According to the Inquirer who aren't necessarily the gold standard in journalism, rumour is AMD will miss the quarter by 300-400 million $s:
http://www.theinquirer.net/default.aspx?article=38487
While they are right that Intel is re-gaining sever share, they are fantasizing how miraculously Barcelona combined with AMD's new graphics/chipsets are going to magically recover their Q3/Q4. I would not be surprised if this is yet again fuzz being thrown around by AMD to continue to make Wall St and investors believe they have things under control. Probably in a moment of laxity having slipped out from under the watchful eyes of his PR and investor relations folks, Hector himself admitted Barcelona would not impact their financials in 2007. But apparently the Inquirer knows more and AMD has magic bullets which will allow them to deliver large volumes of Barcelona.
Here's my take. Yes...AMD is working frantically on getting Barcelona right and delivering as much volume as they can in 2007. Yes...Barcelona will give AMD a (slight) performance edge over Intel - but probably not as big as they are walking around claiming. The net result is AMD may recover some server share if they can get any reasonable volume out. But once Penryn comes in, it's going to be a level playing field and Intel will use pricing and capacity to negate AMD's performance advantage until they can launch Nehalem.
To re-cap - AMD will become a sub 10$ stock very shortly after the earnings announcement. Intel will regain market share in Q1 while bringing in the quarter either in line or above their forecast.
Subscribe to:
Post Comments (Atom)
20 comments:
Whether or not AMD becomes a <$10 stock is dependent on whether they issue more stock to raise capital this year.
The stock is not going to drop 26% as a lot of the Q1 bad news is priced in, although not all of the latest potential revenue shortfall. Remember, AMD has already pre-warned a while back (late Feb?) that they wouldn't hit quarterly estimates.
That said folks looking to make a "killing" by buying AMD stock in Q2 just before Barcelona launch - the stock market reacts to revenue/earnings not product announcements (unless you are in biotech).
Barcelona won't materially impact AMD's revenues until 2008 so you will need to buy and hold for some time - and hope Intel doesn't further drive down prices to limit Barcelona's market penetration. There may be a very short term pickup on stock price but that won't be sustained until the financials improve (2008 best case).
I have no idea what the Inq is talking about - but then again they just "broke" a story about AMD's standing in overall IC revenue - this data has been out and reported for what 2 weeks now? Is Inq required to post a favorable AMD article for every unfavorbale one?
barcelona is the only thing holding the stock above $10, imo. it should go down to $11 in the next 2 quarters as AMD shows gigantic loses. A poor Barcelona benchmark will drive it down to $3 just like before. A good Barcelona benchmark will probably keep the stock afloat until investors slowly realized K10 isn't enough to fend off Intel's aggressive execution.
R600 delayed to end of Spring. K10 delayed to end of Summer. Penryn brought up to late Summer. The shrill of annoying AMD fanboys grinds to a halt as the competition engulfs. 3.6GHz Harpertown in 2007 and 4GHz Bloomfield in 2008 stomp all over Fusion. AMD will soon be worth less than the ex-ATI.
Roborat, I'm an Intel fan, but you're $3 prediction with a Barcelona bomb just isn't realistic from a financial perspective. That would put AMD's price/book ratio so low (significantly <1) that any large private equity firm would be better off buying out the company and liquifying the assets... AMD's price/book is extremely low (1.28) at TODAY's stock price (which is why you keep hear the occasional buyout rumor). Do you realistically think it could fall by a factor of 4 without someone stepping in and buying this company?
Now this price prediction may be more realistic if AMD does indeed double the common stock to raise capital, but barring that AMD would likely be bought out well before it got to $3 (unless FTC/SEC stepped in).
To put this in perspective AMD now has ~0.75 Bil shares (not sure about preferred stock), at $3/share this would value the company at ~$2.25 Bil. AMD's book value (yahoo finance) is ~$10.5Bil. Now even if their current $1.5Bil of cash on hand gets fully depleted that still leaves book value at ~$9Bil. With <1Bil shares (rounding to make the math easy), any price below $9 would leave AMD in danger of a buyout or hostile takeover.
