(FORTUNE Magazine) – T HE CLASSIC WAY for entrepreneurs to break into technology-based businesses is with dazzling innovations, leapfrogging the best products offered by established competitors. Yet Compaq Computer Corp., one of the most successful newcomers in U.S. corporate history, did it by assiduously holding innovation in check. Founded in Houston in 1982, a few months after IBM's PC went on the market, Compaq zoomed onstage with a 28-pound portable personal computer that, in its innards, was as close to a clone of the PC as the law allows. Once Compaq began selling the thing in 1983, sales climbed to what is almost certainly a first-year record: $111 million. When IBM struck back with its own portable last year, Compaq walloped the new machine and continues to outsell it 5 to 1. The upstart's sales last year tripled to an estimated $325 million. This breathtaking performance owes a lot to one exquisite fact: Compaq has yet to make a serious mistake. The company's record is a credit to its cofounder and chief executive, Joseph R. ''Rod'' Canion, a 40-year-old engineer from Texas Instruments so unswashbuckling that even a top aide admits he ''lacks sizzle.'' The game plan that brought Compaq this far, a strategy of piggybacking on IBM rather than leapfrogging it, has required cool tenacity. Canion has plenty of that, and he has proved a master of both the technology and marketing of microcomputers. The company is no mere IBM knockoff, and its strategy of fitting into IBM's world is highly sophisticated. Can Canion keep this up? The question sends Benjamin M. Rosen, the venture capitalist who is Compaq's chairman, into fits of frustration: Will investors, security analysts, and other fascinated observers, he asks, never give this company its due? The stock, which came out in December 1983 at $11 per share, had sunk below $4 by last October before retracking to about $7. Computer industry gurus can't name a company intent on emulating IBM that has earned respectable profits over the long run. The problem is that IBM makes unexpected moves; it can slash prices or come out with machines so tough to copy that emulators get left in the lurch. Last summer IBM slashed prices on its PC just before Compaq came out with its first nonportable line, the well- regarded Deskpro. In August, Big Blue introduced the IBM PC AT (for advanced technology), a dazzler that is not fully compatible with its own PC --and hence not with Compaq's portables or Deskpros. So the question of what Compaq does for an encore is crucial. Will Canion copy the AT? Does he have a new portable in the wings? And what can computerdom expect from Compaq Telecommunications Corp., a subsidiary set up last April to marry computers to telephones? Compaq executives behave like IBM clones in their reluctance to discuss their plans. The policy makes sense. Canion and Rosen remember what happened to Osborne Computer Corp.'s hot- selling portable: consumers stopped buying it after Adam Osborne began talking up a new machine, helping to push the company into bankruptcy. But Canion and other Compaq executives recently relaxed their sphinx-like policy and gave FORTUNE a glimpse of what the future may bring. The basic game plan, to stay compatible with IBM personal computers as long as most software is made for them, is ingrained at Compaq, and neither Canion nor Rosen has any thought of changing it. The strategy was not obvious three years ago when Canion, an industrial designer, and another Texas Instruments alumnus, James M. Harris, sketched out their portable computer on a place mat at the House of Pies in Houston. Canion, Harris, and Compaq's other founder, William H. Murto, also from TI, discovered the strategy while figuring out a marketing plan. It was quickly apparent that retailers weren't interested in computers that didn't offer customers lots of software--any more than they would have been in a record player that played only 27-rpm records. After checking around, the trio concluded that makers of hot software weren't going to redesign it for an unknown product. A lot of software, then, meant IBM software. Compaq's founders began a crash program to make their portable ''hard-core'' compatible. In an industry that worships innovation, they hired employees who could not only live within the limits of compatibility but also become obsessed with the job. ''I don't think you could get a 20-year-old to not try to satisfy his ego by 'improving' on IBM,'' says Vice President J. Steven Flannigan, 38, Compaq's programming chief and another TI expatriate. ''When you're fat, balding, and 40, and have a lot of patents already, you don't have to try.'' The programmers Compaq hired had an average of 15 years' experience. Even so, making a computer compatible isn't