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.
About this "price fixing" scheme:
Threejack,
you do notice that I was right all the time and that it could only have worked if Samsung was involved.
Anyway, whether true or not, it is time the MMs put this thing behind them, but maybe they do not control the pace. (DoJ most likely does)
It has little or nothing to do with Rambus other than to put a (alleged) light on how the MMs or in fact most large companies operate.
I have noted over the past years, that it is most popular to call certain memory makers thieves. In this light so is Intel (one example only: the Intergraph case), so is Nokia, Ericsson, Samsung again (diff. division) : the IDCC case. AMD and Intel (vice versa) have been in each other's hairs many times and the list goes on.
But anyway, back to the subject: it would be good if something definitive is done re this price fixing allegation. Hynix already had the americans and europeans on it's neck for selling too cheaply (dumping); I have not checked the exact periods, but would not be surprised if it is the same period which is in play here;)
Cheers
Cor
Morgan Stanley Semi Conference
Yes Threejack, I was just automatically alerted about this participation in the MS conference;)
You are right, different people see different things in the same information. (called filters) That is fine, especially if they speak up and bend the other people's filters a bit;)
In this case, the CC was scheduled at short notice just after a major legal event, of which the content was very well publicized anyway and therefore one would expect more than a synopsis of that event. Even though I myself did not expect so, I am not surprised that many people expected major news in the form of settlement or other litigation news.
In the financial paragraph they also clammed up, which means Rambus is still stuck at 5-7 cents eps per quarter, which is not news.
The fact they do present at the (pretty huge) MS event, is even more reason NOT to have a CC, as that presentation,(which hopefully is mostly of a market oriented - financially interesting - nature) will be webcasted as well. Of course, shortly thereafter, the silent period will start until earnings in april (amazing that we are already so close to that).
Anyway, we'll be at the MS CC:))
Cheers
Cor
Underwhelmed by this CC
I don't know about y'all, but I was very underwhelmed by this CC.
First of all, they did not tell anything that was not known before the CC but secondly they gave no guidance at all, either on legal issues or financial ones.
More or less taking the fifth on everything.
I may be misreading this, but the loud silence on boards may also mean that I am right here.
OK, tomorrow a new day.
Cheers
Cor
PS I am sorry I stayed up for this "live performance"
What happened to cordob?
Back from Spain, dog tired.
Did not notice the quake , Threejack, but it is some 2000 km away still.
I had a strangle on before the decision and made a profit, not spectecular, but OK. Got left with some march 25 puts, which I have for free now. Maybe they will become something yet;)
Cheers
Cor
PS I have posted at the Fool about the new document at the EPO site. No time for an in-depth study though. First to update my site.
jre, I know this is totally OT but forgot to put this in before my edit time ran out.
I am all for the phonetizising (??) of American English. English is funny in that the same spelling does not produce the same sounds. So maybe we should go "thot" instead of "thought" etc etc.
And this is exactly why "manoeuvre" became "maneuvre", phonetic simplification. The reason why the "old" spelling stuck with me because the French origin of the word is clearly there:
man-oeuvre: work of the hand
Originally from Latin: mani operari --> man-oeuvre --> maneuvre
OK, totally off track but this is how a linguist's mind works and why he so easily learns a new language:)
Cheers
Cor
Hey jre, I did not say you did it wrong.
I know about Websters, I am multilingual and I write and speak a different type of English with Brits than with Americans.
I just pointed out that I was not wrong. There are two rights about this. I don't write harbour, but harbor, but if I write to an Englishman I write harbour. Etc etc.
I am moving from Holland to Spain, Catalunya to be precise, above Barcelona. I will learn to speak Catalan as well as Spanish, to the extent I can.
We can still understand most of it as long as it's written down, but sometimes when they speak we have a hard time keeping up.
LOL, yes they are mumblers.
Cheers,
Cor
Chickens and eggs:)
We definitely agree 100% here:
Put another way, there can be no dispute chickens and eggs are related.
