is...a Libertarian
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.
Street Chatter: Advanced Micro Devices
A longer article, but the conlusion is:
Glancing at the company's Schaeffer's Equity Scorecard ranking (which you can enjoy with all its high/low filter goodness by subscribing to Schaeffer's Gold), we see that AMD garners a 3.5 out of a possible 10. If you wonder why, optimism toward a weak performer is seen as a bearish contrarian indicator. While AMD is in a bit of a rally (on a daily basis), its long-term trends are all on the downward slope, ranking the firm a weak performer.
http://www.schaeffersresearch.com/commentary/observations.aspx?ID=13208&c=obsfeed
It was more than likely a late report and was the MOC trade.
An absolutely fascinating discussion on glueless that has run several days now.
A sun acquisition makes some sense, however it would truly spell the end of SPARC. With Opteron's on the low-end, dual cores in the midrange and Cray/Octigabay at the high-end, there is not much room left for SPARC. Also, the last server maker to buy Cray, SGI, is itself flirting with insovency.
Cray has always been a feast or famine stock, that has had its share of trips through bankruptcy (both Cray's -- one dead and buried, one still on life support). Supercomputer manufacturers are poor investments over the long term. Commodity computers get cheaper and encroach ever further on the territory of supercomputers to the point that supercomputer makers become irrelevant.
The only way cray survives is as a division of a larger company. Though the last company that thought this is itself near bankruptcy.
It is a common oratory technique, usually used by politicians to sway public opinion -- demonizing the object of dislike. For example, it use to be anything "commie" was bad. Now anything "terrorist" is bad. Smooth was trying to demonize anti-Intel behavior by bringing in a controversial term.
It really isn't worth responding to.
As one of the faithful, facts -- let alone Intel's own facts -- don't really matter. That slide is nothing but drivel put out by an AMD spy inside of Intel -- got it. Now stop bothering me with the facts.
@Tenchu I already responded to this BS (#msg-6222958 and #msg-6223596) but it seems you'd rather believe the moon is made of cheese (#msg-6223756).
All you did was put forward your interpretation and proclaim it was correct -- hardly any sort of proof. Doug's interpretation of the tests used is equally valid and I would say that given the context the test were probably SpecIntrate and SpecFPrate, which would be consistent with Intel's own marketing seen here:
http://investorshub.com/boards/read_msg.asp?message_id=6236581
If Intel got it wrong, you better hurry and explain it to them....
Tenchu, if you plead no contest in court, you are found guilty of whatever it is you were charged with -- there are only two verdicts, guilty and not guilty.
Spinning it doesn't change the fact that Intel was found guilty, irrespective of their plea of no contest, of the charges leveled against them by the JFTC.
A 2 GHz, 100W Montecito will always run at at least 2 GHz
and less than or equal to 100W because that is what it is
designed to do, even when running a power virus.
As I said, impressive if true. However, Intel has a history of over-hyping and under-delivering. Let's hope this time out they can put forth a better performance than they did on Itanium or P4 4.0Ghz.
It is a very exciting new technology that is likely locked up across Intel/HP/Idea corp. IP and will probably also appear in next generation Intel x86 processors.
It sounds like they found the silicon equivalent of Ponce de Leon's fountain of youth. Impressive if true.
How do you up the Mhz, double the cores, reduce the TDP all the while remaining 100% compatible with I2? Is it a massive redesign of the chip akin to the P-III to P-4 evolution? Is it some breakthrough process technology? Or is it marketing hype?
I would expect it to dramatically improve upon the price of a Power 5 system.
No doubt true, but I believe the discussion was comparing it to Opteron systems and hence Opteron pricing.
Montecito will make 90 transactions seem quaint by comparison.
The price of a Montecito system certainly won't be quaint by comparison.
wbmw, the problem with your argument is that thinking your not guilty doesn't make it so. If a man who is hungry and poor steals a bag of rice, is he a thief. The law thinks so -- he may not.
