I have compiled a lot of data on RAMBUS. I started publishing this just before I took out my latest position (SHORT at $62 a couple of days ago). You might find some of this helpful.
(You may wish to cut this into WORD to read it more easily. This Fact Sheet is just over a page in length. Just in case there is a new PC user out there, Highlight the text, click your right mouse key, and hit COPY, then go to WORD, click the right key and hit PASTE).
Facts on RAMBUS 2: RAMBUS have serious competition
I keep reading comments on this message board that RAMBUS have a monopoly. Nothing could be further from the truth. Today is a dump of information on the real position.
First, RAMBUS's only technology is a high speed interface to a DRAM, called an RDRAM. RDRAM was developed for games applications, such as for a Nintendo game, where there is only one or two DRAMs instead of the 8 or 16 in a PC. For the past couple of years RAMBUS have been trying to get into the PC market. There are several factors which taken together mean this will not happen to any large degree:
1. The cost of a RAMBUS solution is around 25% of the DRAM price. This is made up of:
a. A RAMBUS licence fee of 2% of revenue from the DRAM manufacturer
b. An increase in chip area of around 10%
c. There is no provision for speed-binning, hence another loss in yield, of at least 10%.
2. In a PC the two biggest pieces of silicon are the processor and the memory. Intel make the processor (and the system controller, and motherboard), for around 80% of the world's PCs.
The profit in a PC is really split between these two pieces of silicon, the processor and memory: If you take away the overheads of a monitor and the minor bits and pieces (things that are cheap: mouse, keyboard, case, PSU, video card, sound card), then it really is simply a profit split between Intel and the memory makers. This is cyclic: at the moment Intel have the lion's share of the profit, and in 1995 the memory makers had the share. In 2000 the memory makers look as if they will be ahead again, hence Intel's interest in LG.
So who pays for the 25% mentioned above: the memory makers and Intel. Neither are going to volunteer to give 25% of their profit away just so that RAMBUS can earn 2%.
3. In line with 2 above, almost the whole industry has got together and commissioned MOSAID to develop a competitor system to RAMBUS. It is called SLDRAM (SyncLink DRAM). You can find information on it on the MOSAID site: www.mosaid.com. Parts are being fabricated now and will be sampled in August.
MOSAID have designed every generation of DRAM for the past 20 years. They are the world's largest DRAM design house and lead the others by miles. Every single DRAM company in the world are MOSAID customers in one way or another. MOSAID are extremely and overwhelmingly competent. MOSAID have invested very heavily in new design facilities in Canada this year. They understand all aspects of the DRAM process: design, production and test. MOSAID also are the world leading vendors of memory engineering and evaluation systems and software.
In contrast with the competence of MOSAID, after meeting all the main players in RAMBUS, I was astounded that not one seemed to know anything whatsoever about a DRAM. I was asked to explain how a DRAM worked, and had to show them how each of the current DRAMs operate. This was astounding: after all, this is public information that every engineer should be able to recite in his sleep. If RAMBUS could not design even the simplest circuit using a standard DRAM, how could they develop the next generation? This ignorance is reflected in the play made by RAMBUS in their prospectus about the importance of granularity in the PC market: granularity (ie. How few DRAMs make a module) is important to games but does not matter one jot to a PC. Also the real speeds in a PC environment are much lower than the headline speeds RAMBUS report due to the large overhead RDRAM carry - there is lots of control for graphics applications, but PCs just want to dump cache data as separate 8 byte blocks, so do not need any of the bit manipulation and masking overhead in RDRAM.
I was so incredulous of the ignorance at RAMBUS of DRAM in general, I asked the CTOs and CEOs of other memory companies who had both high and low level meetings with RAMBUSengineers about what they found, and got exactly the same feedback. RAMBUS is an amazing victory of marketing over real engineering. Unfortunately just as a Boeing stays in the sky because of real engineering, PCs also keep operating on the same principle. Regardless of how many MBAs and lawyers there are, the world goes around because of real engineering. Put MOSAID against RAMBUS like they are now and RAMBUS have not a hope.
The marketing noise is being made by RAMBUS now because they have devices sampled. In August the SyncLink consortium will issue samples, and in January SyncLink parts will be in volume production. The SLDRAMs are faster than RAMBUS DRAMS (including DDR), cost only 6% in area (our estimate), have no significant licence fee, and support fully speed-binning, and are easier to use.
A memory expert.
I attach my last Fact sheet here. I intend dumping one bit of my homework on RAMBUS each day for the next week. By the way I have no axe to grind against Rambus as a company: I am responding simply because I witness a flow of cash from private investors who have not done their homework, to the institutions who have analysts producing unpublished reports based on hard work researching the real facts.
I am a believer in the mass of private investors being able to beat the institutions, but with the foot thumping statements about Rambus on these message boards it looks like I have over-estimated people's willingness to do their homework and under-estimated the mass hysteria. Well, the harder your blow up a balloon, the bigger the pop. Witness Asia.
Facts on RAMBUS 1: The real short position. 24 June 1998
Bloombergs report the short interest in RMBS to be 3 million shares, and the public float to be 3.5mn shares. The Bloomberg data is reported by numerous web services. If this data was correct, there would be a big share price increase caused by those with shorts trying to close their positions on any price rise. This mis-information has caused the recent hike in the Rambus price.
Unfortunately for those buying RMBS, both Bloomberg figures are wrong. There is only 30 minutes worth of shorts, these are held by hedge funds who bought at the August peak, and the public float of shares is much bigger. This means the shares are likely to dive shortly, if you excuse the pun.
Fact: The shorts in RMBS are almost all traded, eventually, by either Goldman Sachs or Merrill Lynch. If you speak to your broker at either of these fine firms, they can check the real position for you. They will report it is around 800,000 shares (i.e. about 30 minutes worth at today's trading volume). They will report that Bloomberg's Data simply is an error.
Fact: there are more shares in the public float than reported by Bloombergs. Yesterday around 10 million RMBS shares were traded. That is pretty incredible stuff if the public float according to Bloomberg is correct: do you really believe every free-public share traded 20 times yesterday? Or every single public share was traded 3 times.
The real public float seems to be in excess of 5 million shares.
The extra shares come from sustained insider selling: I count around $64 million dollars of shares were sold by insiders since last August - the last figures available through the SEC is for 10 June (It can take a month for a promptly reported insider trade to get published). Visit insiderwatch@cda.com, or Yahoo Insider Trades, for the picture until the end of May. Use a broker's terminal for the June figures (10 June is my latest).
So, the insiders are selling like crazy: they sold roughly twice Rambus's total sales for the past year. This has upped the float, and there are institutions also selling during this time. They all figure the average of $50 is more than enough.
As regards people scrambling to cover shorts, they do not need to. The real short position has not changed much: there were no large positions shut down except my own (Yes, I was short - see below). The other positions are held by hedging funds that seem to have taken them out last August, when the price was around $80. With the total short interest being less than an hour's trading volume, there is not going to be a price hike when they close their position, if they close.
The value of $13 for this stock, that I find, and is also reported by Vectorvest (www.vectorvest.com). The Motley Fool value for those that follow this, is, I calculate, $9.50 (yes, nine dollars, fifty cents, NOT nine hundred and fifty dollars!)..
A memory expert.
|
