CNET también está disponible en español.

Ir a español

Don't show this again

Christmas Gift Guide
Tech Industry

Rambus' main man in the memory war

CEO Harold Hughes says it's no fun suing people. But the company will do so if the memory makers won't agree to pay royalties.

It's Rambus on the phone. In Silicon Valley, them's fightin' words.

Emotions run high when it comes to Rambus, which designs chips and technology that enable data to transfer rapidly between two different components. Sony used memory from Rambus in the PlayStation 2 and will use other Rambus-designed products in the PlayStation 3.

The company is also involved in a hotly contested series of lawsuits with memory makers that span the globe. Last week, a jury ruled that South Korea's Hynix Semiconductor must pay Rambus more than $306 million in damages, and that's only for shipments of memory to the U.S. That's not a bad payday for a company with about 230 employees. Other suits against other manufacturers and in other jurisdictions are pending. In the end, the total could top $1 billion.

But CEO Harold Hughes says Rambus wants to move out of the court and become, effectively, an R&D arm for the industry.

Hughes is no stranger to most of these companies, having worked in the chip industry for 30 years. He started at Intel in 1974 on the same day and helped start Intel Capital while there.

Hughes sat down with CNET News.com to discuss his company's plans for revenue, the Hynix suit, and the ongoing arguments about intellectual property.

Q: During the conference call after the Hynix verdict was announced, you and (Rambus general counsel) John Danforth emphasized that the clarity of the decision may push some of the Hynix licensing negotiations forward. Have negotiations with other parties been on hold pending the outcome?
Harold Hughes: I've tried to maintain contact certainly with Hynix for quite some time by virtue of the amount of money involved and we were very confident that we owned the technology represented by the patents. But so far, there has not been sufficient consequence to losing trials.

What do you mean?
Hughes: The DRAM (dynamic random access memory) industry works fairly closely together in their legal representations and there is the theory that we would have to sweep the table across all competitors before we actually get to the point of forcing settlements.

I'm trying to take the discussion away from the courts and into the competitive DRAM market, where we believe that Hynix, in particular, has some competitive problems. In the DRAM market, there are obviously different segments; there is a high-performance segment, usually used in graphics. By working with someone like Hynix on a business basis, we can improve their profitability and by so doing, presumably, offset part of, or conceivably even more than, the royalty they would owe us, while at the same time making them a much more effective competitor and supplier.

I've used the expression that people view Rambus memory technology as an unguarded cookie jar.

So you'd help them design memory?
Hughes: Higher-end memory. Our products have run at higher speeds for quite some time, starting back with RDRAM (Rambus' DRAM). XDR (a high-speed memory from Rambus that will be in PlayStation 3) runs at a multiple of what is available in the market.

But we have to go through the courts, and that's how this started several years ago for better or worse, and there's no obvious way that we would extract ourselves from it.

One of the difficulties about this case is that Hynix could have settled a long time ago or agreed to a royalty for a lot less.
Hughes: Well, that depends. Remember, the $306 million related only to U.S. sales. We may actually even get some interest added to that, which when compounded over five years, in and of itself would become a real number. I would say the number represented by the $300 million is probably the minimal calculation as to how much they "sold in the U.S."

Another interesting aspect of the case is that the evidence favoring your side was pretty compelling. Some technical documents created by the defendants refer to your technology. One manufacturer talks in a memo about how the industry will need faster memory, but hopefully they won't have to pay Rambus royalties. Why have they fought so long? Is it a cultural thing in the DRAM industry?
Hughes: I had that very conversation, and I don't think it's a cultural thing. Who wants to pay a large amount of money? It's a $30 billion business. At 2 to 3 percent, you're starting to talk about very large amounts of money. The cost to defend against a lawsuit is a fraction of that. It compounds for us because we're litigating against three of them (Samsung Electronics, Nanya Technology and Micron Technology), so it's not an irrational process.

But at the end of the day, we're a well-financed, profitable company. We continue to invest in technology far beyond what's currently required by the market and likely to be used in generations of DRAM. Our business model is a sound one. If you have a company that's willing to take the risk that we do and invest in the technology development as far out as we do, why not see that as something that you could use advantageously?