On the Wallmart and teh supermarket strategy, it is basically the same.
Your Supermarket Strategy reminds me of how Walmart opens a store in a community, offers the lowest prices in town, and several small businesses are shuttered as a consequence, victims of capitalism and convenience. Only difference is, as a barrier to future competition, prices stay low ; )
True for the supermarkets too, but they tend to make the blow worse at the beginning by running a lot of loss-leading promotions on staples. These promotions go when the competition is dead.
It also works on a larger scale. In the 70s, my research for my then employer (one of the largest confectionery companies in the world) found that a competitor was going to launch a new competitive product. Our company managed to stuff the channel and the trade had no money left to buy more chocolate (which is expensive) and the product intro failed. Product withdrawn after a while.
Now I am off this thread util we get some really informed comments from Nic for example;)
(although we weren't doing too badly:)
Cheers
Cor
Threejack, the civil antitrust suit (potential)
All the above, just the opinion of a Legal Lamebrain.
LOL, a legal lamebrain and a legal nitwit theorizing on a legal strategy. Well at least we cannot be accused of misusing our knowledge:) Maybe Nic can help us here with a clearer insight.
Unless the DOJ files criminal charges against one or more of the Thieves after its investigation, and Rambus can turn one or more of the MMs prior to filing its threatened civil action, it simply won't happen.
The turning sounds like a good strategy, if it works. The former statement, I am not sure of that is a necessary condition. On the one hand evidience of collusion at some other time for a different purpose might look good in a civil suit, on the other hand it would not prove much.
The problem as I see it (let's assume that both conducts actually happened, for the sake or argument), is that the first case in time(Rambus) is about amigos(more than 3) keeping the price of sdram/DDR artificially low by making too much of it with the purpose of creating a barrier to entry for rdram and the other case (DoJ) would be about more or less the same group keeping the prices artificially high for the sake of profits. (this latter thing is more difficult to do, if somebody builds out a bit too much capacity, they can undercut and still make a profit and poof! there goes your scheme.)
The whole strategy is what I would call the "supermarket strategy". New supermarket comes into town, sells products at a loss until teh little guys are all broke and give up and then raises the prices again. (happens all the time)
We need a lawyer here:)
Cheers
Cor
More about the antitrust
Threejack, you seem to be more sensitive about this as usual. I am just trying to analyse the situation. That comment in the 10K apparently has been there before and is not very strongly worded.
If Rambus makes good on their antitrust threat, the only maneuvering to be done will be among the Three Little Piggies to see which one can race to the pen first to sign on the dotted line.
It won't be only about Hynix, IFX and MU. If it holds at all (very doubtful imo), Samsung is also involved. Then Rambus would have a major problem: aleniating their best customer by suing them, not a good idea, I would think.
Huh? You drinking?
Yes but not much. I typed too fast so lost should be lots.
Manoeuvre is actually the correct spelling as per the Oxford dictionary which I just took out of the cabinet!
And so, Rambus must wait to learn what the DOJ decides.
That is an unrelated case vis-a-vis the suggested antitrust case in the 10K.
Cheers
Cor
Antitrust suit possibility: The actual wording in the 10K is:
(page 73)
In addition to the above, there is at least one additional area where Rambus foresees potential litigation activity, perhaps in the next quarter. Rambus is now evaluating what antitrust or other remedies Rambus may pursue based on evidence indicating that there were concerted efforts in the mid-to-late nineties to deter innovation in the DRAM market and to deter market acceptance of Rambus's RDRAM product. Rambus has not decided whether to pursue such litigation and there can be no assurance that such litigation, if brought, will be successful.
Lost of room for manoeuvering.
Cheers
Cor
I can't believe that nobody today is tuned in to the major news re the EPO???
Cheers
Cor
And how about this eye?
(you all need some diversion:)
from:
http://britneyspears.ac/lasers.htm
Cheers
Cor
Found another nice eye,
how about this one?
