The Numbers Aren't Crunching Cray Research

Analyst Barry F. Bosak has a hand on his forehead and a bilious ex pression on his face. His tablemates stare with consternation at the gluey-cheese "quicheburger" lunch served up following Cray Research Inc.'s May 14 annual meeting in Minneapolis. But the stock market, not food, is what has Bosak upset. Even as he dines, Cray Research shares are trading down, down, down, to end the day 15% lower, at 32 3/4, on bad earnings news (chart). "I had a 'buy' on the stock," moans Bosak of Smith Barney, Harris Upham & Co., his eyes searching for some sympathy.

Like plenty of other analysts, not to mention stockholders, Bosak wonders when supercomputer king Cray Research will get its earnings act together. After a period of flat sales, revenues are beginning to take wing. And the company keeps insisting that revenues will climb at least 10% this year, to $949 million, atop a 7.2% boost last year. But profits have not kept pace.

CHEERLEADER. In April, Chairman John A. Rollwagen told stock analysts the company would scratch out at least a small earnings gain this year, after a measly 3.2% increase in 1991. But on May 13 he changed his mind, announcing instead that earnings might plunge 20%. The surprise announcement sent the stock tumbling the next day. "To have down earnings and revenue up 10% really doesn't jibe," says a puzzled Bosak.

Despite the bad news, the ever-ebullient Rollwagen is pushing the cheer meter to its limits. "Talk is cheap, I know that," Rollwagen says. "But I'm very optimistic about where we're going and where we'll be a year from now. This is one of the great buying opportunities for Cray Research stock."

At last, Rollwagen believes, the grand strategy he began stitching together after founder and former chairman Seymour R. Cray resigned in 1989 is taking hold. Cray left to pursue his own unorthodox design for the fastest computer on earth. But Rollwagen stayed put to push a different vision: He thinks a robust supercomputer company should offer an array of products to provide superpower to a customer's entire computing network--a concept known as network supercomputing. Seymour Cray's troubled Cray Computer Corp. has yet to ship its first product; Rollwagen's approach is piling up converts.

Superfast machines with price tags as high as $30.5 million remain Cray Research's bread and butter, but Rollwagen now is trying to seed new customers with starter machines--an anathema to Seymour Cray--like the new Y-MP EL, which costs as little as $300,000 and will be sold in a cooperative arrangement by Digital Equipment Corp.'s sales force. And while Seymour Cray saw software as an afterthought, Rollwagen has dramatically boosted spending on operating-system, networking, and applications software to the point where it now accounts for half of research-and-development spending.

To help link his supercomputers with networks of science and engineering workstations made by Sun Microsystems Inc., Rollwagen late last year bought networking technology from bankrupt FPS Computing, formerly known as Floating Point Systems Inc. And with the help of $12.7 million in Pentagon funding, Cray Research launched a project to build massively parallel supercomputers, which would link 1,000 or more microprocessors to reach nearly unfathomable computing speeds of up to 1 trillion calculations a second.

FRUITFUL STRATEGY? Over the next 12 months, Rollwagen expects this strategy to bear fruit. Cray Research's brand-new Y-MP C90, which costs as much as $30.5 million, already has taken at least nine orders. At the low end, the company has sold 50 of its new EL machines and expects to build up to a production rate of 350 a year.

But it's earnings, not sales momentum, that has analysts worried. Now, Rollwagen must prove he's as good an operator as he is a strategist. Part of this year's profit drop stems from tougher-than-expected price competition in midrange systems from the likes of Convex Computer Corp. and IBM. But blame manufacturing, too. High scrap rates on the C90's 52-layer, gold-plated circuit boards will trim several million off profits this year. Overhead could use some trimming as well. Rollwagen says he'll keep total employment at current levels but move executives from headquarters into the field. Meantime, he'll be driving hard on day-to-day cost control: Rollwagen himself has abandoned first-class flying.

Wall Street seems to be giving Rollwagen the benefit of the doubt. Although the stock hasn't recovered from its mid-May stumble, most analysts, expecting strong 1993 earnings, still rate it a buy. And Rollwagen boasts that the next time Cray Research stock goes as low as its recent low of32, it will be because the stock hassplit.

That would be good news for Cray Research's luncheon guests. It might even make those gluey quicheburgers palatable.

    Before it's here, it's on the Bloomberg Terminal.