Do you think it might help validate the intellectual-property business model? Right now, when you bring up patents and royalties, everyone gets the willies and starts complaining about the patent office and trolls.
Hughes: Well, obviously there is a big movement afoot to characterize IP as, let's say, a scurrilous activity. I've seen drafts of bills that are very IP-unfriendly. But let's step back and take a look at what is in the best interest of this country. The jobs that we develop and the technology we create is exactly focused on where a high-tech company should go.

We started as a company when there was a significant problem: Intel's ability to increase the speed of a microprocessor far outstripped the ability or the desire, quite frankly, of the DRAM industry to keep up. This meant that Intel would spend a lot of money driving processors fast and simply lose that value by virtue of the fact that more and more DRAM had to go into the box.

Well, in a fairly spectacular fashion. This was a time when Rambus was the proverbial two guys and a dog. I suspect it was profoundly embarrassing to the very capable and proud DRAM companies to have a small upstart like (us).

Things never got off on a good footing and much of the acrimony and the legal dueling going on now goes back to that very fact.

Also, the royalties that we charge have an implicit cost that is not exactly that high. I mean, implicit in the sale of a piece of a software is a royalty probably of multiple of ours. You just don't think about it. They don't package it is an IP sale--they package it as a sale of software.

I want a business deal, I'm tired of going to court.

Does Rambus have ongoing negotiations with PC makers or the graphics card makers--companies that have incorporated DDR (double data rate RAM) or DRAM into their products? And will you be seeking royalty deals out of them?
Hughes: We always have discussions. We meet with Intel as often as we can and discuss our technology going forward and we do the same thing with graphics manufacturers.

So it doesn't sound like you'll be looking for royalty claims against Dell or Gateway for incorporating products from Samsung Electronics and Hynix into their products.
Hughes: There are...I wouldn't state it as broadly as that. At the end of the day, we're trying to act in a way that's in the best interest of Rambus and its shareholders, and obviously we see Intel and AMD as people who are or should be allies, and their PC customers as people who are or should be allies. Our goal is to get a Rambus part as the main memory in PCs.

Well then, because you have a duty to your shareholders, do you think there's a duty to actually sign royalty agreements for the memory they (PC makers) have used or to collect royalties for it?
Hughes: No. I think it's far better that we go directly to the source.

Looking toward the future for a second, XDR will be part of the PlayStation 3, but who else is looking at it?
Hughes: We came out with the second variation called DDR, excuse me XDR2, which is focused on the graphics market. As we negotiate with DRAM companies, including the ability to manufacture high-performance graphics memory at a higher margin, we have to speak to the graphics companies. It's a chicken-and-egg problem.

How about the PC market, where XDR would serve as main memory? I haven't seen XDR show up on product road maps yet.
Hughes: It's even a bigger chicken-and-an-egg problem. The demand, and the need for supply, is at least two to three (times) larger with the (PC) manufacturers. In the graphics market, by virtue of its smaller size, we think that is the easier way for us to (get) into the PC.

Another issue that's come up in some of the other big patent cases is that the defendants have tried to do an end-run around the court system by asking the patent office to invalidate the patents. It came up in the RIM case and the eBay-MercExchange case. Is there any danger of that occurring to you?
Hughes: I think they scoured the Earth for prior art, literally. And they do have a joint defense agreement. I would assume that if they found some prior art, they would have presented it at the trial or the patent office by now.

I think the patent office has seen everything, gone over everything.

Will it ever end?
Hughes: I don't know. I can only tell you that I suspect that in DDR3, amazingly enough, some new Rambus technology will pop up. It won't be new to us but they will have "discovered" it. I haven't seen it; the actual DDR3 spec is not out yet. But I'm willing to bet on it.

I've used the expression that people view Rambus memory technology as an unguarded cookie jar.

Not completely. Infineon settled for several million and you got $306 million this week.
Hughes: Yes, but remember. That's appealable. I want a business deal, I'm tired of going to court.