Actually, there was speculation on SI that they will be rebranding the discontinued parts Celeron. I don't know if that's correct, but it seems like a viable strategy.
Isn't the competitive chip for this market Geode?
Kate,
I agree with your conclusion, that in the final analysis, only the stock return matters. However, you asked me a question and I answered it.
Kate,
I think what happened was Intel was betting on IPF. As a backup plan, when it looked like IPF was going to be delayed, they started 64-bit development of x86. However, Intel was wary of Osbourning IPF, so they denied their x86-64 development and implementation.
Eventually they were confronted with the fact that x86-64 was here to stay and they needed an implementation. They talked to Microsoft about writing Win64 for their implementation and were told that Microsoft was not going to fragment the market and write another version of x86-64. Intel was forced to adopt the AMD design.
For a company who lambasted AMD for years as nothing more than a copycat, it seems a stretch to believe they planned their entire x86-64 strategy around copying AMD. This does more harm than good as it validates AMD's designs for enterprise customers.
Kate,
Hyopothetical question -- if Microsoft told Intel that they were only going to write one x86-64 operating system, what would Intel do? What could they do?
Here is another link for quad core.
http://www.crn.com/sections/breakingnews/breakingnews.jhtml?articleId=161500875
I presume that Intel must be developing quad core as well, but will this be the P-M derivative or would it be a parallel chip development?
Keith, you may be right. But having read about their planned 15-month ad campaign
http://www.channeltimes.com/channeltimes/jsp/article.jsp?article_id=3341&cat_id=883
I thought they may be able to buy customers.
I know that is why I agree that 7-9M is a conservative number. Still the counter argument is how many consumers buy XP Pro and how long will it take businesses to decide they want to use XPPro64 and then to qualify it and certify it as ready for use?
So, 7-9 million may be the right number, though like I said it sounds low.
May well be as every chip sold except Mobile are projected to be 64-bit compliant by year end. Many will still be sold with Win32 as they are going to consumers, but who knows with the recently announced windows advertising campaign.
Tata, HP to build Linux based supercomputer
The project will feature one of the largest supercomputers in India and will include the rollout of the largest Itanium Linux cluster.
Wednesday, April 20, 2005
NEW DELHI: Tata Institute of Fundamental Research (TIFR) has announced a partnership with Hewlett Packard (HP) to implement high performance computing (HPC) solutions at its Computational Mathematics Laboratory (CML) in Pune. The project will feature one of the largest supercomputers in India and will include the rollout of the largest Itanium Linux cluster.
According to the press release, this implementation will enhance CML's ability to conduct world-class research in the area of computer science and mathematics. It will allow CML to run various complex algorithms with up to a billion variables. The supercomputer includes 78 x 2-way Itanium nodes with a high performance Infiniband backbone.
"TIFR is very pleased to partner with HP for high-end research in the field of computational mathematics," said TIFR head (CML) Dr. Narendra Karmarkar. "These technologies will help us in our research on the fundamental problems in mathematics and optimization techniques. With the implementation of this Itanium cluster, we have reached yet another milestone in technology adoption to help "Scientific Discoveries using world-class algorithmic research in Computational Mathematics," he added.
"We are happy to partner with TIFR to set up one of the most powerful supercomputers in India," said HP India Sales Pvt Ltd enterprise marketing and alliances director Pallab Talukdar. "We're confident that our technologies and this partnership with TIFR will contribute to many breakthroughs in the areas of computer science and mathematics that will have tangible benefits for people both in India and around the world," he added.
http://www.ciol.com/content/news/2005/105042002.asp
I suspect Intel management decided to move to plan B long before the de-emphasis of the netburst speedracer philosophy was disclosed.
That is the single biggest risk to an AMD investment thesis. If Intel has undertaken some sort of "Manhattan project", which they have been able to keep out of the press, it could spell long term hurt for AMD.
