The Supercomputer: An Impractical Vision?
The world’s most advanced supercomputer, now undergoing a test at the National Center for Atmospheric Research in Boulder, Colo., represents a triumph of technological vision and entrepreneurial daring. But it may not represent a commercial triumph.
And thereby hangs a tale of a man and his vision, but also of the fast-changing marketplace in supercomputers and every other business today. It’s a tale of the humbling of genius and the explosion of business possibilities that demands we see the future differently from the recent past.
The machine being tested in Boulder is the culmination of the career of Seymour Cray, founder and guiding genius of not one but three computer companies--Control Data in 1957, Cray Research in 1972 and Cray Computer in 1989.
At each of his companies, Cray, now 67, strove to build the most powerful computers, machines able to do billions of math calculations per second. These were computers that could design an automobile or chart a hurricane or re-create photographs sent back from Mars and Jupiter.
In that sophisticated technology, Cray was the U.S. standard bearer in a competitive struggle through the 1980s with Japan’s supercomputer makers Fujitsu and NEC.
Then in 1989, with the competitive battle raging, Cray--a 1950 math and electrical engineering graduate of the University of Minnesota--decided to push technology to a new level. With backing from Cray Research, which spun off the new company to shareholders, he founded Cray Computer in Colorado Springs and set to work on a new kind of semiconductor chip made from gallium arsenide, a material that conducts electrical impulses with less resistance than silicon and thus less heat and energy loss.
The computer being tested in Boulder is based on gallium arsenide, a compound of two volatile metals that is still in the research stage in laboratories here and abroad. It will have an important future in computing and electronics, say computer experts. It is remarkable that a commercial company has produced something so advanced.
Yet there is no rejoicing at Cray Computer. In fact, the company is in trouble.
Difficulties in harnessing the new technology delayed the new computer, and a major order was lost in 1991. Cray Computer has not received another order--the Atmospheric Research Center is using the machine free as a demonstration--and the company needs fresh financing.
Seymour Cray told shareholders recently that the company has enough cash to survive until September, adding that if need be, he will invest more of his own money. He already owns more than 20% of the stock. “I certainly don’t believe the company is going to go out of business,” Cray said.
But Cray Computer stock, now selling on NASDAQ for $3 a share, down from $19.63 in 1991, indicates that others are less hopeful.
What has happened is that the marketplace changed even as Cray was creating his newest supercomputer. Suddenly, smaller and cheaper machines are able to perform supercomputing tasks.
“Companies today design airplane wings or simulate molecular structures using workstations from Silicon Graphics or Sun Microsystems,” says David Frankel, director of technology at Smaby Group, a Minneapolis investment firm specializing in supercomputer research.
Also, companies and government agencies are using masses of cheap computer chips in “parallel processing” to perform tasks that formerly required the sophisticated and expensive circuitry found in Cray machines.
And the marketplace has shrunk as the Cold War’s end reduces Pentagon buying.
So it may seem that the stock market’s baleful vision will prevail over Cray’s more hopeful one.
But that is premature. Howard Watts, Cray’s marketing vice president, points out that government laboratories and the National Aeronautics and Space Administration still need supercomputers--”and we have a cost-effective machine.”
Moreover, Cray is already on its way to producing a more powerful and cost-efficient supercomputer by the end of next year. So it is too early to count the company out.
But more important, Cray and company have learned a lesson that all business can share: In today’s marketplace, many suppliers will have a host of solutions to a multitude of problems. The age of the single supercomputer is over, the time of the many has dawned. “There will be small computers for some uses, large ones for others,” Watts says.
That means the potential market for any one computer will be smaller, and it is risky to focus on a single absolute technology. Seymour Cray made that mistake, striving for the one powerful computer that would put all others in the shade. Instead, customers sought and found other means.
Ironically, Cray’s 1980s competitors, Fujitsu and NEC, followed Cray’s example.
They tried for the biggest, most powerful supercomputers. And they too have fallen in world market share.
But we should keep in mind that if diversity is the watchword of the ‘90s, Cray helped make that possible too.
Along with the supercomputer, Cray also developed the reduced instruction set, or RISC, computer chips that have made markets so diverse and abundant.
That is to say, even if his latest venture proves a commercial loss, we all gain from the vision and drive of Seymour Cray.
Not So Super Cray Computer’s vision of an advanced supercomputer has collided with commercial realities and the company has not been able to sell a single machine. Its stock has plummeted almost 85% from its all-time high of $19.63 pershare in October, 1991.
Tuesday close: $3.00, up $0.38 Source: Dow Jones News Retrieval.