It is becoming fashionable, doesn't it. Think I will have my own eye.
Oh, btw the press release with this eye was:
http://focus.ti.com/docs/pr/pressrelease.jhtml?prelId=sc04029
Hmm, on 2nd thoughts, driving a PRBS pattern? (Public Relations Bull Shit pattern?)
Cheers
Cor
EPO patent neutering.
I don't know about motive, Threejack.
I know there have been almost a 1000 pages of written pleading mostly from Hynix (Opposing #3) to convince the board of it and that they were leaning towards the opposing viewpoints.
Nic also looked at it at the time and did not like the looks of it.
Well, I think Arno took the usual things which are used in infringement about Thiele's translation. Remember this is NOT about infringement, this is about the patent on which eventual infringement would be based. If the patent goes or gets changed, Prof. Thiele's work goes out the window or need to be redone on the new claim language.
I have no time to do a real translation of it, but why do you think Hynix made it public for the first time? Because they could use wording from it.
Not optimistic about this outcome, I am.
Cheers
Cor
That paper contains the "markman" if one can call it that in Europe by professor Thiele (Swiss), which apparently has become public recently.
starts at page 13
Cheers
Cor
O boy:)))
Cheers
Cor
This one should work then:
Hey Elixe and Calbiker, how about this one?
Nice eye. Somehow I cannot get it to display here, the link is:
http://www.uxpi.org/images/front_page.gif
see the uxpi.org site at: http://www.uxpi.org
Cheers
Cor
Tomorrow first day of oral hearing in München at the EPO panel.
On the EPO site (linked above), which is down a lot, there is a fairly new 155 page document from Hynix (Opposant 3) dated 2nd febr.
It's in german, I started reading but had to do some trading, but will get back if there is interest.
Cheers
Cor
Know about http://www.chipstocks.net?
Yes Elixe I undertand it now (always did about PCI-Express, but was thinking of the many possible parallel bitpaths of Hypertransport when I called that parallel).
I have worked in the far distant past with UARTS and USARTS and they are serial too, very much like serdes in fact. Surprisingly I saw last week Philips brought out another improved UART, so this stuff still gets used.
You can add lanes in parallel, but they're all serial.
But it looks like you understand that now.
Yes. I do believe HT handles that more efficiently than PCU-Express, but have not studied it siffuciently to know for sure.
Cheers
Cor
Ha Elixe, Cal will now (correctly) explain to you that PCI-Express is serial.
Now I have a system with minimum interconnects to the controller. It's expandable both within a sub- system and on the controller bus. The sub-systems could use the same layout and be placed anywhere. They could even be on different PCBs.
A lot of advantages at the cost of latency. (Starting to look like Intel's fully buffered DIMMs, though).
You get protocol overhead, like in Hypertransport. Btw Hypertransport bridges have a fifo memory buffer (and prioritized asynchronous traffic), so they could deal with this. Bandwidth would be OK, length if bus also more than enough (maybe this is why CSCO was one of the early HT consortium members), but latency is a problem.
I proposed this for level 4 cache years ago on the TMF Rambus board; just cannot find the link now.
Cheers
Cor
Checked asp s qtr4 relative to qtr 3 approximately
Very rough, because I kept the mix to the qtr 3 mix, but indication about - 7% q-on-q
Cheers
Cor
Elixe: re the sdram market share
Also notice the change in DRAM composition. SDRAM is growing!
Are these volume or $$ percentages? If volume it would explain why sdram is more or less the most expensive dram now (I assume you get the week dramexchange newsletter, otherwise I can fwd to you)
Expensive sdram is good for MU, because they have always been strong in keeping capacity for legacy ram.
Last quarter they reported $6.00~6.50 and expected slight ASP growth in Q4.
I think ASPs were down in Q4.
I am not sure that is the case. But would have to check.
I seem to remember from the Infineon Ccs that fully loaded cost was approx. $ 4 6 months ago and later around $ 3.50. So they are making money.
