Christmas came a bit early this year for supercomputer designer Steve Chen.

On Tuesday, the soft-spoken 43-year-old engineer, who dreams of designing the world's fastest computer, announced that International Business Machines Corp. had agreed to back his fledgling $100 million venture. Chen calls it a sure-win combination that will produce a supercomputer unlike any the world has known.

"We have great design capabilities," Chen said of Supercomputer Systems Inc., his three-month-old company in Eau Claire, Wis. "And IBM has the best technology -- and a lot of money."

The partnership with IBM is the latest example of what Chen's friends and former associates say is his single-minded determination to realize his ambition, despite the obstacles, and to establish himself as one of the enduring leaders of the computer industry.

Chen founded Supercomputer Systems in September, immediately after Cray Research, where he had directed development of state-of-the-art supercomputers since 1979, cut off funding on his latest project. Cray executives said the program had become too costly and futuristic. Chen, who routinely works 60 or more hours a week, disagreed, and set out to prove it.

"The only way to succeed is to go ahead and do something when others say it can't be done," said Norman Winningstad, chairman and founder of Floating Point Systems in Beaverton, Ore., where Chen started his career in computer design. "Steve is bright, dedicated and opinionated, as most people of his stripe are."

Chen said he is convinced that deep pockets -- and an even deeper reservoir of patience -- is what he needs to build the world's fastest computer. And IBM, which has never developed its own supercomputer operation, offers exactly that, Chen said.

"They share the long-term view that it takes time to build a great machine," Chen said in an interview after the announcement. "This is not like a Radio Shack computer."

The computer Chen envisions would be able to complete millions of tasks per second and would be used to solve complex problems in the most sophisticated laboratories, universities and defense installations in the world. Supercomputers are used to simulate wind-tunnel tests and automobile crashes and to perform elaborate chemical analyses and detailed seismographic studies.

When details of the IBM partnership are worked out early next year, Chen said he will have the resources to build such a machine by 1993. The pending deal calls for IBM to provide an unspecified amount of financial backing and technology as well as to assign some of its engineers to the project. In exchange, IBM -- which approached Chen almost immediately after he left Cray -- would get an interest, albeit a noncontrolling stake, in the business.

Although Chen professes to fare poorly in a rigid bureaucracy, he said the arrangement with IBM allows enough autonomy for Supercomputer Systems.

"This is a partnership, not a merger. They'll give us the money, technology and participation and then leave us alone in the Wisconsin woods," Chen said. "Staying separate from IBM is a very significant point for me. They understand that if we were to merge we could lose our creativity and ability to move fast."

Until the IBM financing is arranged, Supercomputer Systems is operating on the money Chen and seven other engineers pooled when the company opened its doors in a office building whose rent is subsidized by the City of Eau Claire. The founders are among the 40 former Cray employes who left when Chen's project was canceled.

Chen is reluctant to talk about his days at Cray, the No. 1 supercomputer company, saying only that the decision to leave was emotionally wrenching, but professionally satisfying.

Chen conceded that he sometimes goes overboard in his attempt to carve out a reputation as an "imaginative and visionary computer architect."

"Maybe sometimes I'm too aggressive. I stretch farther than others can imagine," he said. "They have to tell me to back off and get back to reality."