Hard Drive: Bill Gates and the Making of the Microsoft Empire (18 page)

In time, Ballmer would be called everything from Gates’ alter ego and Microsoft’s No. 1 cheerleader, to the world’s ultimate marketing expert. Regardless of the accuracy of those descriptions, his importance to the company—and to Gates— cannot be overstated.

“If you are going to write about Bill,” said one of Microsoft’s most senior executives, “you must give substantial attention to Steve Ballmer. He is a much more important player than he’s chosen to let people know. He’s so bright, so intense. He’s much more than just a lieutenant for Bill. There’s such a high level of trust and high bandwidth communication between the two. They trust in each other’s IQ. They are equivalently intense.”

It took time for Ballmer to find his place at Microsoft. Until then, he did just what his job title said, assist the president. And he was well prepared to serve. During his senior year at Harvard, Ballmer was head equipment manager for the football team.

“Steve would talk about how important that experience with the football team had been,” recalled that same senior Microsoft executive. “I think he saw himself as not one of the players, but someone who loved making sure the bucket of water was there so the team could win. Bill was going to be the quarterback, and Steve was going to be the guy who would do any damn thing it took to make this work.”

Ballmer did not have long to wait after joining Microsoft before he made himself very useful to Gates. In July of 1980, Gates was busy helping develop a BASIC for Atari. The video game company was entering the personal computer sweepstakes, and Gates had met several times already with chairman Ray Kassar.

One morning in late July, while Gates was preparing for a meeting with Kassar, he took a call from a man who identified himself as Jack Sams of IBM. Sams told Gates he wanted to fly out to Seattle for a talk. He was not specific about a time. Gates looked at his desk calendar. “What about next week? he asked. Replied Sams, “What about tomorrow?” When Gates got off the phone with Sams, he called Kassar in California and canceled their meeting for the next day. Gates then walked in to Ballmer’s office. IBM was coming, he told his friend. Better get out the suits, said Ballmer.

They were the corporate odd couple, on opposite sides of the country. The thought of free-spirited Microsoft climbing into bed with a stuffed-shirt outfit like IBM was about as incongruous as the fact that the greatest computer company in the world was shut out of the desk top computer market in 1980.

In the five years since the Altair had shined briefly in the night sky above Albuquerque, guiding the way for young entrepreneurs with genius and vision, nearly 200 different brands of microcomputers had been brought to the high-tech market place, and those with the IBM nameplate had been quietly withdrawn out of embarrassment. Thomas J. Watson, Sr., would have rolled over in his grave had he known that in July of 1980 high- level executives in the company were suggesting IBM go out and buy a personal computer from someone like Atari and stick Big Blue’s name on it.

It was 1924 when Watson changed the name of Computing- Tabulating-Recording Co. to International Business Machines and created an accounting machine monopoly. Since then, the company had enjoyed a long history of success. In the 1950s, even though Sperry-Rand’s UNIVAC I was the first commercial computer, IBM quickly captured more than ninety percent of the market with its punch-card “Giant Brain” machines. Its 360 series mainframe computer that debuted in 1964 took seventy percent of the market and set a standard that is still a model today. The initials IBM became synonymous with computers. IBM was one of the world’s most successful and admired companies, the bluest of the blue chips, its stock a favorite of Wall Street. But the giant corporation was also straight
f
aced and stuffy, a bureaucratic institution that tended to stifle individuality, smother creativity, and paralyze action. For years, it had an unwritten dress code that executives wore only white shirts and conservative blue suits, hence the nickname Big Blue. Employees posted “THINK” signs in their offices and sang a company song, “Hail to IBM,” which ended with a promise to “toast the name that lives forever”—IBM.

But the federal government didn’t want IBM to live forever, at least in its present form. In January 1969, the Johnson administration, on its last working day in office, filed suit accusing IBM of monopolizing the U.S. computer industry. Three administrations would subsequently argue that IBM’s market dominance should be diminished, that Big Blue be dismembered into Baby Blues, much the same way that Ma Bell was later chopped up into so many Baby Bells. The very life of the company was at stake. When the Reagan administration finally dropped the antitrust suit in January of 1982, the court proceedings had run over 200,000 pages and an entire law firm had been created to try to save IBM.

The suit hung over the company’s head like the Sword of Damocles in 1975 when Paul Allen fed BASIC into the Altair and the machine typed out “Ready,” signaling the start of the microcomputer revolution. In part because of an internal circle- the-wagons mentality that resulted from the suit, IBM had been steadily losing ground in the computer market. By 1980, its market share was down to about forty percent, and that came entirely from sales of mainframes and minicomputers. IBM had nearly drowned in its first attempt to test the waters of the small computer market with its 5100 series in the late 1970s. The machines were taken off the market.