Cheers
Cor
Hi Elixe, on the Samsung FAQs
The corporate demand may start to kick in in 2H'04 as the DDR 2 adoption increases.
I guess the changes of mobos, DDR to DDR2, sockets of P4 do not really help in the short term. A certain amount of "osbourning" does occur in such a changing (for everybody to see) environment.
Cheers
Cor
Cal, I am now totally convinced you do not see the big picture re busses etc. So I will stop sparring with you, it's a waste of time.
I admire Elixe's persistence though:)
Calling Hypertransport a serial link is the limit!!
http://www.hypertransport.org/pr_092903.htm
more details on that site
You can still get a royalty free license:)
(except for things like cpu to cpu xfers, you will have to pay AMD:)
You're backing the wrong horse, bud.
How do you know which horse(s) I am nackering, oops backing?
Your rhetoric also seems to assume that Rambus will be in every PCI-Express. So far I have only seen production chips which don't have rmbs inside, but I may have missed a little pony here or there:)
Careful with the biking, don't think about horses (or lions).
Cheers
Cor
I am well aware of the specs of PCI-Express. It is what is called a mezzanine bus. Now look inside your PC, how many PCI slots are filled? In my PC none, I only have an AGP graphics card.
So comparing PCI-Express volume with what has happened so far with PCI, will lead to disappointing results.
Yes I knew that link from Rambus too. Have they sold any? They are showing "proof of concept" later this month at Intel IDF.
Hypertransport only runs at 1.6 Gbps per lane. It's way slower and requires dual clock signals. You don't need any clock lines with PCI-Express.
yes serial busses have their role too. Also in serial ATA etc etc. There are more busses than there are bikers in California, ha. And you know very well that 1.6 Gbps per lane is very fast for a configurable multilane bus and the clocks are not added for every data pair!
Cheers
Cor
Hey Cal, still have not read the Hypertransport whitepapers, I can see;)
The original Hypertransport already could service several PCI-Express links/tunnels plus AGP8 plus plus.
Afaik the higher speeds are meant for the cpu to cpu links.
Cheers
Cor
Faster Hypertransport,
picked this up at the AMD TMF board, peripherally (!) of interest.
Chips to ride on faster HyperTransport
By David Becker
CNET News.com
February 3, 2004, 5:46 AM PT
...
"The new version will be up to 75 percent faster than the original specification, according to the group, and will support PCI Express, an emerging high-speed standard for connecting PCs to peripherals and each other.
As previously reported, HyperTransport 2.0 promises to dramatically increase the speed at which data can be shoveled between chips. A 32-bit processor using the new standard could exchange data at the rate of 20 gigabytes per second, compared with a maximum rate of about 12.8 gigabytes per second for the current version of HyperTransport"
http://zdnet.com.com/2100-1103_2-5152384.html
the standard can be found at http://www.hypertransport.org
I am still waiting for somebody to get the idea to put an extra memory controller on a HT link:) (a bit of extra latency maybe, although on the A64s the HT link is inside the cpu which helps)
For Cal: Hypertransport uses LVDS, but you knew that already of course:)
Cheers
Cor
Good link, Elixe
I see from:
http://www.dhwg.org/press/2003_06_24
That amongst the founding members there are Philips and Samsung.
Fujitsu, Gateway, HP, IBM, Intel, Kenwood, Lenovo, Microsoft, NEC, Nokia Panasonic, Philips, Samsung, Sharp, Sony, STMicroelectronics and Thomson to Establish Design Guidelines Based on Industry Standards
Samsung working on TV-on-a-chip, Philips with already with a lot of the products, which comprise such a digital home and the chips for it (TV chips, DVD chips etc)
On page 11 of your pdf one gets an impression of which direction the interconnects may take (1394, UPnP)
Honestly, I do not see a cell role in there (or it must be Bill Gates' house:)
see today's PR from Philips e.g:
http://www.philipssemiconductors.com/news/content/file_1033.html
(single chip for LCD TV)
Cheers
Cor
The "...Pentium of the consumer electronics world" sure has a nice ring to it. Uses Rambus IP, too. Maybe Slide 82 is not so far fetched after all??
