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.
alan,
Rumor has it that Intel will introduce a 65nm product like this in late 2007.
late 2007? I don't think Intel is scheduled to have on die memory controller by the end of 2007, not to mention graphics. Could this be a multi-chip module? Or the year is off...
Joe
Snow,
Just a speculation, but I believe the strategic driver behind keeping the flash business is Samsung. As long as Intel has a flash division they can fight Samsung on their own turf rather then letting Samsung get a foot hold in the CPU market.
Samsung NAND division >> Samsung NOR division
Intel NOR division >> Intel NAND division
Well at least as of today. Most of Intel flash losses are in the NOR division, where Samsung is one of the smaller players.
Joe
I may be riding too much on what Apple did. But if you look at how Apple has grown in the past year - 1.5 % to over 7% market share. This is significant (Its pissing Dell off)- I dont think their through climbing. Although, I do wonder when they will become capacity constrained at this growth rate.
I read somewhere that Apple PC units dropped in Q4.
Joe
<Beavis impression>
Mmm heh heh heh. Mmm heh heh. He said Apple PC units
</Beavis impression>
wbmw,
I am not currently putting money on K8L being on spec and on schedule. Even if it is, the time to place a bet would be in June to August time frame, when the market is digesting Q2 results.
joe
wbmw,
I don't think Intel would be hopping on the NAND train if they thought they would end up with obsolete technology. But you are free to believe that if you want.
Maybe they thought wrong?
Obviously, it will take a while for things to work themselves out for us to have the final answer, but in the meantime a steep increase in flash loses is certainly not an encouraging first result of the strategy.
Joe
Klaus,
Any idea what Intel should do with the fabs after CPU, then?
The same as what AMD does - keep upgrading to the latest process technology, and if some of the existing equipment can't keep up, sell it off.
Intel does not talk about mss anymore. And it plans to ramp one fab less in the next node than in the current node. I don't intend to ignite a discussion on diesize-economies, i just mention it because i believe the street connects these nodes in way to explain the market reaction.
I wonder if the mentions of over-capacity and underutilization charges spooked some investors. Or it could be the lack of announcement of exist strategy from NOR - but that's proably way lower on the list of concerns of the investors.
Joe
chipguy,
If Intel hadn't introduced NGMA I would have sat on my
wallet last fall and stayed with my 3.0 GHz Northwood
box, AMD or no AMD.
Are you saying that you would not have upgraded if Intel added the "Ultra" feature to its Hyper-threading?
Joe
George,
The parts exist (they were reviewed by some review sites). But AMD apparently has started very vew wafers (likely because of yield or bin split issues - or possibly a new stepping is in the works). I am judging this (very few wafers) by lack of visible output.
joe
sgolds,
With processors Intel has only one competitor, and that competitor would be happy to share the wealth provided by a price umbrella. Maybe it would cost one or two points of marketshare but both companies (and their shareholders) would be a lot richer.
While listening Q1 2006 CC, I though Ruiz extended (very indirectly) this invitation - to share the wealth. Intel very resoundly rejected the invitation.
Joe
wbmw,
Based on current data, it suggests that K8L is a more of a bust, while Penryn is more of a home run.
I would not (yet) call it a bust, but the schedule looks extremely challenging (to get it out mid-year).
Joe
mmoy,
I was thinking about buying some February 22.5s. Good thing I procrastinated...
Joe
mmoy,
I would go for 64 bit Vista, assuming all your apps work in 64 bit. If not, I would just stick with 32bit XP and an upgrade coupon for later on...
Joe
alan,
I'm pretty happy with the results...
I thought so too. I am a little suprised by the negative reaction on WS. One reason could be that Intel may have oversold the benefits of Conroe - or the timing of its impact.
I think when they finally sell the Colorado fab, and the NOR business (including fab 18) we will see a nice improvement in GM...
I think Intel should dump NAND as well. After Googling a little last night, it appears that Samsung is going to charge trapping technology when the process technology gets to the 40s nm, Spansion will certainly be there, and I think Intel will end up holding the bag with bunch of fabs (by then) and an obsolete technology.
Here is the link:
http://www.eetimes.eu/semi/193000877
Joe
wbmw,
Well, according to you, everyone compares YoY results. So Intel's strategy must have been to tank 2006 so that they can impress in 2007. What genius....
I don't think it was an intension, but the lousy 2006 (relatively speaking) gives Intel the opportunity to look better YOY in 2007 - assuming the growth of PC sales continues in 2007.
