Readers livid, Rambus vulnerable
The issue seemed straightforward: Why didn't Intel go the distance to reduce the cost of computing for its customers--and maybe benefit stockholders, at the same time? Surely Intel wasn't thwarted by technical barriers. Last August, Intel nemesis VIA showed the world how it could be done with a P4 chipset that allowed using lower cost memory. Intel, which won't have similar functionality ready until next year, slapped that company with a patent infringement lawsuit. If you can't outsmart them, then out-litigate them.
Understandably, you were livid. Needing reasonably priced high-performance workstations, TechUpdate reader and streaming media engineer Ryan Osswald said he was "foaming at the mouth." For Harold Bush, chief technologist at a systems provider, Intel's decisions were hitting his company right on its bottom line. "As a provider of computers built around Intel CPU's and Intel manufactured motherboards," says Bush, "I will be forced into selling the 845 mainboards with slow SDRAM memory in order to be price competitive." Network systems engineer George Ou wrote "stupid Intel management decisions are killing P4 adoption." Indicating that he'll vote with his dollars, Jeff Tuttle said "[Intel] doesn't own the computing industry and they're kidding themselves if they think we have to accommodate them. We are the customers."
At least one reader suspected that my column might have only scratched the surface and suggested that I dig deeper. So deeper I dug.
The Rambus conspiracy theory
In my 10-plus years as a tech journalist, I've come to know Intel as an organization that makes extremely well-timed business decisions. Between its processor road maps and systematic price drops, the company has charted a profitable path for itself and its customers. But five years ago, it made a business decision that would come back to haunt the company.
In a 1997 S1 filing with the SEC, Rambus disclosed the details of a contract that it inked with Intel. The contract basically required Intel to pair Rambus memory technology with the Pentium 4. The contract had several other noteworthy provisions.
First, the contract granted Intel one million warrants in Rambus at $10 per share and gave Intel ten years to exercise those warrants. Second, it specifies that Intel's implementation of Rambus memory will be Intel's main effort in terms of developing an interface for accessing main memory from 1999-2002. Third, that Rambus could seek remedies if Intel "communicates to any of the then current top ten (10) DRAM manufacturers that Intel has plans to support, as the primary DRAM for PC main memory applications for the years 2000, 2001 and 2002, any New Interface other than the Rambus-D Interface Technology"
Should shares of Rambus soar, Intel stood to gain enormous profits by exercising its warrants for a million shares at bargain-basement prices. Intel clearly had a financial stake in promoting the more expensive Rambus memory.
But by 2001, things started to go awry. Pentium 4 demand was way off and many analysts cited the expensive Rambus memory as a contributing factor. Worse, Rambus' share price, which traded in the $80 range as recently as last fall, started plummeting in the spring. The company lost over half its value in March alone. The value of Intel's Rambus warrants was evaporating quickly. Meanwhile, VIA was coming to market with a performance competitive solution based on the much cheaper DDR memory.
Rambus, in the meantime, was suing just about every manufacturer of DDR for patent infringement. While some memory vendors buckled, others did not and the courts started siding with the companies that dug in their heels like Micron, Hynix, and Infineon. Things were not looking good for Rambus' patent. Today, the cases remain unsettled. Rambus CFO Bob Ealau told me that litigation is continuing in both the U.S. and Germany and speculated that things are "looking better in the Germany than in the U.S., but that we have a strong case in the U.S."
Bound by its contract with Rambus, Intel appeared stuck in a catch-22. Switching gears to DDR would lower the cost of Pentium 4 systems and might increase demand. But switching gears meant one of three things had to happen. First, depending on your interpretation of the agreement, Intel might be in breach of its contract with Rambus by announcing support for a different memory type. Second, it could redefine its contract with Rambus to allow it to introduce support for non-Rambus memory before 2003. Or, third, it could exercise a poison pill that was included in the contract--that, with written notice, Intel could terminate the contract at any time.
Two weeks ago, on the heels of VIA's introduction of the cheaper DDR-based solution, Intel announced it would introduce its own DDR solution in the first quarter of 2002.
So what happened? Was there a breach? Could the contract have been redefined? Or did Intel take the poison pill? In a conversation last week, Intel spokesperson Chuck Mulloy declined to elaborate, only hinting that a confidential renegotiation might have taken place by saying that the Rambus S1 "is a dated document and any number of things could have changed since then."
Damage control
While the conspiracy theory remains unconfirmed, the turn of events certainly calls Rambus' future into question. And of course there's the question of why is Intel waiting until next spring to release support for DDR. Intel's Mulloy says "we haven't changed our position. We still think the Rambus technology is the best technological solution. If you want to do high performance, we think you should use Rambus."
But, judging by DDR's reported ability to stand up to Rambus' performance, something doesn't add up. If I had to go out on limb here, I'd say that Intel can't bear to see the Rambus ship sink. Rambus is highly dependent on Pentium 4 shipments and if Intel did a sudden about face and started shipping P4 systems with DDR memory--or allowed them to be shipped--Rambus would be finished. Although Eulau was quick to identify the royalties it receives from nine semiconductor companies as a solid revenue stream, Rambus spokesperson Christine Wiseman admitted that the company derives a majority of its revenue from Rambus DRAM.
The value of Intel's warrants and Rambus' survival appeared to be tied to effective DDR damage control. Recall that Rambus sued all those DDR memory manufacturers. Then two weeks ago, Intel sued VIA. (see story). Then, VIA countersued Intel (see story).
Meanwhile, if all the legal wrangling can keep the lid on shipments of DDR memory with P4s, that may be all it takes to get Rambus through to next year when the company plans to release new technologies that will diversify its revenue stream, make the company far less dependent on one partner, and generate some value for Intel's warrants. It leaves me wondering just how coincidental Intel's timing of its DDR support is.
The new deal
Perhaps trying to quell speculation about how the previous contractual obligations were changed, and maybe thinking that all the aforementioned measures were not enough, Intel and Rambus surprised us with an announcement of a new deal that guarantees Rambus $8-10 million per quarter. It's a piece of the puzzle that fits nicely with the need to get Rambus into next year with at least some revenue. I wouldn't be surprised if, in an effort to grab its share of the holiday spending and fourth quarter budget burning, Intel ships its DDR solution ahead of schedule. According to Eulau, "The new agreement supercedes the earlier agreement and now Intel is free to do whatever it wants with respect to memory technologies, as long as it pays its royalty fees." Furthermore, Intel returned all of its warrants to Rambus thus reducing the agreement to a pure licensing deal.
While we wait for Intel to release its DDR-based technologies, you should do two things. Expect to pay the Intel tax when procuring new systems (unless you go with an AMD-based system) and remember that this is just a conspiracy theory. Intel flatly denies that its decisions were based on anything but providing the best solutions for its customers.
What do you think? Share your thoughts with your fellow readers at ZDNet TechUpdate's Talkback, or write directly to david.berlind@cnet.com.
Got a great tip? An industry rumor? Or do you want to submit your own column to ZDNet TechUpdate? Send David your submission, and if we use it, you'll be compensated with some of the cool vendor schwag that arrives in our mailboxes on a daily basis.