UBS analyst humiliated
UBS analyst Thomas A. Thornhill recently upgraded Intel to buy. SeekingAlpha.com dug out his track record. It's hilarious.
posted by Sharikou, Ph. D. @ 6/22/2006 12:10:00 AM
Journal of Pervasive 64 bit Computing
Main Blog Page
Site Feed:
sharikou.blogspot.com/atom.xml
Blog Roll
Hot pages
sharikou.blogspot.com/atom.xml
Blog Roll
Hot pages
- Price War Strategic Analysis
- AMD and Intel Performance
- Intel's false performance claims
- AMD's Architecture Lead Persists
- AMD and Intel Performance
- AMD and Intel Architecture
- Bensley/Woodcrest and Athlon MP
- Bensley/Woodcrest is 2P only
- 64bit CORE Performance!
- Conroe Busted
- Capacity: AMD vs Intel
- Conroe's Impact
- AMD and Intel in China
- A Beautiful Picture!
- Temperature and System Reliability
- Google and AMD
- AMD's Grand Masters
- Intel's Predicament
- AMD and Dell
- NGMA Core and Colwell 6th Gen Arch
- AMD and Intel History
- The Tale of Intel and Anand
Analysis on IT trends and competitive strategies, with emphasis on micro processors, computer systems and networks. Based on latest news, backed up with real data, this site intends to provide a true and realtime picture of the fast changing IT landscape. This journal strives to be accurate on facts and sharp on criticisms. You may email your opinion to sharikou@yahoo.com or post comments here, be cool and intelligent.
About Me
- Name: Sharikou, Ph. D.
Freelance journalist on IT matters. Some of my writings have been published on online IT journals. Any original content on this journal is Copyrighted, but it's free for non-commercial use. Any Trademarks used on this site belong to their respective owners. Some of the pictures are links. If there is any issue with the content of this site, please email sharikou@yahoo.com .
Previous Posts
- AnandTech Busting Update
- Dell Laptop explodes: Intel Inside
- Woodcrest semi-officially fragged before launch
- Inverse threading for AM2?
- Pentium 4 has zero market appeal
- Intel choked distributors to death
- DELL: AMD64 the fastest growing notebook market
- Price war preview
- The AMD64 Rev G Mystery
- On DELL and AMD again
Thursday, June 22, 2006
31 Comments:
Check this out:
http://www.theinquirer.net/?article=32566
I bet the analyst showed the same graph to his boss and got a promotion for screwing so many customers out of their money.
Intel just opened a new plant in Ireland and ready for 65nm production straight away! AMD caught again...
I guess they aren't wrong after all!!
Kentsfield suffers bandwidth woes: http://www.theinquirer.net/?article=32577
Looks like AMD might beat Intel when quad cores come out next year. But this year its Turion X2 and Athlon X2 are not doing well vis-a-vis Merom and Conroe.
bb
AMD have planned to use 65nm technology in Q1 2007.
They're not using it now because they don't have to. They don't need to cram more transistors together, they don't need to reduce heat or power efficiency either.
Thus why this is potential. If Intel are utilising it now whilst AMD don't have to, this means AMD have the edge when it is implemented.
Why do it now when it's not needed? Don't win a football match by going hard for the first 5 minutes and then slacking off.
As far as I see it, Intel were pretty much forced into a position where they had to use this technology to dig them out, not to be ahead.
Also, AMD have planned new fabs which will more than double their production in the coming years. Some of them are already 65nm capable.
Not being a fanboy, but that's my gist of things at the moment. I'm just waiting for Conroe to come out. Yes I hope it's a performanice hitter and hopefully will top AMD as that is what we all want; evolution.
I better thing to reflect on is the forward looking P/E of INTEL, gross margin and cash flow. Then take a look at its competitor.
INTEL has funded all of its 65nm capacity already. AMD is just starting. With all this capacity in place only one of the company has what it takes to survive a price war and still make a proft. That is the company that will has full capacity at 65nm, the fastest design, and full infrastructure to support a 100% market converson thru early 2007.