How much weight do you give to a super secret project in the semiconductor industry - a project that takes years to develop. I am not real close to this subject, but the only recent example that comes to mind is TMTA -- though people had pretty accurate speculation what they were up to.
kpf, thanks for the kudos. As far a price/performance, production costs are less relevant than sales price. As long as Intel is willing to stay competitive on sales price and they have segments like Mobile that they can milk to maintain margin they can remain price/performance competitive. It is only if AMD can get performance way up, while Intel is saddled with EOLd tech, that they can achieve escape velocity.
It sounds like you think this is not a slam dunk by 2007. Intel does have a great deal of resources and a driven management, so I also believe it is unclear. However, given the time it takes from concept to tapeout, it seems that Intel is trying to circumvent the laws of chip design. As I understand it the process is serial and throwing more money or more engineers at the process won't necessarily make it go any faster.
Joe,
Thanks for the response. While I take your meaning with the following:
... speculate today about what happens in 2007 with virtually 0% data, 100% speculation.
I think you have to engage in some educated speculation to make an investment decision other than for day trading purposes. You really have to speculate (by taking a view on a story situation) if you are looking for a potential home run.
Buy RHAT or NOVL if you think Linux is going to dominate, or TIVO if you think their patents will hold up or RIMM if you think their patents will hold up, etc.
I am trying to decide if AMD falls into this category and that depends on the factuality of the processor story going forward. I was originally convinced that Intel would hold its own, but recent news/developments has me reassessing this conclusion.
Smooth, over time the superior tech/$ wins. Look at Compaq or dell or x86 or Apache or for a developing story look at Linux. AMD has already made inroads to the enterprise which will only grow over time. Without lock-in and at competitive prices Intel is projected to have an weaker product -- so other than "we always buy Intel" why will enterprise customers not opt for AMD (even if the advantage is only 10-15%).
Obviously you have a different view on how the competitive landscape will play out, so we are just going to have to differ on that. I think history will repeat itself yet again -- that is unless Intel's 2007 chip is more competitive than it looks today. Marketing dollars can do a lot, but IBM lost the PC battle.
One question, is Intel's 2007 chip projected to be inferior/superior/indifferent from K9 on a performance basis or a $/performance basis?
I agree with your entry point. I too will look to buy back in in the $10-$11 range (I don't need to call the absolute bottom). I think over the next 24 months there is a lot of upside in the stock, but I also it will test the old lows before that happens. Flash will be in their reported numbers for another two quarters, at least, so there is definitely downside risk.
As many of you know I am out of AMD and have been for a while, though I have been looking at buying back in. I have reviewed AMD's and Intel's roadmaps and I find a deficiency on the Intel side. So my question is does anyone want to take a crack at explaining to me what is Intel's future chip strategy and it's likelyhood of success -- hopefully in a non-partisan fashion.
What I see is that IPF is a niche product (let the flames fly)
P4 is 64 bit and dual core, but is EOLd
P-M derivative will be a replacement, but needs:
64 bit
PNI
SSE-3
Integrated Memory Controller
Intel's version of cHT
Dual Core
All this will happen around 2007 when K9 will be hitting the market. Can P-4 with enough life support keep up for the next two years? Will the P-M on steroids be competitive? Is K9 anything more than K8 with 4 cores? Can Intel take P-M to 4 cores in the 2007 time frame. Will customers who invested in Netburst be pissed at P-M?
I see lots of potholes in the road ahead for Intel. Historically AMD has not had the capacity to be much more than a nuisance, but begining in 2006 they should have sufficient capacity to supply half the market (though what they are going to do with this remains to be seen). As an investment thesis, AMD looks like a good long-term (i.e. 12-24 months) buy. Thoughts?
It is interesting that you fail to mention Intel's serial litigation against AMD for IP violations -- that is until they gave up. That's the nature of the legal system.
