Control Data’s Failure Wasn’t Technological
Money talked on Monday when Control Data withdrew from competition in supercomputers. It couldn’t continue financing a money-losing but technologically advanced operation, said its Chairman Robert Price.
The pullout is a bad sign, but not for the reasons you may think. It’s not that the United States--the country that worries about its schools turning out enough engineers--is falling behind in technology. In fact, there is little doubt that the U.S. industry leads in supercomputer technology.
It was U.S. financial markets that failed Control Data. The real story is that the nation that turns out more business school graduates than any other is threatened competitively because it cannot figure out how to finance a business over the long pull, how to stay the course when the going gets rough.
Supercomputers, machines that can perform billions of mathematical calculations per second, are important. They are used today to design airplane wings, simulate weapons systems, find oil deposits or decipher photographs from outer space. In the next decade, as their powers increase, so will their value and importance.
Threw Towel In
Yet sadly, Control Data, the Minneapolis company that first developed such computers in the early 1970s, threw in the towel even though it turns out one of the world’s most powerful supercomputers. Its ETA Systems computer had technical flaws and needed software. But Control Data has sold 34 ETAs worldwide--in a global market that has seen only 409 supercomputers sold to date. It had possibilities.
And with time Control Data might have rejiggered its machine, developed the software. But nobody would take a chance on it. Banks wouldn’t extend its credit lines, and the stock market seemed happy to accept its decision to close the business and restructure; the speculators said the company was “creating shareholder value.”
Control Data’s departure leaves U.S. industry with two prominent supercomputer players: Cray Research--the global leader with more than 200 machines sold--and giant IBM, which has declared its intention to enter the field and is backing entrepreneur Steve Chen’s company, Supercomputer Systems Inc. A variety of supercomputer start-ups round out the U.S. industry.
Japan, meanwhile, has three major entrants in the supercomputer olympics: Fujitsu, NEC and Hitachi. The Japanese companies are not making any money in the business, but they are persevering. NEC, for example, in a joint venture with America’s Honeywell, has just brought out a powerful new supercomputer. It has no orders but it is patient.
What are the U.S. strengths? Technical brilliance and venturesome entrepreneurs, who are advancing the state of the art. Both Cray Research and Supercomputer Systems are going beyond one big machine to parallel processing, which breaks down massive problems, gets many parts of answers quickly and puts them back together--as you might do in developing a satellite photograph.
The U.S. industry also is bringing new power to supercomputing with an effort by Intel Corp., the semiconductor company, to put together 2,000 chips, of 1 million transistors each, to produce the most powerful computers yet made. Intel is working under a contract from the Defense Department on that advanced development.
But most U.S. financing is entrepreneurial. Cray Research, now grown to $800 million in revenue, was founded by Seymour Cray, 62, who left Control Data in 1972 to form his own firm--with backing from Control Data. Two years ago, Steve Chen left Cray to form Supercomputer Systems, and attracted backing from IBM.
Other companies are springing up: Thinking Machines of Cambridge, Mass., and Kendall Square Research of Boston, Tera Corp. of Seattle and companies named Mips and Convex, Masspar and Alliant, Solbourne Computer and Key Laboratories.
Which is exciting, but the staying power of small firms is suspect. Gary Smaby, a supercomputer specialist for Needham & Co., an investment bank that backs start-up companies, believes the next decade will demand larger firms with the resources for a high-stakes game--Control Data, after all, lost $100 million in supercomputers last year.
Without patient financing, home-grown technology may leave home. Two promising newcomers in supercomputers are already owned or backed by Japanese companies: Key Computer Labs, of Fremont, Calif., was recently bought by Amdahl Corp., of which Fujitsu owns 45%, and Solbourne Computer of Longmont, Colo., is backed by Matsushita.
“Capital is a competitive threat,” says Intel’s President Andrew Grove, who sees Japanese financing being offered to start-up companies up and down Silicon Valley. But does it matter who finances technology? Yes, says Grove “because control of technology leads to jobs and development for your people and leadership for your country.”
True enough, but unless U.S. financial markets recognize there are values other than shareholder values, that leadership will fade.