Intel-Rambus New Contract

Format for Printing

Format for printing

Request Reprints


By ptnewell
September 18, 2001

Posts selected for this feature rarely stand alone. They are usually a part of an ongoing thread, and are out of context when presented here. The material should be read in that light. How are these posts selected? Click here to find out and nominate a post yourself!

This is my understanding of the Rambus/Intel press release, and the subsequent Rambus conference call. First the facts, as I infer them, then a little interpretation.

(a) The 1996 Intel-Rambus agreement is dead, as the new agreement supersedes it.

(b) Intel will pay Rambus a fixed amount of $10 million/quarter for the next 12 quarters. (Edelstone's questions were quite helpful here. He noted that the increase in after-tax earnings per share implied an increased revenue of $7-8 million for this quarter. The Rambus CFO, in answer, gave the $10 million/quarter figure for the Intel payment under the new contract. The clear implication is that Intel is currently paying Rambus $2-3 million/quarter).

(c) Among the effects of canceling the 1996 agreement were canceling the Intel warrants. Intel no longer has the conditional possibility (under very strict RDRAM usage requirements) to buy 4 million shares at $2.50/share. It is not completely clear whether the much smaller sized warrants issued to certain DRAM makers are similarly cancelled. In theory, this is good news for current shareholders, since it eliminates the potential dilution.

(d) The agreement is very broad and thus vague about what the payments are actually for, probably deliberately so. That is, it will probably not matter what the courts decide about Rambus's claims to SDRAM and DDR. Intel pays a fixed $10 milllion/quarter for the use of Rambus's patent portfolio. (Of course, at the end of 5 years, it may matter in the renegotiation.)

(e) Rambus has in turn won some patent rights it believed it needed to develop certain Infiniband associated products. It is not entirely clear why a pure IP company that does no fabrication felt it needed such rights.


The agreement puts a solid floor underneath Rambus's earnings and valuation, at the expense of a somewhat lower ceiling. The floor would seem to be the $2/share they have in tangible value (mostly cash), coupled with the guaranteed payments of $200 million ($2/share). Actually the present value of these payments is less than a $200 million up front payment would be, but conversely the payments may well be renewed beyond the 5 year time span if Rambus does not lose its whole patent portfolio in courts. $2/share seems a modest valuation for the agreement, given the possibility that enough patents will hold for Intel to find it easier to renew in 5 years. Thus Rambus should not (in a rational market) drop below $4/share.

However fixed payments do not grow from quarter to quarter. Moreover, it seems almost certain that if Rambus had clearly won in courts the right to charge all makers of DRAM logic controller elements, Intel would have had to pay much more than $10 million/quarter. (The Rambus claims extend to memory controllers, patent rights Intel would have absolutely needed to stay in business). Thus the possible growth in logic element payments has been severely restricted.

Any remaining tatters of a "special" Rambus-Intel relationship seem dead. Intel will no longer have a big potential stake in Rambus shares. The agreement probably occurred now because Intel is preparing to roll out the Brookdale, which will support both SDRAM and DDR. (Incidentally, as I have previously noted, even the PC133 version of the Brookdale apparently violates the original Rambus-Intel contract, see below). Intel could have voided the contract at any time, however that would have created an awkward legal situation (for one thing, it definitely does need permission to use RDRAM, which still dominates Intel's high end at this point).

However Intel was clearly already intending to "let the market decide" between the various widespread memory technologies. And Rambus is now in a fairly strong position. RDRAM is already bringing probably $10-11 million/quarter in royalty payments in January-March, and very likely more now. With this $10 million/quarter (actually only $7-8 million in new money, since recall, Intel was already apparently going to pay $2-3 million for the current quarter), Rambus is at the point where it would be profitable with only a small faction of the PC market (in fact very profitable if not for the legal costs, and solidly profitable even with them). Most of the RDRAM usage is STILL not in PCs. (About 30 million/month, or, say, 90 million for the quarter, 128-Mbit equivalents proposed to be shipped July-September, while P4 usage is probably less than 3 million x 12 chips = 36 million for the whole quarter). Any growth in RDRAM usage in PCs, HDTV's, or elsewhere will still of course result in greater payments from DRAM makers. With the first three licensees of RaSer (the serdes product) all in the last quarter, and the promise of RaSer products shipping in the first quarter of next year, Rambus earnings should certainly remain solidly positive.

In effect, Rambus shareholders now have a solid floor of about $4.00/share beneath them. The sky above is not near as high as seemed possible last year (for one thing, the DRAM market is not near as big, and for another Intel will never have to cough up the big bucks for building controllers). Yet the chance still exists for RDRAM to dominate ultimately (as I think it will), and for QSRL and RaSer to add to the earnings. The advantage of being a pure IP company, of course, is that anything from the latter two will be essentially pure gravy, moving straight to the bottom line (after taxes).


The original Rambus-Intel contract seems to limit Intel's abilities to use even PC-133, which on a 64-bit bus produces 1.06 GB/s transfer rate. The pertinent section of the now-void 1996 agreement which restricts Intel's ability to develop new memory interfaces is below:

"(i) For purposes of this Subsection (b), "New Interface" means any interface for PC main memory applications, other than main memory interfaces on Intel's chipsets shipped prior to the second calendar quarter of 1998, and evolution of such main memory interfaces extending there from. Any DRAM interface which provides greater than one (1) Gigabyte/second/device bandwidth is considered a New Interface."


TMF Money Advisor
Got money questions? Your answers are just a phone call away! TMF Money Advisor puts you in touch with an objective Financial Planner whenever you need it.