Wasn't IBM once the largest vendor (one quarter, though I may be remembering incorrectly) of IPF systems and haven't they always been a top three vendor? This lack of endorsement does not bode well for the multi vendor nature of IPF. IPF will always be a HP chip, because HP has no alternatives, but its viability with other OEM's is apparently being marginalized.
Keith, what exactly is the "commercial client space" in the context in which Rivet was speaking?
sgolds, I agree that when you force a customer to do something they don't want to do they will at a minimum resent it. Customers will evaluate alternatives and this will inevitably lead to defections.
The quote you highlighted is very significant, because it is saying that the decline in PA-RISC is not being made up by an increase in IPF sales. IPF is not only replacing PA-RISC, but also Alpha and MIPS, yet even with these extra sales opportunities and the fact that it runs Linux, Windows, VMS, et al, it still can't make up for the decline in PA-RISC.
I would not be happy if I were an HP investor.
Here is another article with more statistics in it.
http://www.itjungle.com/breaking/bn022405-story01.html
IBM extends lead in server market
IBM continued to march ahead of rivals in 2004 in server sales, a key market in the computing industry, making gains in models using x86 chips and the Unix operating system.
Big Blue's server revenue increased 9.3 percent to $16.1 billion from 2003 to 2004, according to figures released Wednesday by research firm Gartner. That growth outpaced the overall market, which grew 7.2 percent to $49.5 billion.
Dell grew faster, increasing revenue 20 percent to $4.8 billion. That wasn't enough to pass No. 2 Hewlett-Packard, which grew 6.7 percent to $13.4 billion, or No. 3 Sun Microsystems, which shrank 4.0 percent to $5.2 billion.
Servers are powerful networked machines that handle tasks such as storing e-mail, hosting Web sites, making airline seat reservations or recording bank transactions. Comprising a strategically significant market in the computing industry, servers are at the heart of customers' businesses, come with plumper profit margins than many other products and influence purchases of other computing equipment.
IBM has been tops in the market for years. It fended off an attack by Sun, which soared in the late 1990s but fell back afterward. Big Blue is now targeting Dell with higher-end x86 servers and HP's Itanium-based products with its Power5 processors.
"IBM was able to take advantage of the Power5 servers in upgrading its current installed base as well as upselling current (Intel server) users," Gartner analyst Mike McLaughlin said.
IBM showed gains in the Unix server market, which overall declined 2.8 percent to $16.2 billion in 2004. IBM's share increased from 24.3 percent to 26.7 percent at a time when Sun dropped from 32.6 percent to 31.8 percent and HP dropped from 31.9 percent to 30.2 percent.
HP has suffered in Unix as it transfers its systems from its own PA-RISC processor to Intel's Itanium. "The increase of Itanium has not outpaced the decline of PA-RISC," McLaughlin said.
In contrast, dramatic growth took place with servers using x86 processors such as Intel's Xeon and Advanced Micro Devices' Opteron, a market that boomed 16.4 percent to $23.2 billion. HP kept its top ranking with 32.6 percent of the market, but its 14.1 percent growth rate lagged the market. In contrast, No. 2 Dell grew 19.6 percent to capture 20.5 percent of the market and No. 3 IBM grew 22.4 percent for 18 percent of the market, Gartner said.
Dell's gains came chiefly in single- and dual-processor models. In contrast, revenue from its four-processor models shrank 3 percent to $633 million, McLaughlin said.
IBM also led two segments of the server market that grew much faster than the overall market did. Servers running the Linux operating system grew 50.9 percent to $4.9 billion, while blade servers, thin models that slip side by side into a shared chassis, grew 123 percent to $1.21 billion, Gartner said.
In blades, IBM's revenue grew 316 percent to $518 million; HP's grew 78 percent to $437 million; Sun's grew 76 percent to $115 million; and Dell's shrank 48 percent to $33 million.
In Linux, IBM grew 48 percent to $1.5 billion; HP grew 36 percent to $1.25 billion; and Dell grew 44 percent to $748 million.
