Description

Rambus Inc. (Ticker: RMBS)

 

Summary

Investing in Rambus stock is a unique opportunity to own a legal monopoly currently trading at ~4.0x CY09 potential earnings, whose stock has the potential to quadruple because no one knows about and/or cares about the Company.  No major sell-side firm covers the stock; no one has laid out the revenue, earnings and cash flow potential of the Company.  Rambus trades at the ultimate discount to its potential future earnings stream and cash flow despite being a dominantly positioned intellectual property company within its space.

 

Background

Rambus is the premier DRAM intellectual property company in the world.  Dynamic Random Access Memory (“DRAM”) is the dominant form of memory storage used in computers, consumer electronics, and other electronic systems.  2007 Revenues for the global DRAM industry were roughly $32 billion. 

 

While most investors view DRAM as a purely commodity product, the industry has historically developed (SDRAM, DDR1) as well as continue to develop (DDR2, DDR3, XDR) DRAM chips based on core Rambus IP.  Currently, approximately 90%+ of the DRAM chips shipped globally are based on Rambus technology, though less than 25% of the market pays Rambus for this IP.  Rambus recently won a major court case which validated the Company’s actions and essentially granted Rambus a legal monopoly.  However, most investors are not paying attention to it because 1) there is no sell-side analyst coverage of the stock explain to investors the importance of this event and 2) there is a perception that the legal battle will continue forever without resolution.  The bull thesis is that the recently won California District Court case significantly improves Rambus’ leverage against the DRAM manufacturers (especially Hynix), which could act as the catalyst for a settlement in the relative near term.  Worst case, we are now in the later innings of the baseball game from a litigation standpoint, so even without a settlement the last appeal should be heard and decided within 1-2 years.

 

Rambus collects royalty/licensing revenues from 2 primary types of companies: 1) DRAM manufacturers who make chips based on core Rambus IP and 2) microprocessor and graphics companies who must license Rambus controller technology to interact with the DRAM chips in a computer.  Rambus collects money for its technology through either a fee-based licensing arrangement (x dollars per quarter for y years), which is geared towards the controller customers, or on a straight royalty basis (x percent of revenues), which is geared towards the DRAM manufacturers.  Rambus also has an emerging consumer electronics business for its XDR technology, as exemplified by its inclusion in the PS3 and next generation televisions.

 

Currently less than 25% of the DRAM industry pays Rambus any money whatsoever.  Most DRAM manufacturers claim that Rambus acted inappropriately in its interactions with the DRAM standard setting body (JEDEC) and therefore that fraudulent behavior nullifies its intellectual property.  So the DRAM manufacturers (Samsung, Hynix, Micron, Nanya, etc.) have kept Rambus tied up in Court for years arguing that Rambus’ patents should be invalidated and the DRAM industry should not have to pay any royalty to Rambus.  Last week a jury decided in the consolidated case of Hynix, Micron and Nanya vs. Rambus which was incredibly favorable to Rambus; not only was the Company found to have acted appropriately with JEDEC, but also was found to have earned an essential legal monopoly due to the quality and strength of its intellectual property.  The decision was bulletproof in favor of Rambus. 

 

Thesis

The crux of the Rambus bull thesis is predicated upon the ultimate resolution where the DRAM industry pays Rambus a royalty rate for its intellectual property.  While the recent court cases focus on trailing edge technology (SDRAM and DDR1), Rambus possesses even stronger patents and an even better legal positioning for the current and future DRAM technologies (DDR2, DDR3, XDR).  Currently, the DRAM market is ~80-85% DDR2, DDR3, and XDR while ~15% is SDRAM and DDR1 and <5% other.  So, Rambus should be collecting royalty from ~95% of the entire DRAM market. 

 

DRAM Market (Gartner estimates)	2008E	2009E	2010E
Global DRAM Market ($mm)	$26,360	$34,270	$37,015
% y/y	-16%	30%	8%
% of DRAM market not paying Rambus	75%	75%	75%
% of market DDR2,3,XDR	86%	89%	92%
% of market SDRAM, DDR1	12%	9%	6%
% of market using Rambus IP	98%	98%	98%
$ revenues subject to royalty by Rambus
DDR2, DDR3, XDR	$17,002	$22,875	$25,540
SDRAM, DDR1	$2,361	$2,296	$1,703
Total $ revenues subject to royalty	$19,363	$25,171	$27,243

 

With the DRAM market and Rambus’ addressable market quantified in the $25-$30bn range for each of 2009 and 2010, the next question becomes, what is the appropriate royalty rate for Rambus to charge?  We have several data points and comparable metrics on this.

 

1)      Several data points emerged from the Hynix case in 2007

a.       Rambus has a licensing agreement with Hitachi for a 1.00% royalty rate on SDRAM and 4.25% for DDR1, which was agreed to in the midst of litigation between Rambus and Hitachi

b.      Other Rambus licenses (names not specified) which were agreed to without the parties being involved in litigation have terms of  0.75% on SDRAM and 3.50% for DDR1

c.       The Hynix award from 2007 of $133.6mm was based on the Hitachi rate of 1.00% on SDRAM and 4.25% on DDR1

2)      The FTC decision from 2006 limited SDRAM and DDR1 rates, but not DDR, DDR3 or XDR rates.  The FTC rates are:

a.       0.25% for SDRAM

b.      0.50% for DDR1

c.       No limit for DDR2, DDR3 or XDR

d.      Importantly, Rambus has appealed this decision and it is likely that Rambus will win on appeal (decision expected by mid-2008) and these low rate limits will be removed.  It is also important to note that with ~90%+ of the market already transitioned to DDR2, DDR3 and XDR this decision is relevant more for back-payments and legal strategy than future revenues and earnings stream, as the FTC has imposed no limit on the future of the DRAM market (DDR2, DDR3, XDR).