Hey Threejack,
Naturally that is marketing speaking. Not abundantly clear what they mean by that. In the embedded 16/32 bit world ARM cores are leading (almost 1 Billion in 2003)
Note that the average price/royalty of these cores is about $ 0.09 (9 cents). Billions times 9 cents is still quite a lot, but not those numbers from the slides.
If you look below the ARM cores, you have 4/8/16 bit, but mostly 8 bit (e.g. 8051 based) cores. I don't know the total number of those but that is a much higher number. Somehow I doubt that all of these will go the higher node (how much power do you need in a thermostat or a toaster:)
As an example I have seen models from Philips which include a 8051 (=8 bit) core plus all sorts of interfaces incl. sdram and communication and parallel with a list price for the whole thing of $ 0.50.
I wonder if Intel gets some royalty out of that type of product. I never see a mention of royalties in Intel's reports.
Btw I saw the number at IFX for the last FY for royalties was up to some $ 200 M (from $ 100 M the year before); this confirms the tendency to outsource, I think.
Cheers
Cor
(am I glad there is no politics/religion here, pffft)
Hi Threejack,
Still say, it's a big deal.
Why that? Because you think it would command a higher percentage royalty if it did? I find that doubtful. The royalty rates do not really have a relationship to the percentage of IP involved in the chip. There is a lot of IP involved in making a memory chip and if it all went at the rate of 2% for a few features, the total royalty payable would be well over 100% and you know -as a good capitalist- what that means;)
What would be a big deal, would be if Samsung said, hey DDR2 is not DDR so we don't have to pay you DDR royalties. That would be BIG, but somehow I cannot imagine that scenario.
Cheers
Cor
Oh I thought you had seen my posts showing how many units were qualified by Intel (DDR2) some time ago from all the major manufacturers, which includes Samsung.
Understood Tate to say on the last CC existing DDR licenses do not cover DDR2 Rambus IP. Did I hear him wrong? Don't think so.
Even if so (I think he was a bit vague on the issue), why can we not identify with all the clever heads on all these boards together what would this additional IP be?
So, has Samsung signed a DDR2 license and has Samsung been paying DDR2 royalties?
NO, it would have been widely publicized by Rambus, so assume this did not happen. However, Samsung will be paying DDR royalties (whatever the -maybe still reduced rate- deal is) and imo this is the correct way.
So I understand what you are saying, but looking at the matter w/o emotion I have to say that the situation as far as I can see is normal. DDRx are evolutionary developments from DDR, done by the JEDEC partners incl. Intel and that's it.
Cheers
Cor
Threejack, DDR2 from Samsung
Didn't think this was much news.
Seems to me Samsung needs a DDR2 license, and Rambus would need to announce it. It's material because Samsung is a current DDR licensee and as it moves production to DDR2, it impacts an existing revenue stream to Rambus.
I don't know and am very unconvinced actually that is the case. What is in DDR2 which is not in DDR and is supposedly rmbs property? I don't know the answer to that and it has been raised here before w/o a clear resolution.
P.S. One cent ticket one way to Spain? As my younger friends might say, "Schweeeet". Of course, being young and impressionable, they may neglect to ask the all important follow-up question: How much to get back?
Absolutely, but I booked the return for a cent too. The airport taxes total € 30, so not too bad for € 30.02 each way:)
Cheers
Cor
Threejack, sure I can translate when it happens, but not literally as there will be reams of pages.
I hope it is up on the 11th or 12th, because as of the 17th I am going for a week to Spain. (incommunicado)
(bought some plane tickets from Ryanair for 1 cent one way:)
Cheers
Cor
Consolidation of dram manufacturers
(from dramexchange)
Consolidation of DRAM markers: Top five takes 80.90% share
I had to hijack the picture as it would not display directly...
