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.
Hey Nick,
I know from careful study of IFXs annual reports that they already collect about $ 100 M+ a year on dram royalties. This may have something to do with their "superior" trench capacitor technology and probably mainly comes from the Far East.
The main deal they have with SMIC is a technology for capacity deal, which I think is good business wise for them. Reason: on the next downturn in the dram industry they can idle capacity at no cost; now they were forced to keep producing at a little above direct cost, but below fully loaded cost. (including depreciation)
It will not be an "out" from paying for other IP of course because they will be branding and selling the stuff SMIC makes.
Cheers
Cor
Yes Threejack, I know about the wto treaties etc, after all there had to be some give for the most favored nartions status etc;)
However, just assume (for the sake of hypothetical argument), that Rambys wins the infringement in the USA (some time), but loses at the EPO in Germany. (a possibility) The WO patent derives from the EPO patent afaik, so then the whoel Far East also topples.
It would certainly weaken Rambus' negotiating position in coming to an agreed royalty amount. After all, it is all only about money, all the hoopla notwithstanding.
Cheers
Cor
I think that's because you have the image in your browser cache.
Elixe, that is not what I meant. True after the first time, and assuming the cache is not flushed (I do clean it out regularly, but not within a session), the next times it is simply loaded from cache.
What I mean however is that when one loads (also the 1st time), ihub simply sends the some 50byte string: actually they work on it a bit like:
<img src=http://www.chipstocks.net/samsung/bssoweb.jpg>
my site then has to cough up the picture.
Btw they are down again, I have had more trouble with them the past week than the whole last 3 years, must complain again and do something about it.
Sometimes doing business with a party over the internet is like running into a rubber wall repeatedly, (although I have no personal experience of that;)
Cheers
Cor
Hi Threejack
saw your post on the semi board at TMF.
on
Don't really care where they produce it or sell it, as long as they pay Rambus a royalty when they incorporate its IP in their product.
question is how far are the Chinese advanced in accepting IP from USA? There are treaties, but ... I don't know really. If they made it there and sold it there?
Cheers
Cor
(PS IFX also only sells 20% in the USA)
Threejack, LOL on the sauce quote :))
You know iHub has done this cleverly; they spend no bandwidth on these pics, the owner of the site (for example I) does. They just use the link to put it up.
Like Elixe's (samsung) pic is some 50K bytes but iHub only xmits about 50 bytes.
talk about bandwidth widening :)
Cheers
Cor
btw this was a wafer photo I found somewhere and manipulated a bit 3 years ago; I thought chipstocks could only go up:)
Btw somehow I believe they are using the wrong technology to put things in this pdf. Some slides really load extremely slowly.
for example page 11
(using a 700 MHz Duron, not the fastest but usually fine)
Cheers
Cor
Maybe not quite the place to put this, but why not, lots of technology in this MU presentation of today.
http://download.micron.com/pdf/presentations/analyst/Analysts_SV_012304.pdf
warning: very biiiig pdf. (12 MB)
Cheers
Cor
I see it fine now, but there have been problems with my webmaster:( these past few days. Off and on because of dns server stuff. maybe should ask some money back.
Btw case is important that is a unix server, rather than a windows one.
I will put a copy on www.markt21.net/bssoweb.jpg
I know that site is "troublefree" although the other one should also be, see my sign off:)
Cheers
Cor
Hey Stowboat, buy you a sail and some mast material, saves a lot on gas:)
Cheers
Cor
OT: Thanks for that story Cal.
Amazing that one does things like that voluntarily:)
But, seriously, I do understand. My thing is sailing (with a proper yacht) in a near gale or a gale. It is something you never forget: the way the sea builds up, the boat moans (but I know she is more than strong enough, it's the people who are the weakest link), the whitecaps, the rollers, the way one has to switch off autopilot and steer by hand strapped to the boat, meet the waves ugh how can I decribe such experiences to anybody who has not been there?
(certainly my wife does not understand the fun of it)
I have been in force 7 many times (cool), in force 8 maybe 8 times (scary sometimes if there are sandbanks nearby and before GPS navigation became a real problem in driving rain etc), in force 9 three times each time in the "Golfe de Lion" between France and Barcelona which were the really great experiences. The deck of my ship is 2 meters above the water, which is high. The waves however, which were following at some time, were 5 meters high with curls at the top (foamy); scary but great.
Those bone breaks... I had not much experience of them only once broke my ankle parachuting in 1968, but I have a friend who likes motorcrossing and was National champ several times and European champ once; he has 92 breaks in his bones, mostly ribs, collarbones, arms, legs. He kept going when the # of break was over 50 because "I heal very fast"; his wife made him stop at 50 or 52.