Joe
mas,
The only thing that has died is Intel's gross margins and profits from last year. AMD still sold well over a billion and made a profit. Intel should have been using NGMA to make more profits not less.
I think those investors looking for salvation (in form of Conroe) that Intel has been "selling" to the investors (for a year and a half now) may feel a little dissappointed by what has been delivered last Q.
But one of the jobs of the management is to keep investors sold on the positive future of the company. The job of an investors to make an independent analysis and invest accordingly.
Joe
Keith,
It´s still getting worse, just one example:
Sold out, no replacements. Same at CC, BB etc.
Why are you considering to be getting worse? Selling out (with more stock on the way) would not be that bad, unless not much new stuff is on the way. Is that what you are thinking is going on?
Joe
Keith,
If their ASPs rose while AMD's fell significantly, it looks great in comparison. Units rose as well. Possibly less than AMD's which AMD will be claiming as an accomplishment (likelyhood of increased unit market share). More meaningful dollar market share is down for AMD.
BTW, I have not read the report in detail, but the profit seems a little light, in light of good revenues, and prior divestment of some loss making divisions in previous quarter(s).
Joe
mmoy,
Pretty muted respons AH.
How is this for headlines:
• Intel Q4 profit beats estimates
Reuters (Tue 4:26pm)
• Intel Q4 Profit Plunges 39 Percent
AP (Tue 4:26pm)
http://finance.yahoo.com/q?s=INTC
Joe
Looks like a great report by Intel. ASPs up and units are up:
http://biz.yahoo.com/bw/070116/20070116006261.html?.v=1
Flash is $576, but it now includes NAND. Losses there seem to be growing, but also include NAND startup expenses.
The Seagate drives are just 7200, but since they are the recent models, they have a pretty decent data density, and therefore internal transfer rates.
Unfortunately, since my old system was kind of old, there was not much salvageable - DDR1 to DDR2, new mobo, CPU socket, AGP to PCI Express.
Hopefully, my future upgrades will be only one component at the time.
Joe
Paul,
Have you considered RAID 0+1? 2 fast drives striped and then mirrored on another 2 drive striped array should perform much better than your RAID 5 setup.
Initially, I did not, since I thought I could use 3 disks vs. 4 disks for the same amount of storage.
Joe
ced,
I have nVidia 590 chipset, which is capable of supporting 6 SATA II drives. I have Athlon X2 + 4 GB of PC 6400 memory. I was quite content with the performance increase when I set this PC up in the summer (I got a much faster graphics card - 7900 GTX + a dual core Athlon 4800 vs. single core K7 Barton). Initially, I installed on a single drive, but the "downgrade" to RAID 5 have been disappointing.
To get rid of the RAID, I would have to copy some data away, and find a way to shrink the partition and then clone to a smaller drive. But I will just wait for a TB drive, which will involve less work on my part.
BTW, good idea about testing with Sandra or something to obtain objective performance figures, but I think the performance, especiall write performance suffered so much that I can subjectively percieve it.
I haven't noticed any change in read speed, which should theoretically have jumped a bit.
The HDs are Seagate Baracua 10s.
Joe
PS: one of the reasons for RAID 5 was to get some redundancy...
I have 3 hard drives in RAID 5 configuration, and the performance is attrotious. It seems like (almost) no improvement in reads and some 75% degradation in writes.
I have 3 500MB drives for total storage of 1 GB, and I am planning to ghost back to single drive, after this unfortunate experience.
BTW, I am using integrated nVidia chipset for SATA RAID. Is there an alternative, within a reasonable budget, that would be less horrible?
Joe
PS: Yeah, HD seems to be the bottleneck for me, now even more so before I "downgraded" to RAID 5
Hmm... It is hard to read an article where the author seems to impose his ego over the possible news-worthiness of the article...
Joe
Alan,
This thing seems to be a go. Here is a post on SI:
http://www.siliconinvestor.com/readmsg.aspx?msgid=23186614
Joe
alan,
I am CERTAIN Intel was in talks to sell the business to the same investment group that is buying the STM NOR business. The only question is if the deal will actually go through. Intel needs to get enough cash for the business to pay for the assets. I suspect such a deal is pretty complex because Intel has flash factories and R&D in several different global locations... and exactly which ones get transfered and for what price would be a huge negotiation.