Cheers
Cor
Sorry Threejack, I was not lowballing for 2004, because I was talking about the "present" level, say 2003 or even qtr4 2003.
30 cents for 2004 sounds OK to me as an estimate.
Cheers
Cor
Titan, in my opinion it is Vegas time for Rambus
Upgrades don't hurt in the short term, but what is really needed is some positive resolution of licensing for DDR. That might produce a nice pop (could be temporary if you do the math); if nothing like that happens, the p/e is simply too high now.
Say about 20-25 cents profit bottom line now (was a bit lower than the year before, which also does not put a smile on people's faces) is not enough for $ 30-35 share price.
Needs to at least triple.
You say the legal stuff stopped the stock, on teh contrary the legal stuff kept the stock going, otherwise it would be at $ 10 now. The question is (apart from whether they will win this), how much is already priced in?
I am cautiously pessimistic on this.
Cheers
Cor
As Threejack will confirm, there have been times we did that here too:)
But the "bus wars" have temporarily taken first place.
Go ahead, discuss, I'm game:)
Cheers
Cor
If I read this correctly Rambus only get two days to produce these documents to Payne. Is this normal? Or is this the "rocket docket" stuff?
Cheers
Cor
Elixe, re the IFX <> SMIC arrangement
I believe it is very similar to the ProMos agreement, but much bigger in scope.
see this release (hope the link works, otherwise go to press, march 27 2003)
http://www.infineon.com/cgi/ecrm.dll/jsp/showfrontend.do?lang=EN&content_type=NEWS&content_o...
Infineon Strengthens Foothold in China, Expands Foundry Agreement with SMIC
2003-03-27
Munich/Shanghai, March 27, 2003 – Infineon Technologies AG (FSE/NYSE: IFX) is further executing on its sustained commitment to the Chinese market. Infineon and Semiconductor Manufacturing International Corporation (SMIC), Shanghai, China, today announced that they have signed an agreement to expand their cooperation on the production of standard memory chips (DRAMs). Under the terms of the additional agreement, Infineon will transfer its 0.11-µm DRAM trench technology and 300mm production know-how to SMIC. In return, SMIC will manufacture products in this technology exclusively for Infineon.
This new move will enable Infineon to increase its overall capacity by approximately 15,000 wafer starts per month through SMIC’s 300mm production plant, which is currently being built in Beijing. In December 2002, the two companies signed an agreement whereby SMIC would manufacture memory chips in its 200mm plant in Shanghai using Infineon’s 0.14-µm DRAM trench technology exclusively for Infineon. Following the ramp-up of SMIC’s Beijing facility, the addition of the 15,000 wafer starts per month in 300mm technology to the 20,000 wafer starts per month in 200mm will result in a total capacity of 58,000 wafer starts in 200mm wafer equivalents. First products from the new 300mm plant are expected for summer 2004.
“Through the extension of our cooperation with SMIC we can grow our DRAM business without having to invest in production facilities,” said Dr. Harald Eggers, Memory Products Group CEO at Infineon Technologies AG. “At the same time we are further strengthening our regional presence in the promising market of China and aiming overall at a leading market position in Asia/Pacific.”
“We are pleased to expand our partnership with Infineon,” said Dr. Richard Chang, President and CEO of SMIC. “SMIC’s dedicated quality foundry service will provide a win-win business and financial solution as the semiconductor industry moves towards greater outsourcing of production.”
This collaboration will further consolidate Infineon’s position as the third largest semiconductor manufacturer in the DRAM market and further establish the company as a leading provider in the growing Chinese market. According to forecasts by market research firm Gartner Dataquest, the semiconductor market in China alone is set to grow from approximately US $16 billion in 2002 to around US $31 billion in 2006.
Cheers
Cor
Well with the EU and the USA both with high tariffs punishing Hynix, they had already said they would turn to the Far East mostly.
I believe the Taiwanese already made noises also.
Cheers
Cor