The company’s efforts to produce a low-end commercial product were centered at its plant in Boca Raton, Florida, known

as the Entry Level Systems unit. There, several projects were underway. An engineering team headed by Bill Sydnes was working on the System 23 Datamaster, which was a follow-up to the ill-fated 5100 series. The System 23 machine used a mix of IBM and non-IBM parts. A competing group in Boca was considering a small computer sourced
entirely
from outside IBM—in other words, buying a machine from another company.

Most of those working in Boca Raton did not believe it possible for IBM to develop its own successful personal computer, given the existing structure and culture of the company. The personal computer had been created out of an entrepreneurial spirit that didn’t exist at IBM. The personal computers’ designers were hackers and hobbyists, young radicals with long hair who wore faded jeans and listened to loud rock n roll music. The birthplace of the leading personal computer of the day, the Apple, was a garage, not a huge corporation that could not overcome its own bureaucratic inertia.

This was the message, anyway, that Bill Lowe delivered in early July of 1980 to his bosses at IBM corporate headquarters in Armonk, New York, about an hour’s drive north of New York City in the Hudson Valley. In his meeting with the Corporate Management Committee (CMC), Lowe proposed that IBM go out and buy a computer from Atari.

Lowe, director of the Boca Raton lab, was told this was blasphemy, “the dumbest thing we’ve ever heard of.” He was sent back to Florida with orders to do what ever had to be done to develop IBM’s
own
personal computer. He was to assemble a task force and bring back a prototype of IBM’s “Apple” in 30 days!

In the book
Blue Magic,
a fascinating, inside look at the development of the IBM PC, authors James Chposky and Ted Leonsis said Lowe had carefully planned this strategy, knowing the CMC would never accept his proposal and that they would tell him to come up with the company’s own product, free of the current corporate structure.

Lowe went back to Boca Raton and picked his team, 13 engineers not cut from the traditional blue cloth of IBM—a baker’s dozen of free spirits.

“If you’re competing against people who started in a garage, you have to start in a garage,” said Don Estridge, the man Lowe eventually selected as the day-to-day project manager and team leader. Of all those involved in the PC project, Estridge would become the most visible and the best known.

The task force agreed the new computer should be an “open architecture” system. In other words, critical components of the machine, such as the microprocessor, would come from existing technology in the market place and would not be proprietary, like components in the Apple. Industry guru Adam Osborne said he was one of two outside consultants advising IBM to make their architecture open. The other was Portia Isaacson. “I didn’t realize at the time, said Osborne, “that both of us were telling IBM, ‘Don’t build your own hardware and software.’ ”

This radical break from Big Blue’s tradition would turn out to be the key decision that made the IBM PC an industry standard. But it was done out of necessity, to save time.

It was decided that the machine’s software, including the vital operating system, would also come from an outside vendor. Responsibility for securing the software fell to task-force member Jack Sams, and in late July of 1980 he placed a call to Microsoft out in Bellevue, Washington, and asked to speak with Bill Gates.

Sams was not altogether unfamiliar with Microsoft. A veteran of the IBM development team that produced its famous 360 series of mainframe computers, Sams was a software and operating system expert. Most recently, he had been involved with Bill Sydnes on the System 23 project in Boca Raton. In fact, Sams had recommended to Lowe that IBM contract with Microsoft to develop a BASIC for that project, but his recommendation was rejected. IBM came up with its own software for the System 23 machine. “It just took longer and cost us more,” said Sams.

In July, when Sams phoned Gates, the task force had less than a month remaining to come up with a prototype machine and gather information that Lowe would need in making his report to the CMC in August. The corporate brass had not yet approved the TO project. Sams’ mission in visiting Microsoft was merely to look the company over and report back to the task force. Gates would not be told of IBM’s plans. He would only be asked very general questions. But he would have to sign IBM’s famous nondisclosure agreement. The project was hush- hush even within IBM, and anything discussed at the meeting had to remain confidential.

“In that first meeting, we were just prospecting. We wanted to determine whether they had the capability to deliver on a schedule we knew we would have to meet,” said Sams. “If we had gotten out there and found out they were only three guys, we’d have to take a hard look at whether they would be able to do the work or not.”

