Interview: Rambus’ chairman looks to the future

Digital consumer electronics, and the trend for chip makers to license IP (intellectual property) rather than design everything in-house, is helping boost the market for Rambus, according to Geoff Tate, chairman of the Los Altos, California, technology licensing company specialized in I/O (input/output) interfaces.

The IDG News Service interviewed Tate last week when he was in Bangalore, where the company has set up a design center that will have 50 staff by the end of the year. The interview took place before this week’s announcement of a truce in the company’s battle with Infineon Technologies over the intellectual property rights to memory technology. Starting in November Infineon will pay Rambus $5.85 million each quarter for a license to existing and future Rambus patents. The payments will continue until November 2007 with special provisions if Rambus is able to sign licensing deals with SDRAM (synchronous dynamic RAM) vendors.

In the edited interview below, Tate talks about the company’s strategy for making up for the shortage of quality engineers in the U.S., new opportunities in consumer electronics for the company, and the increasing demand from chip makers for licensable IP.

IDGNS: How does your new design center in India fit into your strategy for design and development ?

Tate: Our strategy for 14 years has been to keep everybody in one spot, because it made it much more effective to coordinate people if they were close together. But we reached a point where we decided that the advantages of being in one spot were being outweighed by the cost of living in California, and also that worldwide, talent is developing. It used to be that the U.S. put out most of the top students, whether they were from India or wherever. But now academically places are developing all over the world.

IDGNS: Did you evaluate locations such as Japan, Taiwan, and China, which would be perhaps closer both geographically and culturally to your customers in Asia?

Tate: We have sales and field application engineers in Japan, and have had them there for a long time. Our first subsidiary outside the U.S. was in Japan, and so we have had people in Japan since 12 years ago. In terms of design engineering, there are good engineers in Japan, but language is an issue. It is hard to find engineers in Japan who speak English to talk to headquarters where the technology is coming from, and they will probably have difficulty talking to customers outside Japan. Also, it is very expensive in Japan, just as expensive as California, if not more.

Taiwan seemed interesting, but a big disadvantage in Taiwan is that in hiring university students, there are mandatory military service requirements there. A lot of high-tech companies get the students through the military. So basically if you are not a Taiwan company, you are at a disadvantage in hiring talent out of school. We looked also at China, but China has issues, and one of the big issues is just travel. Our engineers need to go and visit customers, and in China you need to apply for visas to leave the country. If there is a problem with a customer, our customer wants them there the next day, which is not possible if our engineers have to apply for a visa.

IDGNS: Does Rambus now see an opportunity to design other parts of the motherboard of the PC and other devices, besides your traditional business in I/O interfaces ?

Tate: The I/O interface is a very critical, high value part of all of these chips. If you look at companies that are successful, most of them are successful because they found something that was really important, and they do it really well, and they became the best in the world, and then as a result they got a lot of business. Companies that sort of spread out and try to do everything usually fail because they don’t do anything very well.

It is always a balancing act. We have been increasing our product line. We started out just doing memory interfaces, and now we are doing all I/O interfaces. At some point we will move outside of I/O interfaces. But I think we have to be sure that we are only doing that after we have really solidly established ourselves as the number one in the I/O interface area.

IDGNS: Within the I/O space, what are the new opportunities that you will be looking at?

Tate: I think we are already doing almost all the I/Os there are on the high-performance side. We don’t yet have wireless. Today all wireless is done through totally dedicated chips. People want to work with us when I/O becomes a part of a chip. When the whole chip is just wireless, then they want to buy a chip, and we are not in the chip business. So there will come a day when wireless becomes a part of a chip, and that is when we would want to be a supplier. But that hasn’t happened yet, and may not for some time. There are also many low performance I/Os that we do not do, like USB, because they are relatively low value and people can do them themselves. But maybe someday we will offer those as well.

IDGNS: Has the integration of computing functions into consumer electronics translated into new opportunities for Rambus ?

Tate: As long as everything is in the PC, then there is really one customer — Intel. We have been successful in working with Intel. But there are many more devices that are becoming intelligent such as game systems and digital TVs. People in consumer electronics markets, that really didn’t have skills with this kind of high-speed product, all of a sudden need high-speed skills, and that makes them interested in working with us, because of our track record in high-speed computing.

IDGNS: How significant in your opinion is the shift from in-house design to IP licensing in the chip industry?

Tate: IP licensing is seeing a similar trend that we saw in semiconductors. There was no semiconductor business at one point because the systems companies built all their own chips. Then over 20 years companies started to pop up that were specialized, starting with DRAMs, I think. Some systems companies kept trying to build systems the old way, but they got killed by companies that shifted to doing it the new way. Chips are moving in the same direction. What we are seeing today is that almost all chips are involving outsourcing of significant chunks. If you look at microprocessors, most companies other than a few will go to somebody like an ARM or a MIPS to get the microprocessor core rather than design it themselves.