Yes in your story: water is the most important thing. Sailors also sometimes forget. On a yacht it is easier, we have big tanks and never run out. Did you ever read those survival stories of shipwrecked who had to distill sweet water out of salt and suck the inside of fish they caught by hand (the inside is sweet water)? I studied that stuff for training purposes only (because I was making multiday crossings, up to 6 days), but then I studied some more of how to avoid such situations.
XCheers,
Cor
(have to get back to the boat...)
Light in the tunnel for semis
This sounds very much like something I wrote in nov. 2002 :)
See for yourself.
http://www.reed-electronics.com/electronicnews/article/CA376283
Look Who's Buying -- Finally
By Susan Mulcahy, VP of research, Reed Business Information -- Electronic News, 1/22/2004
There is light all the way through the tunnel now. A survey of purchasing managers shows 65 percent expect Q1 chip purchases to increase quarter-over-quarter, while only 5 percent expect component purchases to decline. This suggests better than normal Q1 seasonality, in our opinion.
The survey, conducted by Reed Research Group and Smith Barney, shows that 56 percent of respondents expect lead times to increase in the coming month, about the same percentage as a month ago. High performance analog (HPA) and discrete components showed the strongest product-specific lead time trends, which also is translating into higher prices for these component types.
Interestingly, more purchasing managers are working to reduce chip inventories than increasing them (30 percent vs. 19 percent), which indicates that a lead-time induced chip inventory restocking wave may still be ahead of us.
Overall, the survey results bode well for Q4 chip results and Q1 guidance, especially discrete companies such as Fairchild and HPA suppliers like ADI.....
So far so good. Guidance good for qtr1. I will tell the sad AMDers:)
So now to my little piece:
here: http://www.chipstocks.net/topten/chipmarket.htm
quote:
Opinion: (Nov 22, 2002) There is light at the end of the tunnel. In fact it is visible and has been getting brighter for quite a few months now. This opinion can also be downloaded as pdf file from our download page.
Ok in my piece (14 months ago) it was only at the end of the tunnel but already brightening and in Susan's case it is light all the way through the tunnel so she must have sawed some holes in the ceiling, but I got there first:)
<end of chest thumping>
Cheers
Cor
For just a moment, thought Calbiker was going to have a field day ; )
Well for one thing, we want him to check in daily with his broken down oscilloscope so we know he hasn't been eaten by a mountain lion during a field trip:)
Cheers
Cor
You know Threejack, that is a very annoying way of doing the dates:(
I am early for me, firing up the computer and updating left right and center.
Cheers
Cor
Btw, I am confused as to what the date on that article is.
I have today's date at the top, but it also talks about the 03 spring issue, so this may be old hat ??
Cheers
Cor
CELL: Elixe what is your take on this and are there repurcussions to your ideas about Xdram adoption?
http://www.designchain.com/coverstory.asp?issue=spring03
"These problems may be a few years out, though. IBM has not announced when it will release the final chip, but the Cell project was originally envisioned as taking five years, suggesting that the final product may not be ready until as late as 2007. Although Sony refuses to talk about its future plans for the chip, it does admit that the Cell chip will not be the CPU in the Playstation3, reportedly slated for release in 2005."
???
Cheers
Cor
not be more expansive to make than DDR
should read:
not be more expensive to make than DDR
(edit time is over:(
Cheers
Cor
Well Elixe, that table is one to keep.
What I keep wondering coming away from that, will there be a DDR2A using the 200 Mhz core speed. I mean if they have the faster core, why not use it? This may be because it runs into the speed limitations of the fsb and the interface.
Otoh maybe they are keeping something there?
From this I also surmise that DDR2 will not be more expansive to make than DDR. So a possible price premium would only be a function of (temporary) shortage of supply.
Cheers
Cor
Hey Elixe, did you ever look at Hypertransport?
Just a bit off the memory track, but not really all that far. There are similarities with XDRAM, LVDS for example and a way to minimize clock skew.
I was looking up something on my hd about it and ran into this excellent 2002 presentation, especially good for its illustrations. You may want to take a look at it.
http://www.amd.com/us-en/assets/content_type/Additional/26635A_HT_System_Design.pdf
Note that the opteron has three Hypertransport links on its cpu die. (two can be used to make a neat 4-way interconnect, leaving one for IO tasks)
A 16bit HT link (they can be configured in different widths) has a max bandwidth of 6.4 Gbytes/sec. Mezzanine buses like PCI and PCI-Express can easily be connected through bridges and tunnels.
