FEATURES & COMMENTARY
New York, N.Y. — Donna Fuscaldo reports that Ford Motor Corp. used one to figure out which states had the most defective Firestone tires. Charles Schwab Corp. uses one to run its massive online trading site. Sprint Corp. PCS Group uses one to manage its large database of wireless customer information.
Long gone are the days when supercomputers were used only by government agencies and research institutions. These mammoth computers, with their huge memories and record-breaking speeds, are increasingly being used by corporations ranging from large financial institutions to designers of golf balls.
In fact, two of International Business Machines Corp.’s largest supercomputer customers aren’t government agencies or universities at all, but rather financial giant Schwab and telecommunications behemoth Deutsche Telekom AG.
So why has this sector, once the darling of Wall Street, fallen off the radar screen of many in the investment community?
Industry watchers partly blame the Internet for stealing the sector’s thunder, while makers of high-performance computers say people just haven’t realized that the machines are used in many different businesses, in many different ways.
Supercomputers were historically used for massive data computing, which still aids in weather forecasting, space travel, building nuclear weapons and helping find cures for diseases.
These complex machines, which came into prominence at the end of the 1970s, were once regarded with awe. But like most industries, supercomputers went through a metamorphosis and Wall Street stopped being enamored of them, according to Thomas Tabor, publisher of HPCwire, a Web site that follows the high-performance computing market.
In the early 1990s the economy wasn’t strong, and the U.S. was involved with the Gulf War. Those events and a number of other factors overshadowed the high-end computing industry, Mr. Tabor said. On top of that, the Internet came along and was “shining so bright that most Wall Street money went into that,” he said.
Although getting a big supercomputer contract – known as a “win” in the industry – still garners attention in the media, it is the business-type applications such as data mining and business intelligence that are spurring a new wave of growth.
Evolving are new uses for very high-end parallel computing in parts of e-business where there is a tremendous need to process massive amounts of data, said Brad Day, senior analyst at Giga Inc., a high-tech research firm. Companies are using supercomputers for such applications as data modeling and data warehousing, he said.
Supercomputers are present in “mundane businesses” and the investment community hasn’t fully grasped that trend, said David Turek, vice president of Deep Computing at IBM. “Most still think it’s for science and engineering, which hasn’t been the case since the mid-90s.”
In 1999, the high-performance computer industry, which encompasses scientific and engineering uses as well as business applications, was a $6 billion market, said Debra Goldfarb, group vice president of world-wide systems and servers at International Data Corp. She forecasts that the the market will grow 8% to 10% a year.
“Using a supercomputer means faster time to market, higher-quality products and reduced costs,” said Ms. Goldfarb. Companies are using simulation and modeling in applications ranging from manufacturing toilet seats to building nuclear weapons, she said. “Supercomputer wins are important to a company and extremely important to society,” Ms. Goldfarb added. Ford, Boeing Co., Bristol-Myers Squibb Co. and a hundred other companies couldn’t be competitive without supercomputers, she said. If you tack on the roughly $6 billion market Ms. Goldfarb estimates for support and peripherals such as storage, file servers and software, the market becomes a $12 billion one.
Supercomputers are important for the advancement of business and science, but there is also an element of bragging rights associated with supercomputer “wins.”
There is a certain amount of public relations around the vendors that can create the faster computers, said Giga’s Mr. Day. “You rarely see a press release as these vendors get commercial customers,” he said. “You often see releases when there is a extreme use in a high-end technical environment.”
IBM’s Deep Blue supercomputer is a classic example of hype surrounding an advancement in technology.
In April 1997, IBM’s Deep Blue computer defeated world chess champion Gary Kasparov in a weeklong chess rematch. Kasparov had defeated a lower-end IBM supercomputer in 1996. IBM spent an estimated $5 million on the rematch, including publicity, prizes and building and programming the computer. But the company said it reaped the equivalent of more than $100 million of favorable and free publicity.
“We used the Deep Blue as an event to get visibility,” said IBM’s Mr. Turek. “That doesn’t mean we created it just for that.”
Often the technology behind a supercomputer will trickle down into lower-end projects, he said. When IBM built the ASCI White supercomputer, “we built it with an eye toward bringing the technology to the broadest market,” he said. “We could have just built the ASCI and said here’s our bragging rights.” The ASCI White is used by the U.S. Department of Energy to run nuclear-test simulations.
Recently there has been a rash of announcements concerning high-profile supercomputer contracts. Earlier this month, the Energy Department awarded Compaq Computer Corp. a nearly $200 million contract to deliver the world’s fastest computer to the agency’s Los Alamos National Laboratory in New Mexico.
Compaq is trying to put together one of the fastest computers in the world, but the project has yet to translate into increased sales of its Unix server, said David Bailey, an analyst at Gerard Klauer Mattison.
“The big wins are proof of concept,” said Mr. Bailey. “The real uses like data mining, although important, may not get the buzz that doing weather forecasting does.”