3)      Other memory company royalty rates:

a.       SanDisk single-level-cell technology: ~3.0-5.0% range

b.      SanDisk multi-level-cell technology: ~5.0-8.0% range

 

Financial Overview:

Current share price (3/31/08):               $23.30

2007 Revenues:                                    $180 million

2007 Cash Expenses (ex legal): $155 million

2007 EBIT (ex legal)                            $25 million       

2007 Net Income (ex legal)                  $.18

Cash/Share (12/31/07)             $4.50

Fully Diluted Shares (12/31/07)            104 million

 

Rambus Incremental Revenue Sensitivity Analysis		2009E	2010E
Assuming 0.75% for SDRAM and DDR1 and
2.0% DDR2, DDR3 and XDR		$458	$511
3.0% DDR2, DDR3 and XDR		$686	$766
4.0% DDR2, DDR3 and XDR		$915	$1,022
Rambus EPS Sensitivity Analysis
Assuming 0.75% for SDRAM and DDR1 and
2.0% DDR2, DDR3 and XDR		$2.83	$3.14
3.0% DDR2, DDR3 and XDR		$4.15	$4.61
4.0% DDR2, DDR3 and XDR		$5.47	$6.09

 

It should be noted that this analysis does not include back payments the manufacturers currently owe (or could owe) Rambus.  For example, as a result of the 2006 verdict in favor of Rambus, Hynix currently owes Rambus north of $200 million in back-payments.  

 

Timing

Rambus has been involved in this legal battle for years.  A natural question an investor would ask is, why now?  The bear thesis is that this legal battle will continue for another 5 years.  There are 3 main reasons why now:

 

1)      A major DRAM manufacturer has its back against the wall vs. Rambus:  For the first time, Rambus has put together consecutive victories against a major DRAM manufacturer – Hynix.  In early 2006 Rambus was awarded $134mm in back-payments; now Hynix was a participant in this consolidated DRAM case that Rambus won last week.  As the case last year indicated, an agreement with Rambus outside of the Court can be accomplished at a lower rate than through the legal channel (ex. 0.75% and 3.50% for SDRAM and DDR1 licenses versus Hitachi (and Hynix settlement) at 1.00% and 4.25%).  There is incentive for Hynix to settle without appealing to the Federal Court if Hynix determines the odds of winning on appeal are lower than they previously thought and their probability-weighted option of appealing is more expensive than the option of settling with Rambus.  Here is Hynix’s financial motivation for settling; Hynix is estimated to sell ~$6bn worth of DRAM in 2009 and $6.6bn in 2010; a 1.00% difference in royalty rate is worth ~$60-$65mm to Hynix in each of 2009 and 2010.  The combined effect to Hynix of a 3.50% versus 4.25% royalty rate (remember this is the difference between the Hitachi litigated rate and Rambus’ other non-litigated customer rates for DDR1) on DDR2, DDR3 and XDR in the next 2 years alone is almost $100mm. 

 

Once the first major DRAM manufacturer settles with Rambus, the domino effect begins, as the consolidated DRAM industry loses one if its teammates in its fight against Rambus, and the terms Rambus offers to settle with the next DRAM manufacturer get worse.  The first settlement is likely to lead to another, as no DRAM manufacturer wants to be the last man standing fighting Rambus alone and getting the worst terms on royalty rates from Rambus.

 

2)      DRAM industry almost out of appeals:  After this victory for Rambus in District Court against Hynix, Micron and Nanya, there is only one realistic appeal left for the DRAM industry – to Federal Court (this is assuming the Supreme Court will not hear the case upon appeal from Federal Court).  An appeal will probably take ~1.5 years from now to reach a conclusion, after which there is no where to appeal outside of the Supreme Court.  Unless you take the view that the Supreme Court will hear this case, the bear thesis that “litigation will last another 5 years” is really only ~1.5 years.  So, even with the “fight until death” mentality, we are still looking at a less than 2 year time horizon.

 

3)      Anti-trust case pending:  There is strong evidence that exists which indicate several DRAM manufacturers colluded to keep Rambus-based DRAM (in this case RDRAM) from becoming the industry standard.  This evidence includes emails believed to be in Rambus’ possession, testimony from former DRAM executives whom are currently in jail from the State anti-trust case, etc.  Due to this evidence, it is strongly believed that the DRAM manufacturers will settle with RMBS before this anti-trust case gets into court and the evidence distributed publicly.  While the Judge has delayed putting a date in the calendar up until now, the conclusion of the consolidated case in District Court last week (as well as a procedural decision by the Judge) indicate that we could get a date in the calendar for the end of 2008 or early 2009.  Once this date is set, the clock starts ticking for the DRAM manufacturers to settle. 

 

Catalyst

1) Potential FTC reversal on Rambus appeal (within the next 3 months)
2) Potential settlement with Hynix following the District Court case at the end of March
3) If no settlement with Hynix, then a potential injunction on Hynix chips following the District Court case at the end of March
4) Potential anti-trust trial date