Cray Left His Company to Prevent Internal Conflict

Why did Seymour Cray leave Cray Research, the Minneapolis supercomputer company he founded 17 years ago, to form yet another supercomputer firm?

Some analysts have surmised that engineering genius Cray wanted to escape Wall Street’s fixation on quarterly financial reports. Others, including Cray Research officials, have said the company lacked the financial strength to support two rival supercomputer designs.

But both explanations are off the mark, says Cray Research Chairman John Rollwagen, in his first detailed explanation of the breakup announced last month.

According to Rollwagen, who likened the split to an “amicable divorce,” the division was necessary to prevent a potentially divisive and ego-crushing blow to Cray himself and a resulting clash among the company’s sensitive engineers.

Rollwagen said company executives realized many months ago that they could no longer continue supporting two supercomputer projects: the mainstream C-90 machine under development by one engineering team and the rival Cray 3, the innovative machine being designed by Seymour Cray.

Rollwagen said the two machines were virtually identical in size and performance and would have appealed to the same group of customers.

“A traditional company would have chosen just one design and shut down the other project,” he said. And the logical choice, he added, was not Seymour Cray’s design, because it was too experimental and too far afield from the company’s existing product lineup.

“But not choosing Seymour’s machine would have torn the company apart,” said Rollwagen, who was handpicked to run the company by Cray. “It would have destroyed the momentum of his work on the project and would have polarized the company.”

Rollwagen said the decision to spin off Seymour Cray’s design team into a separate company, named Cray Computer and based in Colorado Springs, Colo., was made after the two men met April 26 to discuss how the two rival supercomputer designs should be handled.

After airing the issues, Rollwagen said the spinoff seemed the logical solution to both because it preserved Cray’s work, maintained the remaining company’s internal balance and gave the marketplace a choice between two machines.

“We created our own competitor,” Rollwagen concedes. “But it was the best choice available to us. I couldn’t imagine not choosing Seymour’s machine.”