I wonder when somebody will get the idea to make a memory controller on the HT bus; latency would not be all that great, but the HT bus has good provisions for streaming data of different speeds. (even if mismatched with the system clock)
let me know what you think.
Cheers
Cor
Threejack,
I have JESD-79-2 pdf on my hd here.
I think Elixe has summarized the points of interest.
Maybe the additional cas latency is something Rambus wants to pounce on.
Cheers
Cor
Hi Elixe, re DDR2 and IP
I read comments like:
Rambus also indicated that there currently are no valid licensees of DDR2. DDR2 contains even more Rambus intellectual property than DDR which could possibly be addressed in a higher royalty rate requirement than DDR.
What is the basis for this thinking? What is there in DDR2, which is not in DDR and could be claimed as Rambus IP? Any ideas?
Cheers
Cor
The R659 is missing from the list. Even if the R659 is PCI-enabled as well, Tate did not strike me as especially confident about the prospect of mass production when he said samples will ship this quarter, but mass production will depend on demand.
Threejack, I agree and also caught those vibes in the CC.
Wasn't Elixe saying it would not appear in a mobo? (losing track here in all this information overload:)
About the SiS965L with PCI-Express:
do we know whether there is Rambus design work in the PCI-EXp part? Is SiS a licensee? (no guarantee, but a necessary albeit not sufficient condition)
I heard S3 graphics mentioned in the CC.
Cheers
Cor
PS had expected Sandy to come to the rescue already:)
Bob,
just after you posted AMD went out of the starting blocks, now suddenly up 0.60.
Cheers
Cor
My estimate is 19 cents for AMD (qtr4), at the Fool AMD board the average estimate is 21 cents and highest 35 cents I believe.
Analysts at 4 cents.
Hmm, Target for stock? Leave that for ZEEV:)
Cheers
Cor
(loooong AMD)
Nic's Claim Comparison
Threejack,
This post on yahoo from Nic also deals with the issue:
http://finance.messages.yahoo.com/bbs?.mm=FN&board=4687909&tid=rmbs&sid=4687909&acti...
The "biggest" of these mysteries surrounds the'804. If this patent was really found not infringed on two separate grounds, including the "integrated circuit" language (which the CAFC DID review) and the "first" and "second" clock issues, how the h*ll did Rambus forget to appeal the second grounds? It makes no sense to dedicate a bunch of pages in the appeal brief to one point, and then not argue the remaining point somewhere.
I mean, THAT one looks like a screwup. The rest look like calculated choices. I think Payne is going to kill claim 26 of the '804 on SJ for exactly this reason.
Da Greek
Cheers
Cor
Threejack,
This was the post on TMF which started it.
http://boards.fool.com/Message.asp?mid=20149135
I will not reproduce the whole post w/o Nic's permission, but this snip is very interesting:
(5) On the other hand, Rambus COULD turn this situation around, by pointing out to Payne more CLEARLY what he SHOULD have done in response to the CAFC ruling. Rambus should tell him that he (Payne) needs to work backwards, by examining what terms the CAFC ruled on, and then looking to see which of the 57 or so rejected claims contain the language in question. The one caveat is that if Payne rejected a claim based on another limitation as well that was NOT argued to the CAFC Rambus would need to forfeit that claim. I've not read every transcript, so I don't know where that would leave us; but I bet its a might more than 4. Please appreciate that there are some claims which are just not that great, because they just don't read well on a DRAM, so the total number is probably still less than 20.
(6) The screwiness of Payne's approach is apparent if you look at claim 26 of the '804 (which he is apparently letting in) and claim 15 of the '214. As concerns the specific limitations argued by Rambus to the CAFC, these claims are essentially the same. If Rambus did not argue the dual clock for the '804 (perhaps they did not care because their proof is ok either way??) why is claim 15 of the '214 any worse??
If I were Rambus, I'd start with claim 15 of the '214, and show Payne the fallacy in his thinking. If you win on just 1, he might begin to see the problem with the rest.
Details about this were put into a pdf by Nic, which can be found here:
http://www.markt21.net/Claims-comparison-frm-NicdaGreek-04jan14.pdf
This is apparently not being seen as important by the Rambus legal team, if you listen to the CC. (Craig-Leapof Faith has sent this pdf to Danforth early yesterday btw)
I am not sure if I am the right person to pick at Nic's analysis (I am indeedo the legal nitwit), but why not ask questions on this thread directly to Nic?