One thing that should help is that there are number of synergies already in place (product interchangeability, existing technology exchange), so there some cost savings that could be realized fairly quickly. I think the problem (for the investment group) is that probably even after the cost savings, there is still quite a way to go to achieve profitability.
Joe
wbmw,
I actually thought these rumors were credible, until nothing came of it.
These things take a while to close. If indeed the transaction would involve STM and Intel NOR units as bases of this new entity, the fact that STM just recently took steps to make NOR unit more independent of the rest of the company would IMO be indicative that the plan is still on.
Fortunately for me, NOR flash is such a small part of what Intel actually does. I am actually more interested in NAND, given the synergy it has with PC products. I'd actually like to see Intel grow in this market, but I'm sure they have a long way to go before they acquire enough design wins to show anything material.
As far as design wins, they should not really be a barrier to NAND ramp. There is such a huge sea of commodity NAND flash out there that if Intel / Micron unit is cost competitive, there is a ready market out there already.
I agree about the synergy with PC products. It seems that finally, someone has finally done the obvious, and put a bunch of these NAND chips in parallel to create solid state disk drive alternative that are a little more competitive with disk drives in read and write throughput.
Flash clearly beats hard drives in access time, and it is just a logical progression until more of the chips are placed in parallel to match disk drives in throughput speeds. One thing in favor of solid state disks is that the advantage of the disk capacity of disk drives becomes less of a factor, IMO. A typical office PC does not need a TB of storage (which is what will become available this year). I think 100 to 200 GB is all you can reasonably need in a typical offic PC (where bunch of the stuff is on the server anyway). Even though hard drives have cost advantage, the advantage would be decreasing over time.
Joe
PS: Spansion is not completely shut out of the NAND market. But it will take a number of steps before they can start competing there. The prerequisites are 300mm fab, 45nm, ORNAND and QuadBit. QuadBit is already here, so is ORNAND. 300mm fab is in the initial phases (building is there, Spansion is probably now in process of placing orders for equipment) and 45nm is still a bit more than a step away. So Spansion in NAND market is there only for a long term investment perspective.
The advantage for Spansion is that floating gate will apparently runs out of steam before charge trapping (basis of MirrorBit) as far as process shrinks - according to some sources. One of which appears to be Samsung, which has it (charge trapping) on their roadmap as one of the possibilities.
Joe
wbmw,
I think about it (flash) as a standalone business, not part of something else (since no longer have any position in AMD or INTC) and do in Spansion. There are some rumors of a Intel + STM NOR flash spinoff, which would completely separate it from the CPU business link.
I don't really get that excited over the actual products.
While I understand it helps if I am generally interested in in the area I invest in, and I would have hard time following investment of a sleeping pill manufacturer, flash is reasonably within my sphere of interest.
Joe
wbmw,
don't know how one could construe this to mean they have not released the parts.
Oops. A brain fart on my part.
Sounds like wishful thinking of the same variety as AMD 45nm products in mid-2008. If they just now switched to 90nm, I wouldn't expect 65nm this year.
I can't predict the future, I can only go by the guidance and track record. Spansion track record of meeting guidance on the technology side has been good since becoming an independent company, the track record of meeting sales (and profit) targets has been fairly poor (consistently dollar short of lofty targets).
Again, this is pretty unambiguous.
I was not doubting you, just asking a question, since I have not been following the NAND side of business.
Joe
PS: If you have any interest in following the NOR side of flash, there is a low volume Spansion thread on SI, where you will find some familiar faces:
http://www.siliconinvestor.com/subject.aspx?subjectid=56255
chipguy,
Keep increasing die size until falling manufacturing
capability matches demand? That's one tactic to
avoid underutilization charges but it will be murder
on cost of sales. :-P
Doing so across the board would be, but doing it only on performance sensitive part of the market (high end desktop, server) would make sense.
Joe
mas,
Short term yes. Long term they need to throw cache and cores at the current K8/L design to be able to continue to sell all they make.
I was thinking short term, until K8L ramps.
AMD are not finished but they are now in a scrap for their very existence yet again.
Yup. "Small" things (like Rev F delay) cascaded down, caused delay of Rev G, and made Rev H (K8L) release a frantic race - subject to potential mis-steps and future delays. As these delays cascade down, each one causes progressively greater financial damage.
Joe
wbmw,
One interesting thing is that they recently released a 65nm NOR chip, which is well ahead of most competitors.
I have been followin the market (since I invested in Spansion) and as far as I know, Intel announced 65nm, not released.