easy. Microsoft Corp. of Bellevue, Washington, had written the operating system for the IBM PC, software that established the playing field for word processing, spreadsheets, and other applications programs to operate on. Compaq could buy some of the Microsoft operating system and the Intel 8088 microprocessor chip that was the PC's brain. But special instructions copyrighted by IBM were built into a microchip to make the operating system run. Compaq's software engineers had to mimic this program, flaws and all, without violating the copyright. A team that grew to 15 programmers worked for nine months and spent $1 million to get the job done. Once the engineering was completed in late 1982, Compaq moved quickly to position itself with retailers before a big national computer show in November. It lined up Sears Roebuck and then went after ComputerLand, a large computer chain. While making that pitch, co-founder Murto got a quick lesson in a different kind of compatibility. A month before the show Murto raced around the country for three days with a fragile prototype, showing it to dealers who were members of ComputerLand's product review committee. In his hotel room in New Orleans, Murto discovered the only three-pronged electrical outlet was in the bathroom, so he propped the Compaq up on the toilet and squeezed five dealers around it for the demonstration. They were impressed. Compaq's distribution strategy, particularly its generosity to retailers, is a key ingredient of its success in the marketplace. Its products are priced to dealers so that they can gross 36% on sales made at the suggested retail price, vs. 33% on sales of IBM computers. Moreover, Compaq is the only major computer maker that sells solely to retailers; it doesn't have its own direct sales force skimming off business with large corporations, even though direct sales account for 25% to 35% of all micros sold. To ensure good relations with retailers, Canion hired away a key member of the IBM team that developed the personal computer, H. L. ''Sparky'' Sparks, 52. Sparks, who developed IBM's network of authorized dealers, didn't come cheap. Canion offered him options on 98,906 shares of Compaq stock at 53 cents each, a $100,000 up-front bonus, and a salary of $150,000--$30,000 higher than Canion's own at the time. The result of all these efforts is that Compaq is one of the few companies other than IBM and Apple Computer to command a vast swath of shelf space in computer stores. Compaq takes advantage of the dealers' loyalty by using them in market research. The company brings key dealers to headquarters several times a year and relentlessly quizzes them on why products are selling or what customers are asking for in the stores. ''I'm always exhausted when I'm done,'' says Jeffrey D. McKeever, president of MicroAge Computer Stores, an Arizona-based chain of 133 stores, who has attended the all-day grillings. Warren Winger, chairman of Dallas-based CompuShop, another large retail chain, says Compaq is a better listener than IBM. He observes: ''Compaq's success isn't from the latest technology or a lot of razzle-dazzle, but from coming out with what dealers want.'' Compaq's early growth was propelled partly by shortages of IBM PCs, but the trajectory continued when the shortages ended in 1984. And when IBM rolled out its own portable last February, it proved a dud. It runs fewer IBM PC programs than the Compaq, is two pounds heavier, and has an orange-on-black screen--Compaq's is green on black--that many users dislike. The power of the IBM trademark caused a temporary dip in Compaq sales. They recovered by September, though, and since then IBM's portable has settled into a slump. When Compaq introduced its first nonportable machines last June, the four- model Deskpro family was a hit. But again, no razzle-dazzle, just some new features that reflect what Compaq learned from talking to dealers and from other market research. D ESKPROS use an Intel microprocessor chip that runs virtually all the IBM PC software, but is more than twice as fast as the chip used in the PC and the Compaq portable. Compaq's research revealed that complicated spreadsheets took a long time to run on the older machines and that users wanted more power. Canion says discussions with dealers helped Compaq build other features into the Deskpro that are valuable mainly to retailers, which he is not above doing. Three months before the Deskpro was released, dealers complained that their stockrooms would have trouble accommodating four new models. Compaq redesigned the line. Though it was priced from $2,500 to $7,200, the same Deskpro chassis can accommodate all four models. Dealers simply plop in extra disk drives or hard disks and whip up whatever the customer orders. While