A Playbook For Taking On Big Blue

Fujitsu's chairman talks about betting on fast computers and chips

Long before there were PCs in every home, Japan's Fujitsu Ltd. battled IBM (IBM ) in the market for giant mainframes. The PC revolution undercut that business. But in early April, Fujitsu again took on Big Blue -- this time with its new Primequest server computers, which run on chipsets Fujitsu designed around Intel Corp.'s (INTC ) powerful Itanium processor. Fujitsu Chairman Naoyuki Akikusa recently talked with Business Week Online writer Burt Helm about the new server and other technology trends. Here are some edited excerpts:

Tell us about Fujitsu's computer strategy.

Servers are the next big product area for Fujitsu. For three years we have been collaborating with Intel, jointly developing a chip set that is going to provide mainframe-class functionality and reliability to servers. These could compete directly with IBM's Power series of computers. Customers who are using big mainframes today will have to phase them out within five to 10 years. Then we will be able to replace their equipment with next-generation servers running Linux or some future version of Windows.

Does that require that you stay focused on the cutting edge in chips?

Yes. Our most advanced semiconductor factory, in Mie prefecture, is producing 300-millimeter wafers with circuits 90 nanometers wide. We were able to get the plant up and running in record time.

What makes the chip business appealing?

The various functions of individual products are gradually converging onto single chips. In the end, the chip itself will be the product. This is an important technology trend. Right now chips are only about 10% of our revenues, but that could go up to 15% -- it depends on what kinds of customers we attract.

You recently curtailed operations in plasma displays for flat TVs. Why is that?

Fujitsu has a lot of patents in this area and excellent engineers. But if you want to make plasma a main part of your business, you need to invest $1 billion to $1.5 billion every year. In contrast to companies that make television sets, we only make a component -- the panels -- so we have no ability to see the entire market. The consumer-electronics business requires different sales channels, different expertise from what Fujitsu has. Ultimately we decided it would be better to put our money into semiconductors.

Why have Korean companies been so successful in the flat-panel market?

It seems to me that the business model for Korean companies is somewhat different from that of Japanese companies. When the Japanese invest $1 billion to build a factory, they focus on high productivity from the beginning, trying to reduce the number of defective parts. Looking at the strategy of Koreans, their priority is summed up by "faster to market and cheaper to market." They take an aggressive pricing strategy, so their competitors give up. Memory chips are a good example. Because of aggressive pricing and marketing by Korean companies, many Japanese companies withdrew from the memory market. Japanese managers tend to focus on getting faster returns on their investment.

Do you see a threat from Korea in advanced semiconductors?

When I talk to top managers at Korean companies, they say their companies are focused on middle-tier technology, not the high end. It seems they would rather leave ultrahigh tech to the Japanese. But at the end of the day, the winners are the ones that make big profits. The Koreans are very successful.

What else does Fujitsu see on the technology horizon?

We have started planning for future supercomputers with a cluster architecture that will produce speeds of one petaflops [one thousand trillion operations per second]. We have put an engineering team together, and we are aiming for 2010.

What do you think of nanotech?

You can read about nanotech in any newspaper these days. Fujitsu, for example, has a battery that uses nanotech. But I think it will still take another 10 to 15 years to develop as a practical business. I strongly dislike the hype about nanotech. But it's O.K. to talk about a technology, even if it doesn't have immediate marketability. That's because it's very important for a high-tech company to have a dream.

    Before it's here, it's on the Bloomberg Terminal.