, truth or lie. Consequently, I do not care at all who is wrong or right at all.
Actually, the only right I claim for anything i post is the right to be wrong. More specifically, whenever, whoever falsifies my musings, I learn - or at least I try to learn . Which I sense as potentially benefitial opportunity to advance on a learning curve.
Now, frankly, the above is more descriptive of how I would wish me to sense than how i actually feel. I am homo ludens and homo lupus as anybody else is. So I look at it as a bar set high enough to ensure I can walk through under it without the need to bend my head down occasionally.
K.
Dan3
Towards the end of next year is the soonest AMD can achieve a run rate of half the market
Hmm. Let's see. AMD builds one 65nm fab in Dresden. Tooling it for 13K WSPM.(?) Plus contracted additional 65nm capacity at Chartered. Max Capacity of Fab7 there is 7K WSPM.
Intel is building four 65nm fabs.
50/50?
Apart from that, it is hard to see anything but 65nm samples from AMD by end of next year anyway.
K.
Joe
If it is done like this, it's not mine, but Fred and his folk's. It just needed today's ihub-discussion (a welcome contrast to the recent high noise, low signal talks btw) to realize it.
K.
doug
65nm production is due in H106
Yes. Intel's. AMD's is a year later, give or take a quarter.
K.
On second thoughts...
Connecting dies at crossbar level would make an opportunity to make use of MCT-defective dies. Now that would make an interesting DfM detail cute enough to put more faith in the Quadcore-by-early-next-year rumour for me:
A spike adding no diecost at all makes it the better spike than Smithfield.
K.
Doug Good point. K. eom
Joe
all that needs to be done is software transition to catch up with hardware.
Sure. Problem is the costs associated are manifolds of hardware costs. And it cannot be done in just months.
Time to invent some blue crystals...
Inventing Solaris-Janusheads for Power and Risc Code would serve to assist materializing your vision far better. :)
K.
Some comments from the street on Spansion
http://www.businessweek.com/technology/content/jun2005/tc20050623_3958_tc024.htm
There is many more options how to deal with Spansion than just those the article points to, I believe.
K.
smooth2o
Yup. Crystal clear, so far. Why do you stop there? Conclusive remainders of analysis are pretty coercive, aren't they?
K.
buggi
whether this is HIGH Demand from the OEM's or bad supply
(bins) from AMD?
I believe it is neither of these. Rather the same kind of "supply management" Intel does to make sure milking out every dollar of its Centrino-cashcow. AMD needs to make look its premium products tight in retail to take care of prices. And OEMs want to see these high prices in the channel show window to make their systems look priceworthy.
Economic reasoning goes along this line: Real shortages move prices. You'd see dealers offer FX- and other highend models hundred dollars higher if there was, and another hundred if shortage persists. Managed tightness is to control prices. Which is what I see. As soon as you look at it from this viewpoint, you don't suffer anymore from the idea AMD would miss any cream because they can't make enough high end chips.
Actually, AMDs problem in this respect is in the dimension that it takes a lot (of time and effort) to get its traditional clientele used to the idea of shelling out a small fortune for a CPU instead of paying half the Intel-price for the same performance. Intels clientele is used to it, but it is not easy to convert these to AMD. Now this would usually lead into one of the comments about it's about time for AMD to tweak its marketing process. I am explicitely abstaining from this currently due to these concerns:
The fact AMD does not intend to waterfall its prices in summer as per reseller emails carries the message AMD believes the current equilibrium in this respect needs currently no action. Insofar, the question you raised should probably be reworded as "Strong demand or poor supply (in volume, not bins!).
K.
rlweitz
Of course, it's hard to draw conclusions with no scale provided.
I couldn't agree more.
K.
Mike Ah ok. Thanks. K. eom
Mike
Thanks again, to begin with. Looking at the date and comparing it with those from three month ago - which is easily possible with jj's archive (thank you jj) - it seems retail demand is shifting from highend Sempron to Athlon64. Considering overlapping pricepoints it does not come at a surprise, naturally.
Beyond this, I see low-end demand shifting from Sempron to Celeron. If only because your data clearly indicate low end Celerons came down couple percent in price and low end Semprons increased up to 20% in prices during recent three months.
Does it feel similar for you?
K.
p.s: This phenomenon might be considered for NPD techworlds projection of brands for the quarter, shown at Analyst day. Although I guess it looks better than AMD's own projections, otherwise AMD would not have used it.
p.p.s: I like this development. Very helpful for AMD's brand perception, at probably no or very little "cost" in terms of income and dollar-share. AMD will most probably concede some unit-shares with it though.
IBM set to roll Opteron-based blade.
http://www.computerworld.com.au/index.php/id%3B615903325%3Bfp%3B16%3Bfpid%3B0
Link is from niceguy/SI.
I'm crossposting it adressed to you because this link causes firefox to crash on my system. :)
K.
rlweitz
Not at all. I talk to both kind of folks at the vole. Reasonably sane minds on both sides, and they talk to each others as well.