Pss.. that is INTEL. AMD doesn't have all the things in place till 2nd half 2007. They need K8L then ramp that baby. By that time they'll be facing a 45nm new chip from INTEL.
AMD looks pretty grim..
Study, understand, and weep
Richard - you are...not smart.
AMD will move to 65nm ad soon as they can - it saves them money. They can make smaller chips, and more chips per wafer - ergo more money per wafer.
Please try to keep up and don't use these ridiculous excuses to try to make AMD look good in the manufacturing dept.
""Pss.. that is INTEL. AMD doesn't have all the things in place till 2nd half 2007. They need K8L then ramp that baby. By that time they'll be facing a 45nm new chip from INTEL.
AMD looks pretty grim..
Study, understand, and weep""
And you're not smart either...
Oh Ok AMD is wasting bilions in new FAB's because it as a grim future... how dumb are you?
And Intel's 45nm will appear only in 2008, and AMD as been already 1 year behind in the manufacturing dept. and that didn't stopped AMD for gaining market share, so now they are 6 months, big deal...
A AMD don't need K8L, the rev.G at 65nm due in December will be more than enough to put Intel back in second...
Study, understand, weep and stop being retarded.
Richard, your absolutly correct!!
Sharikou, I really wish you'd do a piece on Richard's subject... people need to be properly informed (stats, data, etc).
Intel 'may' have 65nm now and 45nm sooner than AMD, but they HAD to in order to stay competitive.
Obviously it would be nice for AMD to have 65nm now... but the longer you can profit from your investment (a.k.a ROI), the better it is.
Brag about Intel's manufacturing ability, blah, blah, blah. But with
with 4MB level 2 cache's, Intel MUST go to 45nm ASAP in order to reduce their cost structure.
Intel is in a race for survival here, AMD is not... that's the difference!!!
"Brag about Intel's manufacturing ability, blah, blah, blah. But with
with 4MB level 2 cache's, Intel MUST go to 45nm ASAP in order to reduce their cost structure."
IMC is AMD's ability to remain competitive. It's simple, straightforward and can make up for the lack of cache RAM.
Intel, on the other hand, has to justify the R&D they spend developing speculative this and that, beefy out-of-order processing, larger caches, etc.
Actually, I remember an interview with one of the head honcho's over at AMD when question about 65nm technology and Intel's take on it.
He really didn't care and that's when stated that 65nm for AMD will hit in Q1 2007. AMD have a said roadmap which has proven itself and remained basically unchanged unlike Intel's. They have also designed their chips ground up with dual core in mind, also unlike Intel.
Why should a company with 65nm technology now strive? It's performance people strive for. Yes it's cheaper to make more chips on the same size wafter, but hang on, who want's chips that don't perform?
AMD can already go to 65nm technology, they even have the resources and financial backing to do so. The only reason why they would move to the process as of immediately and prematurely is if Conroe really does blast them out of the water. In which case they have no choice.
You can also see this as AMD one step ahead, forcing the opposition into a position whilst maintaining their own stand themselves. Sure they can move over to use it now but for what reason when they are in a comfortable position with great market share.
The socket AM2 could've very easily been 65nm technology as well and it wouldn't have surprised me if it was however, I didn't realise AMD were in such a comfortable position. I was also hoping for a great performance increase, but we only got 1 new flagship processor and a series of chips that were more efficient and consuming less power. Now this is great and I did expect a little bit more however, these steps are great and obviously plans for the not too distant future, but I do believe this will be the shorest life Socket AMD have released in many years.
As said before about the price war. Regardless of the facts behind what you've stated with no evidence, if either company should collasp that would be disatrous for the World economy. It would completely monopolise the CPU market and therefore slow down basically everything really. So obviously, according to you because "they can manufacture 65nm chips and can save money" they are going to "win" this "price war"... Look at AMD's prices already. They even had to HAVE a price rise on their Opteron series because they proved so popular, mainly due to their larger cache, but really performance and that's what we buy. Best performance per dollar. At least, that's what I do.