Samsung has a half node design at 80nm or 75nm, but nothing yet at 65nm. Intel could gain back some NOR share next year with this, since their previous designs were 90nm. I think Spansion, for example, is still primarily at 110nm.
Spansion has made pretty hard switch to 90nm last Q and will be at 65nm in 2007. I will have to wait for the CC to get update on the production schedule. The last Spansion update was that they are evaluating 65nm silicon. But Spansion is already starting the ramp of QuadBit, and is ramping ORNAND, which is a play for a portion of the NAND market.
As for NAND, Intel's biggest revenue source comes from the Apple ramp, which hasn't been material through most of the year.
What Apple product is Intel NAND going into? I thought Samsung was (nearly) exclusive vendor for iPods.
BTW, iSupply is predict some turbulence in DRAM and NAND markets, but Buggi says iSupply is always wrong. If the PC market is growing strongly (due to Vista launch), it should absorb plenty of DRAM, and therefore, the suppliers who have flexibility so shift production between DRAM and NAND would not have incentive to switch from DRAM to NAND.
Joe
Buggi,
AMD throws volume
at "any price" on the market just to sell out all the new
capacity which will come online.
That was Intel strategy, while awaiting Conroe, which Intel supporters thought was brilliant. AMD is probably going to repeated, while awaiting K8L.
I am not so sure that the strategy is so great, because the depressed ASPs are hard to recover. OEMs will be resentful to price increases, and will resist attempts of them gaining the driver's seat to Intel going back to the driver's seat.
Another problem I see is that it was hard to sell high performance at substantial premium (outside of performance sensitive niche) last time around (SC), and this time around (DC -> QC) it will be even more difficult.
I think what was helping AMD last time it had performance lead was not only the performance lead, but substantially better combination of thermal envelope and power management - which made AMD CPUs much more OEM friendly than Intel ones - but I am side tracking...
Back to possible strategies, AMD does not have many options other than price everything they can manufacture to sell. Intel did have an option when during the Intel initiated price cuts to not cut prices (so aggressively) and let AMD bump into its capacity limits and stay there, with much higher overall profits to the industry.
Intel's strategy was to sacrifice its own profits in order with the reason of limiting AMD profits (and unit market share losses, which were really bad PR for Intel).
As far as AMD not having any other options other than price everything to sell, the reason is that unlike AMD last time, Intel this time is not going to bump into capacity limit, since Intel is (likely) carrying extra capacity.
As far as Intel limiting AMD profit to limit AMD capacity, I think that cat is out of the bag now, and we are entering uncharted territory. With Q2 demand seasonality, AMD capacity growth and 65nm conversion (substantially in Q2), AMD capacity is likely to exceed 30% of the market demand...
Joe
wbmw,
You can still make money without having a stock position in either. I buy and sell options, which I think are perfectly good substitutes, if you choose the right ones. If you are interested in this, we should talk some more.
I would be willing to gamble a small amount on Intel's good Q4 results and Vista launch, which would probably mean February expiration. I would certainly welcome Intel going down in sympathy tomorrow (to get a better price).
The danger to Intel results is that the spotty reports of demand weakness in December may be real, and that it may have affected Intel as well.
BTW, I did have some Intel LEAPs earlier in 2006, when Intel was in the teens, but I sold them fairly quickly long time ago, which was probably a good idea, since Intel has not gone anywhere since then.
Longer term, I am more cautious, since all I see is more price competition in the low end, the vast middle of desktop and notebook.
Intel has been taking over the high end of desktop and 1 to 2S servers, where the performance competition is the most intense, and where performance advantage of Core 2 matters the most.
But in the competitive area, there is AMD no longer capacity constrained, giving discounts left and right (more emboldened by cheaper to produce, 65nnm X2 - which are bound to appear any day now) and OEMs very receptive to discounts.
BTW, I will probably take a look at Feb 20 and 22.5 calls...
Joe
wbmw,
Actually, it was more like over $200 prior to the bubble burst. I think it took a deep dive in 2001 or 2002, and then continued its steady decline since then until present day.
You could be right about the $200 pre-bubble, but I would be curious about 2001 to 2006. I believe there was a multi-year period of stability in $150 range prior to 2006 dive.
NOR has a different problem. It's not like the market has an alternative CPU type that is rapidly gaining acceptance over x86. And if there were, and Intel tried and failed to improve x86 up to the levels of success the new architecture was having, then I would expect them to do what they did in flash, and start implementing the new architecture into their business plan. Intel is doing this with NAND today, and I think the payoffs down the line will be good.