Cheers
Cor
Elixe, when I was brushing my teeth this morning it also occurred to me this might be the solution. I will replay the tape recording I made later on to check what they said exactly.
Adjust rdram to 2.865 (because that was not given precisely) and we got what we need. (good for future reference:)
Can we now check our earlier calcs on royalty rates?
Cheers
Cor
Hi Tim
We (3Jack and I) were not being sloppy. probably the newsservice.
We did not mishear anything, the quote as 3jack had in his post was on the net, I saw it myself. It was changed later.
I think you are both being sloppy. $20billion/~100million shares = ~$200/share.
Exactly why I was commenting on it.
Here is the precise quote from post # 1354.
Should Rambus prove that the memory makers have violated its patents, royalties alone could amount to $20 billion annually, or $2.50 to $3.00 a share, said Erach Desai, who follows the company for American Technology Research. Desai, who made his remarks during an interview on CNBC, said earnings of that magnitude imply a stock price of about $60 a share.
Desai said he is optimistic about Rambus' chances in court. (ATR does not have an investment banking business.)
http://www.thestreet.com/_yahoo/tech/billsnyder/10137240.html
That was the way it was on the net. (before it was corrected)
Cheers
Cor
Threejack,
I knew somebody would come up with highest revenue and lowest costs, LOL;)
I bet higher than that. Given the Velio charge, let's say 6 cents. You can remind me if I'm wrong.
We'll know about 15th april then:)
It is not just the Velio charge, Bob Eulau clearly said "temporary lull in litigation expense" (back to what it was is good for -2 cents)
Cheers
Cor
PS do you think we should post the link to Nic's document about claims, which I put on one of my sites, here? There were hundreds of downloads so far, but little discussion because people do not understand the importance.
(I am sure you have seen it on the Fool)
Threejack, yes higher royalty revenue but also higher costs:
your So, a 15% increase in 4Q PC shipments, mostly using DDR, should translate to higher royalty revenue for Rambus next quarter.
e.g. take 30 M (lower end) minus 22 M = 8 M operating income which is less than this quarter's o.i. of 11 M.
Or take mid 32.5 M minus o.e. mid 24 M equals 7.5 M is worse.
Take my word for it (they may be lowballing Intel style), the eps for qtr 1 will be something like 4-5 cents.
Cheers
Cor
PS they can't push slide 82 too much otherwise somebody may go to prison;)
Threejack, the article has been corrected (at yr link)
"Should Rambus prove that the memory makers have violated its patents, royalties annually could amount to 2% to 3% of the entire DRAM market, said Erach Desai, who follows the company for American Technology Research. Because the DRAM market is now valued at about $18 billion, those royalties would add from $2.50 to $3 to earnings per share, he said."
Cheers
Cor
Ok, Elixe, 2% blended for dram seems OK (for the part of dram Rambus gets ....)
What is your take on those "unexplained royalties" ?
Cheers
Cor
Well Erich Desai went a bit too quickly after that CC;)
Snip:
Should Rambus prove that the memory makers have violated its patents, royalties alone could amount to $20 billion annually, or $2.50 to $3.00 a share, said Erach Desai, who follows the company for American Technology Research. Desai, who made his remarks during an interview on CNBC, said earnings of that magnitude imply a stock price of about $60 a share.
Either he was misquoted in the article or he has gone off the deep end.
$ 20 Billion would be almost $ 2 a share (a bit less as the diluted shares are increasing rapidly) minus tax some $ 1.20 a share left. Also $ 20 Billion would be about 100% of the dram market!!
Very sloppy stuff indeed. (somewhere)
Did you also note, Threejack, that the common supposition that XDR would have higher than 3.5% royalties was sunk in this CC?
The reply was that they had always said that rdram royalties were between 1 and 2% and that they expected to get (somewhat?) higher royalties for XDR. Definitely not more than 3.5% as Desai asked, there was considerable humming at that point.
(Elixe you may have to adjust your earnings calcalutions here; I think 2.5% is a good number for XDR)
From Eulau's comments on the financials we can see that the profits (eps) will probably go down again (this was the best qtr profit wise since dec.2000, btw the trailing one-year eps is now 21.8 ct versus 23.8 ct a year ago for the ttm eps- that is -8% despite this one good quarter); litigation expense will be up again
For qtr one the range given for revenue was $ 30-35 M and for op.expenses $ 22-26 M. There will probably be an increase in contract revenue (again), but from the numbers we see that contract revenue is a loss item recently, the cost of contract revenue is higher than the revenue. (this qtr. $ 5.15 vs rev. $ 4.60)
Legal and dates:
- the usual about IFX Richmond
- some more dates about Micron and Hynix
(it might be a good idea to put all those dates in one post and link it above some time)
- the Mannheim case (Landesgericht, is after the EPO patent thing) is scheduled for May 14th. Rambus has requested a later date of June 18th. (not clear if that is still pending)
RaSer/ PCI-Express
about 5 more licensees, now 25+, was 20+ three months ago (and fact has been 20 for some time).