Sun has launched an effort to spread its Solaris version of the Unix operating system to x86 chips--chiefly AMD's Opteron. The company made some headway with that effort, though it's still a tiny sliver of the overall x86 market, with revenue increasing 299 percent to $59 million.
However, even though Sun promotes Solaris at the expense of Linux, the latter fared better at Sun, with 93 percent growth to $59 million.
AMD pioneered the addition of 64-bit extensions to x86 in 2003 with its Opteron. Intel followed suit halfway through 2004. Despite AMD's earlier arrival, more revenue came from servers using Intel's 64-bit Xeon chips, McLaughlin said: $1.3 billion for Xeon servers compared to $838 million for Opteron servers.
Intel could make more gains soon. It plans to release in coming months new 64-bit Xeon processors, code-named Cranford and Potomac, suitable for four-processor servers, an area where Opteron already is available.
http://news.com.com/IBM+extends+lead+in+server+market/2100-1010_3-5587722.html?tag=nefd.top
IBM dominates dull Q4 server market
IBM continued to dominate a slow-growing worldwide server market during the fourth quarter.
Server vendors moved a total of 1.87m units during the period - 6 per cent more boxes than they shipped in the same quarter one year ago. All told, customers shelled out $14.2bn in the fourth quarter compared to $13.5bn one year earlier. For the full year, the vendors shipped $49.5bn worth of servers on 6.7m units. This compares $46.1bn in revenue on 5.6m units in 2003.
Click Here
IBM showed the best mix of high-end and volume systems, as it held the revenue lead but ended only third in total shipments. IBM shipped $5.1bn in servers during the fourth quarter and was followed by HP with $3.9bn in sales, Sun with $1.4bn, Dell with $1.3bn and Fujitsu with $654m in sales.
Dell enjoyed the strongest growth at 18 per cent. HP also had a strong quarter, growing 9 per cent. IBM grew 6 per cent, and Sun saw revenue shrink 5 per cent.
HP outpaced all vendors in total shipments, moving 1.9m systems. Dell followed with 1.4m shipments, IBM moved 1m, Sun shipped 337,000 and Fujitsu shipped 229,000.
http://www.theregister.co.uk/2005/02/24/server_q4_04/
CombJelly, one example that comes to mind is Quicken. After MS failed to buy them, they came out with their own product. Now many years later, Quicken is still the dominant player.
@wbmw, Intel has been developing enterprise solutions and reference designs for years now.
I'm sure they have, but other than blades which are recent and still a small market, how many of the Tier 1 OEM servers are based on those designs? You don't get the best performance from reference designs, and Intel doesn't employ a battalion of engineers to tweak and improve their reference designs. Further, they don't interact with end users at the same level to be able to timely improve and enhance their offerings. Even Dell, the poster child for using other peoples expertise, does some of their own R&D.
This is a question about CPUs, not platforms
I mentioned this point to highlight the fact that you have to walk before you can run. It's hard to build a platform around an unannounced new server chip, and premature. They can start pursuing the platform after the chip is done, but how long is that going to be? If what you say about P-M is true, how long does it take to make the changes you mentioned and get it to market. Only then will they have engineers and inclination to start designing a platform. I just don't see it happening in 3 years.
wbmw, the problem with your argument is that you are consigning servers to the commodity bin -- no different than oil or wheat or desktop pcs. The Tier 1 OEMs differentiate themselves by the designs and features they put into their servers.
I don't think that Intel has the same level of expertise (or patents) that HP/IBM/Sun do in building the whole server and I don't think they will invest to develop that expertise -- just as you argue they won't in the IPF world (Intel does offer a chipset for IPF).
How long does it take to develop a new chip? Over the next 3 years what is Intel's transition plan? They have announced the EOL for P-4, so it is not their long-term solution. According to the Anand Report, P-M does not scale well for desktop/server use -- so Intel will be busy devoting resources to designing the Xeon replacement (whatever that might be). I can't see them also spending dollars to try and out muscle the Tier 1's in server design.