I never really ran into inconsistent communication from Redmont. Occasionally it seemed so, but has always turned out to be more a misunderstanding on my side in talkbacks for clarification.
K.
Tiger64
I just cited what Microsoft said (short after the Release of X64-OEM-OS).
I agree if MS used the term "market penetration" correct, it would need a significant run-rate next year. That's why I said I understood it as it hoped to achieve this.
However, the term is very often (mis)used for incremental market share in a period by marketing spinners. Understood so, I'd call it a reasonable assessment.
K.
rlweitz
Within the next year you won't be able to buy a PC that's not x86-64 compatible. By then most PCs will ship with Windows XP x64.
Huh? Microsoft expects to achieve a market penetration of 4% next year with its X64 client OS. Which I read more as "Microsoft hopes to achieve...
K.
wbmw
Never mind, I did not take it offensive at all. K.
wbmw
I'm just not that fast, you know. :)
K.
mountain dazzie
Many thanks for the update. K. eom
mmoy
My thoughts are exactly along your lines. Btw I have seen systems offered in Germany, availabilty 06/16 iirc. Did not follow though if these are delivered already.
Probably not a bad idea to make sure there are only a handful chips out for a couple of weeks, to prevent from much damage if there is platform issues beyond these known already.
We have seen there is serious issues on the platform side as well for 840EE..
K.
Doug
Thanks. I've seen reports there is board issues. Could be AMD deliberately limits supply for showroom quantities until these are kinked out.
K.
Mike
Sparse availability in Germany.
Erste Retail Athlon64 X2 in Deutschland lieferbar:
4200+ und 4600+ in stock bei alternate:
http://www2.alternate.de/html/shop/productListing4C.html?cat1=003&cat2=223&cat3=000&tree...
4200+ bei http://www.mix-computer.de
Tx amdyou@w:o for the links
There is systems offered already as well.
K.
AG
Your estimates for the current quarter are pretty close to mine and independent market surveys. Digitimes sees 14M mobile CPUs for current quarter, I model 1,25M for AMD. Looking forward, two consecutive quarters of 25% mss gains (from nine to 14) would be pretty impressive. But it is not impossible: It would just need a scenario of market growth continuing to exceed projections, Intel remaining tight on Dothan and/or mobile chipsets - and AMD and its chipset partners have volumes to benefit from it.
K.
Keith
I'm not sure it's the CPU availability. We have seen Turions in notebooks at Lidl-stores (the targa traveller thing) already, however on a fullsize design. I am leaning more towards supply and/or design issues on the platform side currently, in particular with integrated chipsets.
K.
wmbw
I should have expected you smell the rat....
))
K.
wbmw
Any idea why I'm getting this message?
That's an easy one, mate. Slow software on slow hardware. Try firefox on AMD. )
K.
I stand corrected
http://www.siliconinvestor.com/readmsg.aspx?msgid=21412756
Sorry, Marc.
K.
Keith
We have seen two dozen or so offerings from various vendors already, so notebookmakers not publicly talking too much about Turion obviously doesn't mean they don't build with it.
K.
Piper Jaffrey can't compete with MorganStanley's Semiconductor Analysis
"MorganStanley also has approximately a six- to nine-month lead on the learning curve of comprehensive assessments"...
Jules
I did not mean amortization in a fiscal sense. (Taxes are not a prominent problem for AMD anyway. ).
Depreciation period for tools is five years, most facilities of the fab are depreciated over ten years, and the shell over 25 years.
K.
Joe
I believe there is barely a tool left from initial toolset in 1999 today. AMD is currently investing 150M for Fab30's third major toolchange currently. And it expanded cleanroomspace twice, from 90K to 115K squareinch and then from 115 to 150K to make all this possible. (I am impressed how they do all this - as many others in the industry are, awarding the fab frequently.)
Btw, for a moment I thought about spinning upndowns idea ahead to "Saxonian circle", including a third fab plus n+1 utilization of each. But it could not be claimed as a novelty: Intel does it at some of its locations, others do as well. Apart from that, crop rotation is used for ages anyway.
K.
upndown
Thanks for your thoughts. I like the "Saxonian Ping-Pong" analogy. It is not feasible though imo. Playing this game would only allow for amortization periods of three years or so.
(I certainly cannot dispute migrations are possible even in an running production fab - AMD has demonstrated it twice recently. However, it is a relatively costly way of manufucturing high volume parts.)
K.
chipguy
I see your point. Breaking ground for a new 45nm fab by end of next year or so would require another 3B finance. I agree it is doubtful AMD will be able to show enough free cash flow (if at all) next year to find it, even implying subsidies.
But then, Fab36 is planned for a two stage tooling. 13K WSPM first stage, 65nm. 7K additional WSPM second stage is probably intended for 45nm tooling in O8. I believe AMD can stem this, but it is not enough. It would need a second source. IBM comes to mind. Or one of the asian members of the 45nm development consortium with deep pockets.
But at the very end I agree in your assessment. To compete successfully against Intel in the long run is close to hopeless without steadily creating free cash flow to allow building a fab for every node.
K.