Ok, Mr. Hotshot Coolposting Insufferably Smug Dude,
let's see what you (or somebody who actually knows something) can make of this...
rambusite.com
"My Thoughts On Valuing Rambus
More On The Great Memory War Of 2000
By Patrick T. Newell
June 28, 2000
The DRAM sector is a quite cyclical. Strong ups and downs modulate powerful underlying growth and contraction.
DataQuest, which has the best track record for forecasting in the industry, predicted in late 1999 that the DRAM market would be $29 billion in 2000 and $70 billion by 2002. Rambus Inc. RDRAM was predicted to have a 10% share in 2000 and a 50% share by 2002.
Actual DRAM sales are exceeding predictions. The most recent projections are for a $36 billion market in 2000 . That would translate to $3.6 billion RDRAM using the DataQuest 10% projection. The world’s largest memory maker and biggest producer of Rambus Inc. products, Samsung, now predicts a $38 billion market for Rambus Inc.’s RDRAM by 2002.
DRAM market $36 billion in 2000 (6/15/00)
ebnews.com
Samsung estimates Rambus Inc. production, June 22, 2000
koreaherald.co.kr
The recent settlements by Toshiba and Hitachi suggest that Rambus Inc. will be able to charge royalty on all synchronous DRAM, whether RDRAM (Rambus DRAM) or not.
Valuing Rambus Inc.:
By 2001, Rambus Inc. will trade on 2002 earnings. According to indications in the press, the royalty Rambus Inc. collects is about 1.5% on RDRAM, 1.25% on SDRAM and higher on DDR. In addition, Rambus Inc. collects higher royalties on the DRAM controllers (3-5%, although these are cheaper). As a result, Rambus Inc. earns about 3% on DRAM sales, half from controllers. The latter half is often conveniently ignored by Rambus Inc. skeptics. Since the DRAM market is currently running 20% ahead of projections, I will up the 1999 projection of $70 billion in 2002 by 20%.
The Calculations
3% of $84 billion times 60% (40% tax rate) = $1.51 billion. Divide by 100 million shares and estimated 2002 earnings are $15.10/share after taxes. Contract fees from licenses adds perhaps $0.50/share. Total: $15.60/share in 2002.
If the legal victories hold up, Rambus Inc. will have a monopoly more widespread and more complete than Microsoft or Intel. What should a company earning $15.60/share after taxes in 2002 be worth in 2001? With high margins, high barriers to competition, and rapid growth, perhaps 40 times forward earnings, or $624 share.
Key variables:
(1) Growth rate of the DRAM market is notoriously hard to estimate.
(2) Non-RDRAM products like serializer/deserializer sold by PMC-Sierra are not counted in the above. These are currently small but fast growing markets.
(3) Recent legal victories might take months or even years to encompass entire industry. It is possible that established precedents could be overturned.
(4) The future market share of RDRAM is still controversial. Note that Samsung, the world’s most successful memory manufacturer, and DataQuest have more credibility to me than any other analyst.
A useful price estimate for Rambus Inc. requires estimating three factors:
(1) The probability that RMBS will take a large and growing percentage of the DRAM market – or else win the legal rights to the existing market (or both).
(2) The size of the DRAM market
(3) How fast the DRAM market is growing.
Simply because Rambus Inc. earnings grow 100 fold or more in a year does not make Rambus Inc. stock worth a forward P/E of 100 or 200 if that implies a price target higher than if Rambus Inc. owned the whole market. Suppose that the DRAM market was fixed at 1999’s $20 billion. Then the most RMBS could get from royalties would be 1.5% x $20 billion /100 million shares x 2 (for controllers) times 60% (taxes) = $3.60/share. Since earnings would be stagnant, forward P/E should be 10, so a price of about 10 x $3.60 or $36 would be appropriate (well below present value). Even if Rambus Inc. stock growth went from 0.5% to 50% (100 fold growth), no price higher than about $36 would be reasonable. The market always looks forward (or eventually does anyway) to see the future stagnation.
But in fact the DRAM market is growing fast; indeed faster than the most optimistic estimates. When a 1999 DataQuest estimate put the current year 2000 DRAM market at $29 billion and the 2002 market at $70 billion, many PC-centric observers couldn’t believe it. The PC industry is by far the largest user of DRAM, and it is growing only about 10-20% a year. DataQuest must be wrong many speculated.
Yet as of June 2000, it is clear DataQuest actually underestimated the strength of the growth So did the DRAM companies. Year 2000 DRAM Sales now looks to be $36 billion. In June Samsung estimated that the 2002 market will be about $76 billion, half RDRAM.
This dramatic upswing has a powerful effect on the worth of Rambus Inc.. If the DRAM industry were growing by the 15% growth rate of the PC industry, then a fair valuation of Rambus Inc. would use a forward P/E of about 15-30. If the market is growing at better than a 25% pace as appears to be the case, then a forward P/E of 25-50 is quite reasonable.
NOTE: Since Rambus Inc. collects royalties without doing the manufacturing, all these estimates assume Rambus Inc. employees do nothing except sit around making bonfires out of the cash that pours in. Others have commented on this. Rambus Inc. may find a better plan.
The key point is too understand why the DRAM market is growing so fast, and whether it is likely to continue.
The PC market has the biggest share of DRAM sales, but is probably the slowest growing segment. Here are some others.
(1) DSPs/small process control processors. Originally DSPs (Digital Signal Processors) were used for audio and video signals, and to perform certain mathematical algorithms fast. They still are, but there is an increasing trend for DSPs to show up everywhere. Think of them as a pint-sized CPU or chipset. An example of how they are spreading is Analog Devices recently convincing Whirlpool to use a DSP in blenders to save energy controlling the motors. The market is already large, and is growing by better than 30% a year. InStat and others have estimated that in the next 5-10 years this will be a bigger business for memory than the PC market. Tiny processors might appear in a hundred places – or a thousand – in your home controlling almost everything electronic. Sun Microsystems is hoping to break into this market with its new MAJC which uses Rambus Inc. technology. Note, however, that Sun Microsystems is not currently a major player in this market. TI, Motorola, Analog Devices and DSP (the company) are the major players. All currently use plain SDRAM. This is the biggest potential long term gold mine out there. If it continues to grow as predicted, and if Rambus Inc. wins royalties either through legal means or by the adoption of RDRAM, the win will be big.
(2) Internet consoles/game players/DVD. Rambus already is in PlayStation 2, predicted to dominate the market. Enough said.
(3) Fiber optic storage area networks. Brocade dominates this market, currently a few billion, but the market is expected to grow to tens of billions of dollars in the next few years (according to Briefing.com). Brocade’s Silkworm uses RDRAM.
(4) Network connectors. Specifically, the next generation of OC-192 may have to use Rambus, although IBM is hoping to make 200 MHz DDR work. This is a large, fast growing, but currently fragmented market. Rambus Inc. already has a small stake through companies like Vitesse, SwitchCore and PixelFusion. Other companies are still using made to order SRAM (static RAM) parts which cost 10 times what RDRAM does. Rambus Inc. is also hoping to make an impact through its new QRSL (Quad Rambus Signal Logic).
Short term, the biggest driver of Rambus Inc. profits will be the Willamette and/or settlements with SDRAM producers. But PC sales alone could never make Rambus Inc. the next Cisco or Microsoft. For Rambus Inc. to be the stock of the decade, instead of just a stock that was undervalued at the start of 2000, DRAM usage will have to continue to spread far beyond the PC. Fortunately for the company and its shareholders, that appears to be happening at a pace faster than the most optimistic estimates."
tekboy/Ares@tosleep,perchancetodream.com |