I think $10-$12 is a more realistic bottom ($6-$8 if AMD does a stock issuance). That's the downside - if Barcelona is even only marginally better than K8, I think AMD could creep back to the $15 levels toward endQ4/early 2008 as they will be forced to give up the price war and (I think) would cede market share back to Intel for better margin/profitability. Of course if Barcelona is as good as advertised AMD has some significant stock upside in 2008 as that would potentially allow them to retain market share AND raise their ASP's.
"Penryn's just a shrink of Core2 it will not yield that much difference."
"Intel will not be able to release anything above 3GHz until 2008"
"Intel will not have IMC/CSI until 2009"
"When fusion comes out AMD will reinvent the market and put Intel in catchup/copy AMD mode"
If I had a nickel for every time I heard one of these AMD fanboy statements recently, I would actually be able to by a fair amount of AMD stock...or better yet short it for the short term! Perhaps now the random fanboy SPECULATION, with no supporting data will stop.
On a serious note - AMD needs to re-think this "30% market share at all cost" strategy so the "monopoly is broken". There are a few flaws I see with sustaining the current strategy of slashing prices ridiculously to maintain/gain share until hopefully K10 comes in and replaces that share at higher prices.
1) The average consumer doesn't generally upgrade the chip, especially in the notebook and desktop arena, so a short term win with low end K8 parts doesn't necessarily lead to a K10 replacement in 2-3 years. Also the upgrade cycle is pretty long. It;s not like people buying these K8 chips now will be upgrading to K10 inside of a year! (except maybe enthusiasts)
2) K10, especially in the desktop arena will not appreciably ramp in 2007. These means they will need to sustain the ASP hits likely into 2008, with the potential exception of the server area. AMD themselves have said they expect limited revenue impact in 2007.
3) Slashing prices will eventually have some negative branding effect - if sustained for a long enough time, AMD will be synonymous with the cheap solution or the low end solution.
4) The earnings/revenue hits will slow down AMD's expansion (or prevent them from trying to pull equipment in/furthering accelerating their ramps). It will also cause AMD to spend effort trying to save nickels in the short term (see: AMD delays Austin consolidation until 2008 articles - oh wait, that was to prevent distractions to Barcelona launch - my bad!)
It seems like AMD's choice right now is higher unit volume and little/no earnings or raise prices, give back some share but get better ASP's, GPM and earnings. Given the need to continue to integrate ATI, convert F30 and push forward on K10+ plans (fusion, ALU, etc..), I would think giving back a bit of share if they are able to raise prices would be the smart decision.
Ruiz seems to think that unit share is the key for the long term (and he may be right). Of course, as AMD is a business, the board may tire of this strategy and decide to replace him - it's not like Ruiz owns the company. Other scenario is if the stock falls enough AMD may get bought and Ruiz may be replaced that way.
Thoughts?
"If I had a nickel for every time I heard one of these AMD fanboy statements recently, I would actually be able to by a fair amount of AMD stock...or better yet short it for the short term! Perhaps now the random fanboy SPECULATION, with no supporting data will stop."
The truth will set them free. The truth will soon unravel;)
Anonymous intel fan:
You're forgetting about AMD's massive debt and depleting cash on hand after at least 3 quarters of significant loses.
I was always told that book value of a semiconductor company with Fabs/Assembly site doesn’t directly correlate to cash. It is a lot different from say buying a car factory. Liquidation isn’t a straightforward option.
There are big risks involved in acquiring a running Fab which is why you never ever seen one happen when it is in peak production. It’s crazy to buy a Fab when tools are depreciating, lots are running, people still working while at the same time not being profitable. That is like jumping in and paying for the quarterly losses yourself.
Typically buyers wait for the Fab to go under and then buy it at 1/10th the cost. The best time to buy a Fab is when it’s producing nothing and only a skeleton force is left.
AMD has gone below $5 just recently which tells you that its book value doesn’t really mean anything.
"Penryn's just a shrink of Core2 it will not yield that much difference."
"Intel will not be able to release anything above 3GHz until 2008"
"Intel will not have IMC/CSI until 2009"
"When fusion comes out AMD will reinvent the market and put Intel in catchup/copy AMD mode"
the thing is that this fanboy would claim that himself is not a fanboy and would use his lame logic to argue over you if not deleting your in his blog. i don't think to reveal his name here, but a clue here: occasionally he would use 1+1=2 statement is true, and thus his other 1+2=4 staement is also true too logic and alike. :)
oh, btw. Pat from intel just revealed the roadmap clearly indicating the Penryn in 207 and nehalem in 2008 as well as the >3Ghz thingy. But i don't think the fanboy will stop ...