Sams made the trip to Microsoft with two others from IBM, an executive from purchasing and someone from corporate relations. They were along more or less as “witnesses,” Sams said. IBM was paranoid about security. When they got off the elevator on the eighth floor of the Old National Bank building in Bellevue, the receptionist notified Gates, who was in his office.

“I knew Bill was young, but I had never seen him before,” said Sams. “When someone came out to take us back to his office, I thought the guy who came out was the office boy. It was Bill. Well, I’ll tell you or anybody else, and I told IBM executives this the next week, that by the time you were with Bill for fifteen minutes, you no longer thought about how old he was or what he looked like. He had the most brilliant mind that I had ever dealt with.”

Gates would later say that the IBM team asked him “a lot of crazy questions” at that first meeting. The next day, Steve Ballmer typed up a letter for Gates to sign, thanking IBM for the visit.

When Sams got back to Boca Raton, he gave a favorable report about Microsoft and its young president. “I recommended that we base our plan on using their firm,” Sams said. “We left after the first meeting feeling they could respond to what we wanted them to do. I felt their answers were open, and I thought they were correct.”

A few weeks later, in the dog days of August, Bill Lowe flew back to Armonk. With him were Bill Sydnes, Lewis Eg- gebrecht, and the prototype machine that would become known as the IBM PC. Sydnes was engineering manager for the project. Eggebrecht, an engineer, had worked with him on the prototype. After seeing a demonstration of the machine and hearing the report from Lowe, the CMC gave the go ahead. The project would be code named “Chess,” and Lowe would have exactly one year to get the new machine to market.

The engineering team was to be cloistered in Boca Raton and would work as an Independent Business Unit, or IBU. Former IBM Chairman Frank Cary, in an oft-quoted remark, said the IBU was “IBM’s answer to the question, ‘How do you make an elephant tap dance?’ ”

This was an outfit of technological Green Berets, a Dirty Dozen who would soon grow into hundreds.

A second meeting between Microsoft and IBM was held in late August, not long after Project Chess had received corporate blessing from headquarters in Armonk. Sams told Gates on the phone that he was bringing along four others from IBM, including an attorney. When the five “suits” arrived at Microsoft, they were met by an equal number of representatives from Microsoft. Gates was joined by Allen, Ballmer, Nishi, and an attorney from his father’s law firm.

Once again, Gates and his team were asked to sign the nondisclosure agreement, and after this formality they were then

told what they already suspected: IBM had a top secret project underway to develop a personal computer.

Numerous published accounts over the years have said that it was Gates who, after hearing about the project, told the IBM group their microcomputer should be built around the 16-bit Intel chip and not the old 8-bit 8080 chip. Gates knew the more advanced chip would allow Microsoft to write much more powerful software for the new computer. The preface to Microsoft’s own MS-DOS Encyclopedia states that Gates and Microsoft convinced IBM to base its machine on the newer and faster chip. “IBM was, however, unsure of microcomputing technology and the microcomputing market . . . the preface says. “One of IBM’s solutions—the one outlined by Sams’ group—was to base the new machine on products from other manufacturers. All the necessary hardware was available, but the same could not be said of the software. Hence the visit to Microsoft with the question: Given the specifications for an 8-bit computer, could Microsoft write a ROM BASIC for [IBM] by the following April? Microsoft responded positively, but added questions of its own: Why introduce an 8-bit computer? Why not release a 16-bit machine based on Intel’s 8086 chip instead? At the end of this meeting—the first of many—Sams and his group returned to Boca Raton with a proposal for the development of a low-end, 16-bit business workstation. The venture was named Project Chess.”

Microsoft’s official version of history is exaggerated, however, according to key members of the Project Chess task force. All affirmed that IBM engineers in Boca Raton had decided to use 16-bit architecture long before Gates was ever told about the machine. Bill Sydnes, engineering manager for Project Chess and the first man to be hand-picked by Lowe for the 13- member task force, had this to say about Microsoft’s claim:

“That’s a crock of bull. Absolutely a crock of bull. We had already done the System 23 with an 8-bit architecture and the hardware was completely done and ready to ship and the operating system was about a year or year and a half away when we started the PC program. So I moved over to do that and we had firmly settled on the fact that it had to be a 16-bit architecture because of where we thought the machine was going to go over time.” .

Several chips were considered for the IBM personal computer, according to Sydnes, including Motorola’s 68000. But this 32-bit chip had some technical flaws. “Basically, Motorola wasn’t ready,” Sydnes said. “They were still six to nine months off from where we needed them to be.” The task force eventually decided to use Intel’s 8088 chip rather than the 8086. The 8088 was a 16-bit chip with some characteristics of 8-bit technology. The main reason for going with this chip was that it was not as fast as the 8086, which Sydnes said had a little too much horsepower for the machine they envisioned. By using the 8088 chip, engineers could easily upgrade to the 8086 in time.

