Cray Research is eyeing a whole new market and a whole new computing approach. The CS6400 question is, will commercial customers accept it?
by Michael Finley
Exclusive to Inside/PC Week
Copyright © 1995 by Michael Finley. All rights reserved.
Many companies do a double take when they first hear about it: Scientific supercomputer giant Cray Research is dipping its toe in commercial waters? That's like the National Rifle Association switching to bows and arrows, or Cher taking a shine to older men.
Yet, as Cray nears its second full year of making and selling commercial superservers, that's what its Business Systems Division is doing. One small corner of the company has turned its back on the smallish ultra-high-end computing-intensive market, and is busily hawking its CS6400 superservers to any and all business takers, and at prices as low as $225,000.
Business Systems is off to a good start, selling 34 systems in 1994, and earning an estimated $119 million, and showing signs of ramping up rapidly up this year, especially as visible new customers are announced. Corporate is determined not to smother its offspring, which was upgraded this January from a subsidiary to a division. Cray's installed base of commercial customers is short, but impressive: Bank of America, Britain's Churchill Insurance, and Fuji Bank.
One of its biggest catches so far is the good old Automobile Association of America. In a contract with an estimated value of $1 million, AAA installed a CS6400 as the hub for an online information system that makes available to its clubs air schedules, hotel rates, car rentals, maps, trip planning information. Says AAA CIO Lou Fresquez: "We were satisfied that if the system was reliable enough for Cray's corporate customers with truly mission critical needs, like Wall Street firms, it would be plenty safe for us."
It's not a natural thing to switch from selling $30 million machines to $400,000 machines overnight, though. It is a move that many companies (Kendall Square, Thinking Machines) have yearned, but failed, to do. Why is Cray still standing after other players have fallen?
They begin with a formidable core competency: a knack for stupendous scale. Bobbi Hazard, VP and general manager of Cray Business Systems, insists the idea for a commercial superserver was the customers', not the company's. "They were coming to us saying, we know you're good at solving huge computing problems. Can you do solve huge data problems as well?" Cray researched the market and found a hefty niche for itself: selling high-end superservers that combined the openness of SPARC and Solaris with beefed up throughput and stronger memory connects. The result was the CS6400, a four-bus behemoth scaleable to 64SuperSPARC processors, with the muscle to move 5 Terabyte relational databases without breaking a sweat. Interesting feature: the CS6400 boasts power supplies that can be replaced while the machine is still up.
Beyond throw-weight, Cray is stressing the CS6400's affordability. Many an IT manager, hearing there is a sales rep for a company that sells $30 million computers out in the lobby, will eye the nearest window. Jeff Pancottine, director of marketing at Business Systems, says Cray itself was stunned by results of audited transactions-per-second tests that showed the Cray box outperforming roughly equivalent-class systems from IBM, Pyramid, Hewlett-Packard, and DEC.
Cray is giving its young guns a free hand. Last year Cray purchased Floating Point Systems, a maker of SPARC scientific boxes, and used their expertise and their Beaverton, Ore., site as the foundation for ambitious new team with deep experience in the high-end UNIX world. Bobbi Hazard previously served as VP Sales for nCUBE, and before that with Digital and Data General. Cray gave the designers autonomy. Business Systems has its own marketing staff and distribution channels, and they have made great headway for a first-year product -- $119 million in sales. That's a 5 percent drop in Cray's overall revenue bucket today, but it could swell to flash flood proportions as commercial markets get wind.
The company is fronted by "a respected graybeard." That's how one analyst described incoming Chairman and CEO Phil Sampere. Sampere was the driving force behind Sun's successes in the 80s. His presence gives Cray not just his technology marketing experience from his days with Sun Microsystems. he also sports a bankable Fortune 100 pedigree as recent president of Eastman Kodak. Whatever dimensions Sampere adds at Cray's corporate level, his background in UNIX boxes and commercial apps makes him a gold mine for Business Systems Division.
It's a hard cold world to sell to when you don't already have customers in that market. Cray has "a bird in the hand" -- an existing base of scientific customers nearing the stage where they were encountering the same sort of data handling chores as commercial customers. So Cray is able to sell looking forward to new customers while accommodating existing ones. Altogether Cray has identified three target industries that its sees as a good fit for its machines: financial organizations, telecom companies, and the government.
The big difference between Cray's older customers and the new ones it's courting are expectations. The old scientific customers were on the same page with Cray technologically. If a system went down, they whipped out a screwdriver themselves, dove in, and fixed it on the spot. Corporate IT departments look less kindly on reliability problems. So Cray has had to beef up not just its hardware reliability, but its hand holding systems.
Competitors don't want Cray's hands anywhere near them. "They've got a hotbox, period" says Bob Robinson, product marketing manager at Sequent, makers of the Symmetry 5000 SE30 and SE70 systems, just five miles down the road from Cray in Beaverton.
"But customers want more than that today. They want complete solutions, a relationship they can trust. Cray can't offer that as yet. That's why they sold a total of seven systems in the entire first quarter."
Cray replies that it is in the market to win, and if competitors don't like it, tough. In January the company redesignated the commercial computer group from a subsidiary to a full-fledged division.
"Even my sister in law knows Cray makes fast and powerful scientific computers," Pancottine says. "Sure, that kind of familiarity can be a double-edged sword." Cray must now convince commercial clients that it can be a player in the everyday business hurly-burly. It is a do/die challenge, says Jeff Liebel, research analyst for Minneapolis' Smaby Group. "We see the supercomputing market shrinking 8 percent annually through 1999. Cray has been working hard to improve both its top and bottom lines, trimming expenses and moving into these broader markets."
This foray out of labcoat country and into the business bazaar is Cray's best chance to reverse that trend. Five years from now, who knows? We may hear the name Cray and not even think of supercomputers. x
Michael Finley is a communications consultant in St. Paul, Minn. He can be reached at mfinley@skypoint.com.