Many investors remember 1999 when Qualcomm (QCOM) had a run from $3 (split adjusted) on Jan 1, 1999 to $83 (split adjusted) at the end of the year. QCOM was a 27 bagger in a single year. The QCOM run changed the lives of ordinary people. It made millionaires out of schoolteachers and preachers who invested and held throughout 1999. In short, it was like winning the lottery. Of course, those were the bubble years and something like that could never possibly happen again…..or could it?
On June 8th, when Rambus was at 14, I shared an article with Seeking Alpha readers telling you that RMBS was about to make a run. Less than a week later the stock is now in the mid 17s and headed higher. As recently as February 20, 2009, just 4 short months ago, Rambus stood at $6/share. Is it even remotely possible that Rambus could be a 27 bagger in a year like Qualcomm was in 1999? That would mean Rambus has to get to $162/share by February 19th 2010. The surprising answer to this question is yes. While it may take Rambus two years to get to that level, it certainly is possible that Rambus could be there in 8 months.
How can that be?
First, you don’t have to believe an anonymous internet poster (NukeJohn), even though I have given you recent predictions on TSRA and TIVO that have come to pass. In 1999, there was a fund manager (Kevin Landis of Firsthand Capital Management) that picked QCOM as the tech stock of the year and guess who he picked as the tech stock of the year in 2009? You guessed it: RMBS. Watch the last minute of this 4 minute video.
How can Rambus possibly have the potential to be a $162 stock?
First, I should tell you that the late Fred Hager of Hager Technology, one of the top technology advisory services in the bubble years, predicted that one day Rambus would be a $1000 stock. Although Fred took a tremendous amount of abuse for that ill fated call (who knew the U. S. justice system would take 10 years?), Mr. Hager's heirs might one day still have the last laugh. So how is it possible for Rambus to get to $162 in a year or two….or to a $1000 in 3-4 years? Please read on….
The History of Rambus
In the 1990s, two PhDs by the name of Mike Farmwald and Mark Horowitz had dinner one night in Palo Alto, CA where they came up with idea that solved the biggest problem facing the computer industry at that time. Computers were running faster than the memory connections, and even the engineers at Intel (INTC) and IBM (IBM) couldn’t solve this memory bottleneck problem. But two young enterprising American inventors came up with the solution and formed a company called Rambus.
The story is too long to tell you from the start, but this link can give you most of the information. Unfortunately, after these brilliant engineers founded Rambus, the memory manufacturers, along with JEDEC (the industry standard organization) simply started “cherry picking” Rambus’ intellectual property and incorporating it into the standards. You will hear many people tell you that Rambus tricked the industry into standardizing on their technology, but this was simply not the case.
In fact, Judge Ronald Whyte of the Northern District of CA held an extensive trial on this very issue. Rambus’ conduct while a member of JEDEC was examined with a fine tooth comb, and Rambus was completely exonerated. After a seven week trial, the jury came back in after about three hours with a unanimous verdict.
The FTC was another major roadblock Rambus had to clear. Even after the FTC’s Chief ALJ completely exonerated Rambus in his Initial Decision, the politically motivated FTC full Commision overturned their own ALJ and convicted Rambus of “exclusionary conduct” and attempted to get them to give up their patent rights. But the Court of Appeals of the Federal Circuit slapped down the FTC, and then the U.S. Supreme Court refused to hear an appeal. The FTC dropped the case recently, which has helped clear the way for Rambus to start collecting royalties on the DRAM Market and the controllers associated with every DRAM device. Infringement has been proven in the Northern District of CA court, where Rambus was forced to win all three phases of a trifurcated case.
This case started in 2000 when Hynix filed charges against Rambus as part of a coordinated scorched earth litigation policy to put little Rambus out of business. Rambus won all three phases, but the final verdict did not come out until March of 2009. If they really want to appeal the loss, Hynix now has to pay the 250 million dollar supersedes bond and sign a lien of a 300 million dollar fab by the end of this month. Of course, one alternative to this financial pain is to just settle with Rambus on these new reasonable rates Rambus has offered the industry.
In addition, Rambus has an antitrust case against Samsung, Hynix, and Micron (MU) in California State Court in San Francisco. This case started in May of 2004, and it is finally coming to trial in September. Over the past five years, the defendants have tried numerous ways to get this case dismissed, or to delay the start of this case.
But the Honorable Judge Richard Kramer just two weeks ago confirmed his tentative ruling that this case will start in September. This news has not been reported by the mainstream business press, but it was confirmed in Rambus’ recent UBS conference presentation.
Significance of the AT Case
Unless you have followed Rambus closely for 10 years like I have, you may not understand the significance of this Antitrust price fixing case they have against Samsung, Hynix and Micron.