Sams confirmed that the Boca Raton group had already decided to use a 16-bit chip before Gates was contacted. But he said Gates was not told this at the second meeting, because of the IBM secrecy protocol, so Gates may well have believed he was making a recommendation that IBM acted on.

“We had selected the 8088 for reasons of our own, but we did not tell that to Bill,” Sams said. “We essentially asked for his recommendations and among his recommendations was that we use the 16-bit architecture. I’m sure he believes he suggested it to us. But we clearly never intended to use the 8-bit CPU again.”

Gates also recommended that the IBM computer have color capability, Sams said, and he made several suggestions about the keyboard.

IBM wanted Microsoft to supply software development tools such as BASIC for its personal computer, and Gates agreed to do so when he met with Sams for the second time in late August of 1980. Don Estridge, the leader of the Project Chess team, would later explain to
Byte
magazine why IBM did not go with its own BASIC: “IBM has an excellent BASIC—it’s well received, runs fast on mainframe computers and it’s a lot more functional than microcomputer BASICs were in 1980. But the number of users was infinitesimal compared to the number of Microsoft BASIC users. Microsoft BASIC had hundreds of thousands of users around the world. How are you going to argue with that?”

As far back as Albuquerque, Microsoft’s slogan had been “We Set the Standard.” That vision and commitment was now about to pay off.

Since there was not enough time to develop its own software for Project Chess, IBM not only needed to buy BASIC, but more importantly a microcomputer operating system from an outside source. And the most popular one at the time was CP/M, the operating system developed by Gary Kildall of Digital Research. Sams said he was under the assumption that Microsoft had the rights to the CP/M source code when he first contacted Gates in July. Microsoft had just released the SoftCard, which allowed products written for CP/M to run on the Apple II. “Our target market was the Apple software,” said Sams. “People were buying the Apple for business purposes and we knew the products would sell. Many of these products were based on CP/M, so it was clear to us that we needed a CP/M or compatible machine. Our proposal, the one we originally talked to Microsoft about, was to use the source code for the SoftCard they were then marketing for the Apple. ... I presumed they were able to offer us a 16-bit version of the operating system. But we really didn’t discover until the second or third meeting that was not true, because we weren’t able at the first meeting to ask the sort of detailed questions we wanted to ask.”

Microsoft not only did not have the rights to sell or license the CP/M source code to IBM, but Digital Research, which owned CP/M, had not yet developed a 16-bit version of the software. Obviously, Sams and the IBM team were a little out of touch with what was happening in this new industry.

Sams flew to Bellevue to meet with Gates several more times during September. At what would have been their third or fourth meeting since the first contact, they talked at length about the operating system for the new PC. IBM wanted Microsoft to supply it with not only BASIC but also languages such as FORTRAN, COBOL, and Pascal. Stand-alone BASIC could function without an operating system, but Microsoft’s other languages could not. IBM needed to make up its mind about what it wanted to do for an operating system, Gates said. Time was short. “Bill told us if we wanted a 16-bit CP/M, we would have to deal with Kildall,” said Sams. “We said ‘oops.’ We had really only wanted to deal with one person. But now we had to talk with Kildall. I asked Bill if he would make an appointment for us.”

Gates picked up the phone in his office and called Kildall at Digital Research. Gates told Kildall he was sending him some important customers, “so treat them right.” He then handed the phone to Sams, who made an appointment for his team to visit Kildall the next day.

Gary Kildall and Bill Gates went back a long way, to Computer Center Corporation when Gates was one of the bunch of crazy kids from Lakeside hired to try to crash the C-Cubed computer.

Little did Gates know how often in the coming years his path would cross that of Kildall, who had occasionally paid nighttime visits to the young programmers at C-Cubed, or how profound the consequences would be. A Seattle native whose father owned a navigation school, Kildall had enrolled as an undergraduate at the University of Washington with thoughts of becoming a high school math teacher until a friend showed him a computer program written in FORTRAN. He was immediately hooked. He later designed a program to help his father calculatetide tables, a job previously done by hand for a local publishing company.