When Intel switched to 65nm, they didn't just reduce the process therefore the same chip was only merely smaller. They restructured like any other company would and added more transistors and another core. So this means that basically the chip size is still the same and therefore the same amount of CPUs are still produced on that same wafer. If AMD switched to 65nm, they could either completely restructure or simply reduce the cost of their already cheap CPUs but this in turn could be less revenue because it's good to keep your CPUs at a price of the competitor so that you can maximise it. There's also the money you have to consider, not performance.
I'd also like to know why my words are "ridiculous excuses". They're not excuses and merely the truth. You can even look at AMD's roadmap if you don't believe me.
To sum it up, AMD have a strong integrity which is being held together nicely. They are not promising anything, nor do they advertise a lot and basically monopolise the market; but in the end they shove it in the fact of Intel. Just like the "dual core battle" where Intel didn't rock up. This is because they don't believe in such tactics and don't have to. Look at the advertising campaigns, they're nothing more than hilarious and truthful jokes.
People need to think about the biggest picture, money and performance. What else matters? How they do it, why they do it? You don't care do you. You want the best product for the cheapest price and that's about it really.
Apologies for not backing up the facts above straight from an interview. I would try to source it but it was from about the time Intel switched to 65nm.
You can also look up when Intel switched to 65nm technology.
I remember it was trustedreviews.com and I believe they stated that this was a good move because the previous generation were already producing too much heat. Whilst AMD on the same process had no troubles at all (they still don't) and still have the performance to back them up.
"Why do it now when it's not needed?"
I'm spitballing here but ~2X die/wafer for a company that is capacity constrained right now would seem to be a good reason. If they are truly sold out of all capacity for the foreseeable future a faster transition to 65nm which would roughly double the number would seem to be a good move...
Also AMD is claiming 40% improvement in performance (which is complete crap, but the transition will definitely give some performance benefit) - they are holding it back because they "don't need it"?
"Also, AMD have planned new fabs which will more than double their production in the coming years. Some of them are already 65nm capable."
For those having a hard time reading this let me translate:
- "some" = 1 (partially)
- "AMD have planned new fabs" = they are planning to CONVERT an existing 200mm fab to 300mm (there is speculation on 1 new fab in NY)
"Not being a fanboy, but that's my gist of things at the moment." = I have no actual knowledge in this are so I will just creatively misinterpert things and misinform people.
"Sharikou, I really wish you'd do a piece on Richard's subject... people need to be properly informed (stats, data, etc).
Obviously it would be nice for AMD to have 65nm now... but the longer you can profit from your investment (a.k.a ROI), the better it is."
~70% of semiconductor equipment is reused during a technology node transition (ex: 90nm - 65nm). So you can reuse 70% of your equipment and get roughly double the die output
....and you're saying the ROI is better to stay on the older technology?
Since you clearly have a "solid" finance background all you need to do is ammerotize depreciation on ~30% of the capital over a 4 year period and compare that to get getting rougly double the die out...then you will realize the ridiculousness of your comment. (or you could go on blissfully thinking that it's a good thing AMD is waiting because it wouldn't help from either a capacity or gross margin perspective)
Oh and all of their 300mm equipment is brand new and they will end up throwing away about 30% of that equipment when they fully transition to 65nm, so yeah it's a good thing they didn't just start up on 65nm.
Guys are you in the dark ages...
INTEL needed the process advantage to stay close with the inferior net burst Pentium IV.
In case you haven't noticed.. Conroe, Woodcrest, and Merom will absolutly crush anything AMD has for the next one year. Even when AMD gets to 65nm they will be behind. Their only hope is K8L. That will keep them ahead in the 4 ways market but on the desktop and mobile which represents 90% of the volume the won't have the benchmarks. By 2nd half 2007 INTEL will be on 45nm again 9 months ahead of AMD. they will have a new design so I expect they will maintain their lead on the desktop and mobile..