- No income from RaSer in the past quarter
- I read from how things were said also no income from RaSer in qtr1.
(I have to go back and listen again to some pieces there)
- Velio serdes is validated in UMC process, Rambus on PSMC process so they can serve more customers flexibly.
- two of the new licensees are 1) Stargen has something in development and 2) S3 Graphics.
- they are very cagey about RaSer royalty questions. (be prepared for that to be further into the future than you thought)
royalty snips
sdram+DRR+controllers were 46% of royalties ex. Intel, i.e. $
That makes it 0.46 * $ 17765 = $ 8172.
rdram less than 10% of total revenues (< $ 3237), were up 4% vs. last qtr
so:
- Intel 10000
- sdr/DDR 8000
- rdram 3000
- rest 6765 (what is this? XDR too early..... )
- total 27765
Joe TASR up again a few pc in the pre market.
Another wild day coming up?
Cheers
Cor
Hi Guys, did you see the new SiS chipset with PCI-Express?
see:
http://theinquirer.net/?article=13634
The chip, called the 965L, is compatible with a range of chipsets for the PC, including the 661FX, 648FX, and others.
SIS said it will support two PCI Express ports, eight USB 2.0 ports, eight channel sound, 10/100 LAN, two Serial ATA drives and four Parallel ATA drives
Sandy, do you know if SiS is a Rambus licensee for Raser?
Cheers
Cor
Hi Threejack,
I nailed the Intel earnings to the cent. I had 33 cents w/o extraordinary items, but the $ 600 M charge had already been announced by Intel. So I published 33-6= 27 cents.
The extra 6 cents from the tax angle were unforeseeable and will also not repeat.
I consider these earnings blowout earnings, although I expected them. What the market thinks is always a crap shoot:)
Cheers
Cor
For those who do not understand "inquirer speak", this:
And in 2006 Intel will release Tukwila, formerly known as Tanglewood
may mean that new thing is going to be called Tukwila or it may be a fantasy name of Mike Magee at the Inquirer related to Tequila:)
Cheers
Cor
(Itanium is called Titanium there, Intel is called la Intella etc; but he has sources that Mike)
You're definitely a good newshound, Sandy.
Thanks
Cor
Hi Sandy, looked at the Juniper router and could not immediately identify the Rambus part.
You probably know what it is; do they use rdram memory maybe in some of the PICs?
see also
http://www.juniper.net/techpubs/hardware/t640/t640-pic/t640-pic-TOC.html
tia
Cheers
Cor
Elpida launches 2GB DDR2 memory module
Let's hope they get it to work, Cal:)
http://www.dvhardware.net/article2332.html
"Elpida announced the volume production of its 512 Megabit based 2GB DDR2 registered memory modules. These are the highest density currently available for the server market, and have a speed of 4.3 Gigabytes per second (PC2-4300). Elpida developed a new stacked BGA (sBGA) technology, allowing to create slim modules (4.8 millimeter thickness).
2 Gigabyte DDR2 DIMM Features Elpida's new DIMMs (Part numbers: EBE21RD4ABHA-xx) are based on Elpida's 512 Megabit DDR2 SDRAM devices that operate up to 533 Megabits per second (Mbps). The modules are organized as 256M words x 72-bits x 2 ranks, and they contain thirty-six 0.11 micron, 512 Megabit DDR2 devices that are stacked and mounted on the DIMMs using Elpida's unique sBGA stacking technology. The mounted devices have a CAS latency (CL) of 3, 4, 5 with a burst length of 4, 8. The standard 240-pin DIMMs have a JEDEC-compliant outline with a slim thickness of 4.8 mm, and they support Error Checking and Correction (ECC) necessary for high-end server applications. The modules also offer 1.8V operation - one-half the power consumption over DDR. "
On the same site Nanya tells that they find it difficult to get DDR2 validatd to Intel's exacting standards. They expect the price of DDR2 to be some 50% over DDR for some time due to shortages.
(that sounds familiar too:)
Cheers
Cor
What do yuo all think of Taser (TASR) technically?
Fundamentally the eps is sorta exploding.
Cheers
Cor