The INQ just mentioned about 3.33GHz Peryns which are not really a dumb shrink after all!
http://www.theinquirer.net/default.aspx?article=38566
"Intel kept repeating the >3GHz mantra, but with performance numbers let a few things slip. For desktop, they mentioned 3.2GHz, and the demo today was done on a 3.33GHz part."
To Dr Yield and Dr Roborat,
I noticed your comments in one of AMD fanboy's blog, while it is ok to discuss matter technically for the fun of making those fanbois look stupid, but i guess we shall not talk in too much details, potentially revealing what others do not know about in the process or what so ever technology. Sometime things like using dual patterning and immersion in 32nm node might be something that is clear to you and your what so ever company, it might not be so clear for your company's competitor.
The Inquirer was more accurate than the much touted analysts ! Nothing beats insider info.
AMD fanboys wrong again, strike 69. Next they say Q1 is just another slip and the 2nd half is going to be roses.
Moon pie. Penryn coming to crush their dreams!
"Sometime things like using dual patterning and immersion in 32nm node might be something that is clear to you and your what so ever company, it might not be so clear for your company's competitor."
To the folks who actually do the process technology development (Intel, IBM, AMD, TSMC...) the info on these blogs is trivial - it is only the idiot fans who think moving to immersion early is a good thing (because the press who reports on this stuff are technical idiots) who are uninformed. Any competent R&D group would at a minimum do feasibility work on both scenarios.
That said Intel has always demonstrated a strong capability to extend "old" technologies further than expected. While moving to new technologies may sound good and be good for a headline, unless it is adding value (yield, clockspeed, cost, etc), it is better to reuse existing technologies as much as possible as this lowers cost, risk and generally allows for faster ramping.
As immersion does not do any of these things (that I'm aware of), moving to it earlier than your competitor is NOT an advantage. Last I checked the 45nm features (actually, smallest CD on this node will be in the 30-35nm range), will be the same between Intel and IBM... as there are no actual 45nm products available now there is no way of verifying this physically but I think it is safe to assume both companies will follow the ITRS roadmap on feature sizes.
You know, if you really want to have a conversation about what I say you could just ask.
"Penryn's just a shrink of Core2 it will not yield that much difference."
As far as I can tell; Penryn is a 45nm shrink with extra cache. The cache is particularly important for quad core because Intel is still trailing on memory bandwidth. There is no indication that Penryn will increase IPC however it does look like Intel will get some higher clocks and lower average TDP.
"Intel will not be able to release anything above 3GHz until 2008"
The latest desktop roadmaps still don't show anything higher than 3.0Ghz. Is HKPEC an AMD fansite?
"Intel will not have IMC/CSI until 2009"
I think Intel will have IMC/CSI out on Itanium in 2008. It is possible they could get this out on X86 servers very late in the year but I still don't see it on the desktop.
"When fusion comes out AMD will reinvent the market and put Intel in catchup/copy AMD mode"
This isn't something I've said. I would never use that phrase. It certainly looks like Intel recognizes the advantage of having the GPU in the same package and is working toward the same goal. I wouldn't call this copying since Intel likely had this in works when AMD announced Fusion.
Pointer
I'm sorry if you feel bitter about not being able to comment on my blog. The simple fact is that I don't talk about you behind your back; why do you feel that you can talk about me behind mine? TheKhalif is strongly pro-AMD and I gave him the same warning when he personally attacked other commenters. I expect a certain level of civility or the comments break down into playground taunts and superlatives which are a waste of time. If you really want back on the blog make a post and show that you have some personal integrity. I would not make this offer to Red.
Why am I not an AMD fan? The simplest test of whether or not someone has a slanted point of view is whether they recognize qualifying factors. Fan statements never included these. For example:
Barcelona will increase IPC however it isn't going to be clocked as high as C2D.
Intel seems to have overestimated its ability to deliver > 3.0Ghz on 65nm however Intel will have a six month lead at 45nm.