So.. in the end you weep. They are spendind billions, but its so sad. They should weep somemore. They should have put that capacity in place on 90nm. At that time a year ago Dell would have moved completely to AMD if AMD had capacity. Tomorrow AMD will have some more capacity but the market won't have a compelling reason to choose AMD as INTEL will have more capacity and better pefromance.
AMD is bringing up capacity 2 years too late.
The only way they won't be in a lot of hurt is if 65nm shrink of Opeteron gives them 30-50% speed up. Frankly that isn't possible. Or if K8L pulls in 6 months.. and that won't happen either.
Did you notice AMD stock is down 30% since IDF and Woodcrest annoucement. Its going to go lower.
INTEL is the buy...
You can be stupid... and back AMD
Or you can be smart and rich and have bought INTEL at 17 bucks. Even at 18-19 bucks its a buy.
AMD has EOLed their single core A64 3000+ and 3200+ (plus scrapping one of their 1MB cache production line in favor of 512KB cache) which indicates capacity issues at hand. Also Opterons will be in severe shortage for sure...
This means AMD will leave the "value" segment market wide open for Intel's P4/PDs to dominate (which is AMD's traditional stronghold and big volumes here). And Intel has the volume to flood this segment, a good reason why Intel can afford to slash prices up to 60%. This will kill Semprons (and Intel's own Celery)...
As Taiwan mobo manufacturers have indicated in the HKEPC article, there is no way AMD can fight Intel in the value segment once the price wars started. That's why they have increased their ratio of Intel mobos to AMD mobos.. (can be noticed at Computex 2006 as well). The value segment is pretty significant due to volume...
Hiring another FAB (Chartered) for production also means AMD has to pay that FAB as well, which goes into the pricing also..
Since AMD is unable to reduce prices further, they gave up the low profit areas (value segment) and will try to fight it out in the mid range segment (where profits can be maintained). I don't see this as a good move as volume would not be as good as the value segment..
With their 2.5-billion investment on their single FAB upgrade and is not even ready yet.. Let alone AMD's profit last year was only around 300 million? Would take them years to recoup (return on investments)...
Intel ain't survivin.. Remember that their profits last year was 8 billion! That's good for at least 3 FABs.. Intel can afford them.
Maybe it will be even smarter to buy Intel when it is at $10.
Even with Core Duo 2, Woodcrest, etc., there is no profound change that Intel is offering to the market. It is all still FSB chips that run hot. Intel's premiere vendor is now anti-mainstream Apple, so one really cannot expect any market miracles there.
After Dell kicks in with its AMD offerings... then we will see what Intel stock is really made of.
Some of you really do believe everything you read. It's called propaganda!
How can AMD be capacity constrained? What! Your going to tell me that such and such person said it and since [s]he works for one of the largest banks in the world, then it must be true! I pitty the fools :)
The margins are not as great, but AMD has Chartered to back them up if that was the case.
Since you clearly have a "solid" finance background all you need to do is ammerotize depreciation on ~30% of the capital over a 4 year period and compare that to get getting rougly double the die out...then you will realize the ridiculousness of your comment.
People, we have a real genius here!
How much does it cost? Something like $1 billion? Or is it close to $2 billion? Amortize that over 4 years, but in two years you'll spend another $1 billion to change it to 45nm. Ah what the hell, amortize that too on top of the old amortization!
We call this Enron accounting!
You can double your production at half the price, but as long as you don't have buyers, your in shit!
Or maybe you can try to entice buyers by lowering the prices by 50% on all products? This would just collapse the industries ecosystem. This is not building a solid company with a long-term vision.
I don't claim to be a financial expert. Nor do I have the numbers to back-up my point, but I do trust AMD's operational judgment.
Food for thought
Last year INTEL was at its worst competitive position. Invested 5 billion and made 8 billion
AMD was at its best postion, got 3-5 points in MS.. and made a few 100 million.
This year, the bottom will fall out. INTEL's investmet rate is abou the same, AMD's is what 2x as they walk into a price war and their most uncompetitive position in 3 years..