First, you should know that Samsung and Hynix signed plea agreements admitting that they fixed prices, and they both had several executives go to jail. Micron turned the other cartel members in to the US Department of Justice in exchange for amnesty. This story has been covered by major news media in the past. For example, Business Week first started putting together the pieces of the puzzle over three years ago in this excellent article. Then, just a year ago, Bloomberg did an expose on the DRAM Price Fixing Cartel.
In essence, these defendants have already been proven guilty of DRAM price fixing, and now Rambus can tell the world (and a California jury) why they fixed prices. In fact, a few emails have already been made public in the early stages of this case. The damages in this Antitrust price fixing case are huge. Rambus lost billions of dollars in royalties when the DRAM manufacturer’s actions convinced Intel to shy away from RDRAM as the standard memory for future Intel PCs.
In fact, in recently revealed court documents, Rambus alleges the damages to be 4.32 billion dollars, which are then subject to automatic trebling because this is a Cartwright Act case.
Therefore, the total damages facing the DRAM manufacturers approach 13 billion dollars. With Rambus having approximately 104 million shares outstanding, that is over $120/share in past damages alone from DRAM manufacturers (this doesn't include controller manufacturers).
Until last week it appeared that the stakes were so high that the memory manufacturers couldn’t afford to settle, and it was possible the cases could go on for years with appeal after appeal.
So what changed last week?
Just this past Thursday in after hours, it was reported that Rambus and the European Union have reached an agreement whereby the EU would drop charges against Rambus (following the lead of the FTC).
In exchange, Rambus would license the DRAM manufacturers and controller manufacturers with very reasonable rates. What is reasonable? Back in 2000, Rambus signed a license for DDR with Samsung for 3.5% and with Hitachi for 4.25%. Most Rambus long term shareholders (such as myself) were expecting at least 4.25% royalties on DDR2 and DDR3, and possibly 5-6% or more.
But Rambus has decided it makes business sense to bury the hatchet and move on, so they have offered the EU a very reasonable rate and this offer will be available to any and all takers. However, if a company doesn’t accept this rate (which includes a "most favored nations" clause), they will face much higher rates in litigations with Rambus.
Although Rambus has negotiated these rates with the EU, they have essentially offered these rates as “most favored” worldwide rates to any company who is ready to sign.
The terms and rates Rambus has negotiated with the EU were printed in an official EU publication and are as follows:
11. Rambus's 13 May proposal offers a bundled five-year worldwide licence for future DRAM products for all of its patents for SDRAM, DDR, DDR 2 and DDR 3, whereby it commits not to charge for SDRAM and DDR. The offer does not cover past royalties.
12. For chips, the following would be the maximum royalty rates:
(a) SDR DRAMs — subject to compliance by the licensee with the terms of the license, the licensee will be granted a royalty holiday during the term of the license on SDR DRAM devices;
(b) DDR DRAMs — subject to compliance by the licensee with the terms of the license, the licensee will be granted a royalty holiday during the term of the license on DDR DRAM devices;
(c) DDR 2, GDDR3 and GDDR4: 1,5 %;
(d) DDR 3: 1,5 %.
13. For Memory Controllers, the following would be the maximum royalty rates:
(a) SDR Memory Controllers: 1,5 % per unit of selling price until April 2010, then dropping to 1,0 % ( 1 );
(b) DDR, DDR2, DDR3, GDDR3 and GDDR4 Memory Controllers: 2,65 % per unit of selling price until April 2010, then dropping to 2,0 %.
14. The commitment will be valid for a period of five years from the adoption date of the Article 9 Decision. The licence grant will expire at the end of this five-year period, irrespective of the signing date of the licensing agreement. Licensees will have, after a one year minimum licence period, an unconditional opt-out to the licenses before expiry of the duration of the contract.
This weekend, the New York Times had an article that stated the following:
Any company that wants to make DRAM, or dynamic random access memory, has to pay Rambus for the design it developed. The chips are used in personal computers, servers, printers, personal digital assistants and cameras. Worldwide DRAM sales were $34 billion last year.
Rambus said Friday that it would cap the fees it charged for licenses over five years for certain memory types and memory controllers. The European Union said Rambus also promised that any future fee cuts would benefit the entire market.
The above statement from the New York Times article is correct, but they left off a couple of items. First, they said "personal computers, servers, printers, personal digital assistants and cameras". They left off some major products that use DRAM (or graphic DRAM) chips including cell phones, HDTV televisions, and camcorders. These are high volume products that make the size of the controller markets huge. The size of the controller market dwarfs the 25-35 billion dollar DRAM market (which Rambus will get 1.5% on).
Here is an exchange from Rambus’ Q1 2008 conference call transcript (pages 8-9) which is available from Seeking Alpha. It appears that the Total Available Market (TAM) for controllers is in the "hundreds of billions" as was stated in this Q1 2008 conference call.