Intel is likely to bump clocks over 3.0Ghz in early 2008 however AMD has a history of late clock bumps so higher 65nm speeds are possible.
Intel is almost certainly going to lose server support from IBM however this may not matter since Intel could have increased support from Sun and Dell. Secondly, IBM's support for Opteron is not likely to be as strong as it's support for Power.
There are lots of factors like this that prevent the landscape from being black and white. I never here qualifying factors like this from true fans. For example, I noticed that Ho Ho corrected an anonymous poster on Roborat's blog who had claimed that K8 didn't support SSE or SSE2. A true Intel fan wouldn't have corrected the poster.
In fact, Ho Ho overstated the correction by claiming that K10 would get SSE4 before Intel. Actually, K10 only gets a few of the SSE4 instructions so Intel will have a number of SSE4 instructions that K10 does not heave until at least the first K10 revision. Shanghai perhaps.
Pointer
"Pat from intel just revealed the roadmap clearly indicating the Penryn in 207 and nehalem in 2008 as well as the >3Ghz thingy. But i don't think the fanboy will stop ... "
Intel Press Release, March 28
Higher Frequencies -- Penryn family of products will deliver higher overall clock frequencies within existing power and thermal envelopes to further increase performance. Desktop and server products will introduce speeds at greater than 3GHz.
Here is a quote from Pat Gelsinger:
The technology will enable Intel to offer desktop and server processors at speeds greater than 3 GHz and a 20 percent performance improvement over the current line of "Merom" and "Conroe" processors, said Pat Gelsinger, senior vice president and general manager of Intel's Digital Enterprise Group, in a U.S briefing Wednesday.
Neither of these mention > 3.0Ghz clocks in 2007. Greater than 3.0Ghz in 2008 seems perfectly reasonable.
The quote about Nehalem from the same Intel press release is:
After Penryn and the 45nm Hi-k silicon technology introduction comes Intel's next-generation microarchitecture (Nehalem) slated for initial production in 2008.
The phrase "initial production" makes me think this is going to be late in 2008, and Nehalem still hasn't appeared on any roadmap. I simply don't expect Nehalem to match the release of C2D where the server chips appeared at mid year. I'm sorry if this seems like fan point of view to you.
anonymous wrote:To Dr Yield and Dr Roborat,
I noticed your comments in one of AMD fanboy's blog, while it is ok to discuss matter technically for the fun of making those fanbois look stupid, but i guess we shall not talk in too much details, potentially revealing what others do not know about in the process or what so ever technology. Sometime things like using dual patterning and immersion in 32nm node might be something that is clear to you and your what so ever company, it might not be so clear for your company's competitor.
Anonymous, I think you mistake my points. Any info I discuss is common knowledge in the industry. IBM and AMD have oodles of very, very smart process engineers, just like Intel. My only point is that all patterning approaches have tradeoffs, and the typical blog reader doesn't understand those tradeoffs. Around here, and other sites, newer==better. In actuality, newer==different. I want people to understand that AMD and IBM chose one path, Intel another. Same thing happened with copper wiring. Intel managed to achieve end performance with Al metal that was equivalent to the AMD/IBM performance with 1st gen Cu, at lower process integration costs. In general, Intel always chooses a more conservative approach at higher than baseline costs because at their volumes, they it makes better fiscal sense to postpone introducing new technologies. Longer times to debug the process makes for higher yields and binning when they are finally introduced.
"The phrase "initial production" makes me think this is going to be late in 2008, and Nehalem still hasn't appeared on any roadmap. I simply don't expect Nehalem to match the release of C2D where the server chips appeared at mid year. I'm sorry if this seems like fan point of view to you."
You are so unaware of you're bias it's starting to get funny. You state your point on CSI being 2009 because you don't see any roadmap with it on it... do you have a roadmap for 2009 saying that is the scheduled release? If not why not say CSI in 2010? 2011? You ASSUME that if there is no data to contradict you're statement, that it is by default correct, yet fail to support YOUR OPINION with any fact! And you continue to make this statement as if it is a fact as opposed to an opinion from an AMD fan.