You tell me who is in better shape to invest in the future..
By the way AMD is buying IBM's process. You think IBM is tailoring it to AMD or just retrofit what it already has. Do you think IBM brings a wealth of high volume high peformance technoloyg know how.. I don't think so. When was the last time IBM had to push volume and lots of performance.. Never. How about charter.. they are a second rate foundry that also has little experience in really high volume performance CPUS.
Things aren't good.. K8L better be a big rabbit. IBM 65nm needs to be great, then AMD needs to execute a transfer and ramp faster then they ever had... and they deal with droping prices..
That is the competitive landscape!
You can basically write Intel off your book. It will be reduced to 10% of its current size in two years. It's simple economics, just try figure out Intel's revenue and profits in the next 7 quarters. What you will see is straight losses. Intel is betting on Conroe, but Conroe won't be released until July 23. AMD's 65nm Rev G parts will flood in 4Q06, followed by Rev H (K8L) in 2Q07. DELL is not stupid. So it goes AMD full force. The result is Intel will lose massive amount of market share, coupled with price collapse, you see at least 7 straight quarters of losses. The result is BK.
"People, we have a real genius here!
How much does it cost? Something like $1 billion? Or is it close to $2 billion? Amortize that over 4 years, but in two years you'll spend another $1 billion to change it to 45nm. Ah what the hell, amortize that too on top of the old amortization!"
Uhh... this would be a good argument if you were COMPLETELY retooling the fab every generation (but you're not). Typical fab cost for something the size AMD is building is 2-2.5Bil; what you are ignoring in your analysis is to go from one generation to another you only have to change out ~30% of the equipment. This means you get double the output for that ~30% cost increase (if you do the math this means positive ROI, assuming yields are similar which AMD claimes them to be)
There is a reason, other than performance, that people shrink technologies...I don't claim to be a genius but what I'm stating is that for ~600mil (depreciated 150mil/year over 4 years) AMD could get double the capacity from one technology node to the next.
One other (stupid) thing in your analysis - fabs generally skip a generation to get full benefit of depreciating the equipment. If you look at some of the Intel fabs they generally go from 90nm to 45nm (skipping 65nm) or they may go from 65nm to 32 nm (skipping 45nm) - this means changing out ~50-60% equipment every 4 years.... but then again what do I know as I come from the Enron school of accounting.
My point is AMD will go to 65nm as soon as technically possible - they are not waiting or holding off as some have suggested ("as there is no need") there are just too many economic and performance reasons not to.
"How can AMD be capacity constrained?"
Why did AMD cut cache on desktop products?
Do you think AMD is outsourcing some of their chipmaking to Chartered because they have an abundance of capacity in their existing fabs? (Why would they do that if they had plenty of in house fab capacity?)
They have 1 200mm fab and part of a 300mm fab (still ramping) - that is not that much capacity given their ~20% market share. Until they get a 2nd 300mm built out they will be constrained unless Intel takes market share back (which they might with price cuts and Core 2)
"You can basically write Intel off your book. It will be reduced to 10% of its current size in two years."
I assume 10% size (or 90% recduction) would translate to ~8-10% of overall market share? (roughly).
So you are saying AMD will be able to supply 90% of the market's chips or ~4.5X their current capacity in 2 year? No way.
To put things in perspective for you, most advanced (193nm) lithograhy equipment has an ~18 month leadtime, meaning if you ordered the equipment today you would get it in 18months (or at the end of 2007); if AMD goes the immersion litho route, leadtimes for that equipment could be even longer than that!
So you are saying AMD will be able to supply 90% of the market's chips or ~4.5X their current capacity in 2 year? No way.
AMD will expand its capacity by 4X in two years, this was info from AMD's recent analyst meeting.
I don't believe the analysis; I'm also pretty sure AMD did not say 4X at analyst day. I couldn't find the foils easily on AMD's web site.