While there is a lot of talk, NAND replacing NOR, it has apparently not materialized go a significant extent. It is just that the natural NAND market has grown very fast, much faster than NOR, and NOR has been under more intense price pressure after Intel's 2004 price implosion. It is not natural for all market competitors to lose money at the same time, for number of years in a row. That's worse than the airlines.
Cell phones are growing, but it's not like there's any fear of using a cell phone processor as part of your PC.
I didn't mean to suggest any kind of competition between PC and a cell phone. I just meant to point to some parallels: -- One component of the cell phone (flash) has been struggling, following the price implosion, while the cell phone market has been growing. It has not recovered fully yet.
- One component of the PC market (CPUs) could be in the middle of similar price implosion even though the overall PC market may be healthy...
Joe
wbmw,
So has Intel substantially changed their product mix? Yes, actually, they have. And it's been in the positive direction with a strong Core 2 ramp, and new quad core skus introduced towards the end of the quarter at higher price points. If anything, that ought to give an ASP lift for the quarter.
These are certainly hopeful signs for ASPs to be at least steady, OEM discounting can skew things. With AMD giving discounts and growing units (we will see who grew units faster) I don't think Intel just stood by and not matched some of AMD pricing. Even if Core 2 was not discounted, there is still a substantial non - Core 2 volume.
Up until now, I would have been unsure about AMD's product mix. I was half expecting them to do alright, given large discounts, but strong demand for dual core. Obviously, the mix wasn't good at all, and the discounts dragged ASPs down heavily. For them to crow about substantial increases in unit sales, and yet still only grow 3% in revenues, their ASPs must have tanked.
Yup. Big time. Which is why I wonder if strong unit growth and substantial discounting left Intel unaffected.
Q4 06 and onward in 2007 could be another round of price declines like that of Q1 to Q2 06 with the roles reversed. AMD behind, fighting for unit market with low pricing just like Intel was in Q1 and onward in 2006.
Joe
wbmw,
Maybe if you start following my advice, you'll be making money on INTC and AMD as well.
What's your advice for someone who has no current position in either?
Joe
mas,
I know, isn't it ridiculous and all just because Intel can't stand a natural duopoly, pathetic really.
Well, I can understand Intel's desire to prevent duopoly from taking place, but Intel's arsenal this time around is limited only to legal methods, because of the microscope of the lawsuit.
Q2 should be very interesting quarter from unit share point of view and also from a potential of a disastrous price competition...
Joe
wbmw,
Which is lower ASP than in 2004, which was lower ASP than in 2003, which was lower than in 2002, etc.
I could be wrong, but I think Intel ASPs have been around $150 for a decade prior to 2006.
If I ever see another slide showing it's 10 year constant decline, I'll be sure to post it.
I would be curious to see it, if you come across...
And though it's still preliminary, it appears as if things have worked out better than they could have possibly planned.
Maybe I am a pessimist, but I see a parallel between Intel's 2006 price cuts in CPUs and 2004 price cuts in NOR flash. BTW, more than 2 years later, NOR is yet to recover.
Like in CPUs this year, mobile phone unit sales have been growing over the mentioned period, with little benefit to NOR flash vendors.
Though I don't think Intel could have done so well if not for AMD's help. They owe at least some of their prosperity to AMD's languishing execution over the past 6 months.
More than 6 months. Rev F (DDR-2 version) arrived effectively in Q3 vs. Q1 expectation (probably pushing out Rev G by similar time), so sub-par execution has been going on for year.
Joe
wbmw,
Maybe if you assume the market slowed down (which goes against all indicators), and assumed that Intel forgot to warn (which goes against their entire history), you might have a case.
I see market (PC market) doing fine (unitwise) but ASPs under pressure.
I don't expect Intel to miss their (fairly wide) range of guidance, but given AMD warning, I suspect Intel will be below mid point of the guidance.
But that would point to low acceptance of Core 2 (which goes against all indicators), or a lower than expected ramp of Core 2 (also not expected). Actually, if anything, AMD's warning suggests even better news from Intel this quarter. Tomorrow might be a good time to buy some calls.
I think Core 2 is doing just fine, I just worry about ASPs. If Intel experienced same or higher unit growth as AMD without falling ASPs, there is a potential for Intel to have a great quarter.
Joe