Sharon Holt - Senior Vice President of World Sales, Licensing, and Marketing
Yes. Hi, Mike, this is Sharon. So… yes, and I realize when we say DRAM controllers that that could be confusing. Let me give you a couple of examples before I jump to the general. So a system that most for us are familiar with is a typical PC. So in a PC, obviously you have a CPU usually from Intel or AMD that’s either connected through a bridge or directly to memory. Whether it's the CPU or the bridge that’s connecting directly to the memory, that device would include what we would call the DRAM controller. So it's basically some logic that helps transmit messages back and forth between the memory and the CPU. So in a PC system, that's what the architecture looks like. In other types of systems, it could be an ASIC or some other kind of chip set that is talking to the memory and in that case, that ASIC or chip set would include a memory controller. Does that help?
Mike Crawford - Riley Investment Management
Is there any way to kind of put a TAM on that?
Sharon Holt - Senior Vice President of World Sales, Licensing, and Marketing
So it's… first of all, when we look at it, we are looking at certain types of memory devices at a very high level. If you are looking at the market itself, it’s quite large. As you know, the DRAM market is $30 billion a year. The controller market is multiple times that much larger if you just add up all the Intel microprocessors, all the AMD microprocessors, other chips sets and ASICs and other specialty devices that talk to memory, we're talking hundreds of billion of dollars, that's at the top level and then of course for us, the TAM would be some percentage of that depending on the royalty.
Now, let's review what the Wall Street Journal said when they announced the EU Rambus settlement...
Rambus will offer lower royalty rates to customers of its chip technology, but it will ask for more customers to pay royalties, this person said.
Rambus will ultimately get the 2.65% (2% after 4/2010) for chips with embedded controllers that interface to DRAM chips. If the companies don’t pay willingly, they will ultimately pay a lot more unwillingly. This is actually a good deal for the controller manufacturers because Samsung had signed a license in 2000 that had a 5.5% rate on controllers. If we are conservative and say the controller market is 200 billion (instead of hundreds of billions) and we assume that only 30% of the market signs up in the first year, that is still 2.65% on 60 billion dollars or 1.59 billion dollars in revenues from controllers and 375 million in revenues from DRAM. Remember, this does not include past damages for DRAM or controllers. Past damages could easily exceed 1.5 billion on DRAM and 3 billion on controllers when one considers the amount of products sold in the past decade.
Hynix and Samsung are now virtually forced to agree to these rates, because if they turn them down, they are on the hook for 12 billion dollars in damages in the AT case. If the parties are willing to pay for back damages, Rambus will probably let them out of the AT case.
The other thing I was thrilled to see is that DDR4 is not included in this EU settlement. Just recently, Rambus put JEDEC and the whole industry on notice that they had five new technologies available for licensing. Rambus knew that, once again, the DRAM manufacturers were considering using Rambus IP for the next standard currently known as DDR4. Rambus expects to be paid on these technologies if they are included in the standard. I suspect that Rambus has already sent JEDEC a letter offering Reasonable and Non Discriminatory (RAND) rates. I can’t imagine the rates Rambus will demand for their newer technology will be below 4.25%, and I suspect the rates are closer to 5%.
Summary
Rambus has lowered their royalty demands, but when they start getting paid on all the products that have DRAM, even these lower royalties will justify a price well into triple digits. Rambus may never be the $1000 stock that Fred Hager talked about, but it could easily be $300-$500 in 3 years.
In my opinion, we will see settlements soon, likely starting with Elpida. Rambus has drawn the line in the sand. They have said “here are our rates”, take it or leave it. Once one or two companies start taking it, the flood gates will open up, and most companies will have to take it. Those that don't sign licenses will wind up paying a helluva lot more.
As much as I hate to admit it, litigation has become a core competency at Rambus. As recently as a week ago, Harold Hughes (Rambus CEO) said he could pass the bar exam, and I can sympathize with him. As a Rambus investor for ten years, I've learned more about law than I care to know, but it has been fun learning how to handicap patent cases. This has served me well in other investments like TSRA and TIVO.
Although I was initially disappointed with the rates, I now see the brilliance of this strategy being executed by Rambus. The part I like best is that the 5 year licenses the DRAM manufacturers will be forced to take stretch out four years after the F/H patents expire. Long before then, the DRAM market will transition to DDR4/XDR2, where I believe Rambus will be getting paid 4 to 5% royalties.
Rambus should soon be at $35-40/share. I would not be surprised to see a run similar to the run in January 2006 when the stock went from 16 to 36 in 11 trading days. I am raising my 3 month price target to $45 and my 1 year price target to $120.
While we may not quite repeat QCOM's 1999 performance, Rambus will still be one helluva investment over the next year.
Disclosure: Long RMBS, TSRA, TIVO; Short MU