The 300mm fab does not have as many wafer starts per month as their current 200mm so it is not ~doubling the capacity due to 200mm vs 300mm as you have previously argued in other posts.
Also the 2nd 300mm fab is taking the place of the existing 200mm fab. so this is a net output increase of ~30-50% by replacing F30 with F38. I'm assuming F38 will have similar size as F36. I'm also factoring in the higher F30 WSPM in the calculation.
Finally factor in a greater migration to dual core and quad core (as you have mentioned will be out soon), this will further limit die/wafer. Unless they will be getting ~1/2 of their chips from Chartered there is no way they can do 4X in 2 years. If you read my last note they will barely be able to order litho equipment needed in the 2 year timeframe you suggest.
"AMD will expand its capacity by 4X in two years, this was info from AMD's recent analyst meeting." - Taking in AMD's marketing crap is easy..
Did you see that new Irish plant Intel opened, wait.. there's another on the way at Oregon. Many things can happen in two years time.
AMD is already delayed too much as it is.. and unlike 90nm, the newer 65nm process is a different "beast" altogether. That is if it were so easy to transition from 90nm to 65nm, wouldn't it happen earlier already? Do you know 65nm on SOI is very difficult? No to mention 45nm, or 32nm in the future... And SOI may prove to be AMD's own achilles heel.
Intel transitioned to 65nm without the need for SOI because they found another way.
Can't imagine now when Intel starts to flood the market with P4s/PDs and Core processors... even before Rev.G is released. The value segment is already theirs for the taking...
And.. Beware! AMD is sleepin' with the enemy, in this case the enemy is Dell!
My point is AMD will go to 65nm as soon as technically possible - they are not waiting or holding off as some have suggested ("as there is no need") there are just too many economic and performance reasons not to.
And I couldn't agree more!
But my point is that you can't say Intel is a better company because they are already at 65nm, which is what most analyst use as their argument for investing in Intel vs. AMD. And that is the first argument most Intel fanatics will use too!
Why did AMD cut cache on desktop products?
In part, it may have been to increase capacity, but this doesn't mean they are constrained. There are many alternate motives to this this decision.
AMD's architecture is much better than Intel's and they don't need huge L2 caches like Intel does to compete.
The 'extra' 512K cache is equivalent to an extra 200MHz, so the idea is cut back the cache and sell faster processors.
Doing this you can increase capacity (I'm guessing by 10+%).
You become more competitive for this so called 'price war'.
Besides, when you make the move to quad-cores on a single die, such as AMD is getting ready to do, your yeilds will take a hit.
Chartered is just an insurance policy for AMD's limited resources.
If AMD really wanted to, they could start using 65nm technology. Anyway, they have planned finanicially not to use it and as said before stuck to their roadmap. Ofoucrse using 65nm tech is a great improvement and I never said it wasn't.
Don't forgot AMD still owe millions to pay off their loan. I forgot the total amount. Was in an article because they are considering buying ATI.
I'm just waiting to see how "great" Conroe really is. Seems like a make or break.
"AMD's architecture is much better than Intel's and they don't need huge L2 caches like Intel does to compete."
What would happen if Conroe had an "IMC", and dumped half of the L2 cache? Therefor dumping memory latencies and having shared L2 cache.
Isn't it something that AMD would have trouble competing against?
"What would happen if Conroe had an "IMC", and dumped half of the L2 cache? Therefor dumping memory latencies and having shared L2 cache.
Isn't it something that AMD would have trouble competing against? "
No... because they still don't have Hypertransport and are still limited by their Front Side Bus architect.
The chose to exclude IMC because they can change (or make us consumers suck it up) their mobo/memory every 2-3 years, as that would require AMD respin a new silicon for an updated IMC on their chip.
Intel has stopped innovating, ran by marketing non-techies, and only out to swash AMD...
Just think about what Intel has done for the industry lately...?? Besides monopolize and squeeze out everyone. Those analyst are as bad as Intel as they're out to make a buck, and can care less about us little people.
Post a Comment
<< Home