Compaq was gearing up to make the Deskpro, it was also beginning an ad campaign that cost $19 million last year. And it was starting to sell in Europe, through 330 outlets by year's end. In April it cut prices, anticipating the IBM move that was soon to come; with the new expenses, profits sagged. In the second quarter of last year, Compaq earned a mere $892,000 on sales of $66 million. While the company was still reeling, IBM introduced the AT. The AT is a true powerhouse, faster than the Deskpro. Significantly, it is also incompatible with some of the best-selling software that runs on the PC and Compaq's machines. The AT uses an advanced Intel microprocessor that allows several people at remote terminals to use it at the same time. As a so-called multi-user machine, it appears to reflect a strategic shift by IBM. Some dealers are clamoring for Compaq to turn out its own version of the AT. This time Canion isn't accommodating them. ''The personal computer market is absolutely not going to multi-user machines,'' he says. ''The basic reason for the personal computer's success is that it gives you flexibility you can't get in multi-user environments. With multi-user machines you have to share computing power and data storage capacity. If any one of the users runs software with bugs in it, that will shut down the entire machine.'' Canion may be right again. Winger at CompuShop agrees ''100 percent'' with him about multi-user machines. ''In the old days you went multi-user to share processors or expensive peripherals like data storage disks and printers,'' Winger says. ''All those are so absurdly cheap now that there's no technological or economic reason for multi-user machines.'' The AT is priced from $4,000 to $6,000, without extra terminals, depending on the amount of data storage and memory. Stephen Cohen, an analyst with the Gartner Group, a Connecticut IBM-watching firm, predicts IBM will introduce a single-user machine at midyear, running on the AT's microchip, and that it will be a stunner--he envisions IBM selling a million of them before the year is out. Canion doesn't rule out copying IBM's single-user AT, though he refuses to be stampeded. Meanwhile, IBM has run into problems getting enough disk storage devices for the AT and has had to curtail production. This stroke of luck has helped sales of the Deskpro, many customers' second choice. On the portable front, Compaq spent nine months trying to develop a small, lightweight ''laptop'' computer last year. In the end, Canion killed the idea. ''Most computer companies decided on the size and weight of a laptop and then tried to see what they could fit inside,'' he says. But Compaq's research showed that it was more important for the computer to perform a lot of functions than to be tiny or featherweight. Canion concluded nobody could come up with all the necessary functions in a machine priced low enough to find a market. So far, his marketing instincts have proved correct. Laptops have poured onto the market, but none has won as many fans as its maker hoped. Compaq portables are devilishly heavy to lug around, and the truth is they are only slightly more mobile than mobile homes. Consumers want them lighter, and Canion says he will respond with products this year. At least one new product will be ''more than 10% lighter,'' promises Canion with deliberate vagueness. O THER NEW PRODUCTS seem sure to emanate from Compaq Telecommunications, which is designing ''telecomputers'' that combine the functions of computers with those of telephones, clocks, calculators, and calendars into one desktop console. The 50-employee subsidiary is headed by C. Murray Francois, 42, yet another recruit from TI. Sparky Sparks is also there, studying how the products should be sold. The group has developed an IBM-compatible machine that will carry both data and voices. The first product will be designed to work with old-fashioned analog switchboards still used by many businesses. The telecomputers will also communicate directly with each other on intra-office networks. Francois says his product will be aimed at upper-level executives who don't need a lot of spreadsheet-calculating power but want something to help manage their hectic schedules. That's a group that has avoided computers so far because they're too difficult to master, he says, so there will be an emphasis on software, some of which Compaq will write, that makes the machine easy to use. It will be able to collect data from corporate mainframes and, for those so inclined, do spreadsheets. Most important, it will function as an electronic bulletin board to help executives schedule meetings. A device to turn a Compaq computer into a telecomputer is probably also on the way. Canion insists that telecommunications products