Intel is sampling CSI on Itanium this year, I do not think (OPINION!) it is a stretch to think that when Nehalem is released in 2008, CSI could be ready - yes Intel may delay it and focus on 1P/2P server space which may stick with FSB, but it is hard to say with authority what will happen (though you seem to have no issue speaking with authority on these topics!)
"As far as I can tell; Penryn is a 45nm shrink with extra cache."
If you truly believe this than i have vastly misjudged your knowledge of architecture and ability to read. While Penryn is not a major re-design of Core2 , it includes SSE4 instructions (which can clearly be seen on the divx benchmarks now on the web) and some minor FP changes. IT IS NOT A SIMPLE (DUMB) SHRINK! You are either misinformed or intentionally trying to spin this. (After looking at your blog I'm now starting to believe it is the latter)
"It is possible they could get this out on X86 servers very late in the year but I still don't see it on the desktop."
Your comments made no mention of desktop - you simply stated no CSI on x86 architecture until 2009. Honestly IMC/CSI is not likely needed for the desktop - Intel in fact has said they will only be implementing it initially in server space (2P+? or 4P+) anyway.
Scientia - your blog has become a joke! Your latest article on thermals is hilarious and approaching Sharikou territory! Especially when the person who commented that he is running under 60C on his 6600 and you said he is out of design spec and validating your point! He was freakin running a 6600 at 3.3GHz and with a higher than normal Vcore - now drop the Vcore to "spec" and run at normal frequencies - what do you think will happen to the temp? And the data you were quoting off Anand was peak Temp (not average) so while the application under load may have averaged 88% utilization, for your outstanding calculations shouldn't you have used AVERAGE temp during that test? And you fail to mention where 55C became the thermal spec - you linked to a THG forum page meant to help out newbies!
Can I look forward to your Q1 earnings: good news or bad news? After all your Q4 analysis article on earnings was pure genius. (apparently AMD's Q4 #'s were good news and was not pointing to any major troubles in Q1). Perhaps you can start factoring in VIA again and explain how Intel is actually still losing market share!
Anonymous said...
...
If you truly believe this than i have vastly misjudged your knowledge of architecture and ability to read.
...
wow, we share quite the same viewpoint! :)
anyway, you comments might make him think that you are me as his post was directed to me. Anyway, nice comment.
Scientia from AMDZone said...
Pointer
I'm sorry if you feel bitter about not being able to comment on my blog. The simple fact is that I don't talk about you behind your back; why do you feel that you can talk about me behind mine? TheKhalif is strongly pro-AMD and I gave him the same warning when he personally attacked other commenters. I expect a certain level of civility or the comments break down into playground taunts and superlatives which are a waste of time. If you really want back on the blog make a post and show that you have some personal integrity. I would not make this offer to Red.
nope, i didn't talk this behind your back. I was giving my opinion on you, on the few blogs that you visit. you can freely defend yourself if you think that it is not true.
I'd not post in the blog where the owner would just do anything, name labelling and post deleting on defending his own view point. Check my comments in your blogs and others, I have never used bad words and such. while i was giving my thought why you view was wrong with my detail reasoning behind it, you simply name labelling me as fraud and i also saw you deleting post.
btw, side notes, i visit Sharikou's blog for fun. While i'm not agreeing with his post 99% of the time, I mostly do not respond to his post but other's comments. I just simply don't want to respond to thing that is clearly can be classified as jokes. I will responds (if feel wanted) to comments that mix truth with false statements, miss-conception, etc. I would not claim that i'm right, but will say my mind with back up statements. others are allowed to argue over it without me being pissed off (as long as they are civilized).
and as of the khalif case, did I use a bad word to responds to him (sometimes i will , but my bad word is limited). I responds with statements, with back up data showing why booting to Windows is a feat but showing task manager is not. I even provide a batch program and assembly for people to test their task manager. and I wonder how my comments can be labeled as fraud?
Yes, i do posted AMD jokes in my blog. But what is it to do with my comments. Secondly, those jokes, are their marketing jokes, saying by their very own marketing folks, full of marketing BS. While others claiming that Intel is a marketing company and thus and thus, I just posted in my own blog to show what happen with AMD marketing. Nevertheless, what is it to do with my comment or credibility?
that's the reason why I think that you just can't really take comments that disproving your view and has bad blogmanship.
Post a Comment