won't make much of a splash | and that in his planning he isn't counting on a big boost from them ''this year or next.'' According to International Data Corp., a Massachusetts computer research firm, only about $60 million of telecommunications work stations were shipped last year. The business is so new that Sparks says it's hard to be sure what distribution channels will be used. Compaq is likely to link up with manufacturers of PBXs, telecommunications switching systems for offices. These companies have salesmen who are comfortable selling complicated networking gear and who have the right contacts with corporate communications executives. In implementing Compaq's strategy, Canion and Rosen all but ignore the threat that IBM could abandon its ''open architecture'' policy in microcomputers. To encourage software developers, IBM has always made technical details about its personal computers public--the kind of information that helps Compaq make machines compatible. If IBM were to abandon that policy in favor of a proprietary architecture, a move that some predict, makers of compatibles would be in serious trouble. This would be especially true if IBM were to use proprietary microprocessors it produces instead of buying the off- the-shelf variety. Even if compatible makers like Compaq could figure a way to copy IBM, it might take so long that it would be pointless in a business that moves as fast as this one. IBM presumably wouldn't shift to a closed architecture unless it was sure it had the software to support its proprietary machines. The Gartner Group believes IBM intends to have a big business selling its own software. ''IBM would like a situation where, to quote the cliche, it sold the razors and rented the blades,'' a recent Gartner report said, ''and software products are its most promising blades.'' The group calculates that practically all of the $250-million increase in IBM's programming budget last year was spent on personal computer software. Canion and Rosen don't think IBM will close off the architecture, but even if it does, they believe the new machines will be compatible with the old ones and therefore with Compaq's. ''There are two million people out there that have bought IBM personal computers,'' says Rosen. ''The company can't just abandon those machines.'' Yet with the AT, IBM has a line that is only partly compatible, which is making the tea leaves harder to read. ''The confusion they have caused with that machine is classic,'' says Canion. C OMPAQ has more immediate problems. Expenses in 1985 will be high: for tooling and advertising of new products and for manufacturing facilities. And, of course, there is the prospect of more price cuts--and even signs of a post- Christmas industrywide glut. Though Compaq has been profitable since the third quarter of 1983, its profits, like its products, aren't razzle-dazzle. Last year it earned an estimated $11.4 million, a decent 11.6% return on $98.3 million of shareholders' equity. But Compaq has an estimated $58 million of bank debt at work as well, at rates that will rise if the prime does. Though the stock has doubled since October, it is still 36% below the public offering price, so it would have to climb a lot before Compaq could comfortably raise money by issuing new shares. To finance its growth, Compaq will need to raise more capital somehow, preferably from profits. Pondering Compaq's profit prospects, some industry analysts damn the company with faint praise. ''Compaq is good, it won't die out overnight,'' says Douglas Cayne, another analyst at the Gartner Group. ''It will probably suffer from a lingering malaise.'' Ben Rosen hates that kind of talk. But Rod Canion tries to ignore it. He's got a job to do, and he can't afford his first big mistake.

CHART: TEXT NOT AVAILABLE

CHART: TEXT NOT AVAILABLE WHAT JUNIOR ATE IBM feared that its PCjr--announced in late 1983--would steal business from the PC itself. But Junior wasn't much of a cannibal; its design was so badly flawed hat IBM had to overhaul the machine last summer. After that, the new machine ful illed IBM's hopes and fears: as the PCjr's market share took off, the PC's dropp d. Compaq's machines, compatible with the PC, haven't suffered nearly as much. L st year was great for the Apple II line, nearly eight years old, whose newest me ber, the IIc, was introduced in April. BOX: INVESTOR'S SNAPSHOT COMPAQ COMPUTER SALES (LATEST FOUR QUARTERS) $268.6 MILLION CHANGE FROM YEAR EARLIER UP 355% NET PROFIT $11.3 MILLION CHANGE LOSS YEAR EARLIER RETURN ON COMMON STOCKHOLDERS' EQUITY 11% FIVE-YEAR AVERAGE N.A.* RECENT SHARE PRICE $7 PRICE/EARNINGS MULTIPLE 19 TOTAL RETURN TO INVESTORS (12 MONTHS TO 1/15) -48% PRINCIPAL MARKET OTC *Initial public offering 12/9/83. Explanatory notes: page 98