After getting his graduate degree in computer science in 1972, Kildall moved to the California coastal town of Pacific Grove on the Monterey Peninsula, where he taught computer science at the nearby U.S. Naval Postgraduate School. Kildall bought one of Intel’s first 4004 microprocessor chips and wrote a simple programming language for it, which landed him a one- day-a-week consulting job at Intel up the coast in the Silicon Valley. When Intel released its 8-bit 8008 chip, the same microprocessor that Gates and Allen were working with in Seattle to build their Traf-O-Data machine, Kildall developed a programming language for it, too. In return for his consulting work, Intel gave Kildall a small computer that he set up in his classroom at the Naval Postgraduate School. Kildall, who did not have the necessary hardware expertise, brought in John Torode, a friend from the University of Washington, to develop a disk drive for the computer, which had since been upgraded with Intel’s 8080 chip. Kildall, meanwhile, worked on a simple operating system called Control Program for Micros to store information on To- rode’s computer disk. By 1974, they had a crude microcomputer with an operating system, and they sold two of the machines to a computer company in the Bay Area. This was several months before the Altair appeared on the cover of
Popular Electronics.
In time, Kildall began selling his software and the CP/M operating system to computer makers through Intergalactic Digital Research, a company he had formed with his wife, Dorothy McEwen. They quickly shortened the name to Digital Research. In 1977 Seymour Rubinstein, marketing director of IMSAI Manufacturing in San Leandro, California, paid Kildall $25,000 for the right to run CP/M on IMSAI 8080 microcomputers. IMSAI had been started soon after the Altair debuted in 1975, and quickly seized the lead in microcomputer sales.

Meanwhile in Albuquerque, Gates was doing everything he could to help make CP/M an industry standard.

For a language company like Microsoft, supporting different microcomputer operating systems was a business nightmare. Each operating system had its own way of doing things, such as managing memory and file systems. If Microsoft’s programmers could write software for the same operating system on each computer, all they wpuld have to do was modify their code slightly for whatever specific devices a customer’s computer might have.

“When we were talking to another OEM, a hardware customer who wanted to run BASIC or any of our products, we got to a point by 1977 or ’78 where we were always trying to get them to go to Digital first and get CP/M running because it made our job a whole lot easier,” recalled Steve Wood. “When we were doing custom things like the General Electric version or NCR version, it got to be a real headache. It made our lives a lot easier if someone would just go license CP/M and get that up on their machines and then our stuff would pretty much run as is. And Gary would do likewise. If someone went to him to license CP/M and they were looking for languages, he would refer people to Microsoft. It was a very synergistic kind of thing.”

In fact, there was an unwritten agreement between Gates and Kildall that Microsoft would stay out of the operating system end of the business, and Kildall would not get into microcomputer languages, according to industry sources.

Not that Microsoft did not want to get into this market, too. “Early on in New Mexico we talked about developing an operating system,” said Wood. “We asked ourselves if we should really be referring all this business to Gary. We always came back with the same answer: We have all this other stuff to do.”

At one point, probably in late 1978, Gates considered a merger with Digital, which certainly would have changed the face of the industry. Gates flew to Monterey and talked about a possible deal over dinner at the Kildall home. “It was a fairly serious discussion,” said Kildall. “At the time, Microsoft was thinking about moving but they couldn’t decide where, either back to Seattle or to the West
Coast....
I thought it was an okay idea [a merger], but we weren’t able to come to any final agreement. I don’t know how our personalities would have mixed. I got along fine with him, but we would have had to explore it
more....”

Microsoft was not the only company trying to promote CP/M as an industry standard. LifeBoat Associates, a New York software distributor that had started out as a club for CP/M users, pushed it as well, publishing a catalog of Digital’s software. Microsoft also sold its CP/M-based languages such as FORTRAN through the catalog. Later, LifeBoat Associates did CP/M implementations. OEM customers would license the operating system from Digital, then ask LifeBoat to have it ported to their computer. Microsoft worked closely with LifeBoat in trying to make CP/M a standard. “There was this vigorous effort to get all the computer makers to support that operating system, and Microsoft’s languages too,” said one Microsoft programmer. “There was intense lobbying from Bill and from Tony Cold of LifeBoat Associates. What really created the microcomputer industry more than any other thing was that all these hardware manufactures didn’t really have a clue what the industry as a whole was going to be doing, and they all got lobbied real hard to support a consistent operating system so there would be a market for applications.”

But in late 1979, the synergistic relationship between Microsoft and Digital began to unravel after Kildall packaged his operating system with a BASIC that had been developed by Gordon Eubanks, one of his students at the Naval Postgraduate School. Eubanks’ CBASIC had been on the market for about two years, and it represented the only real alternative to the BASIC offered by Microsoft.

“It wasn’t a real strong competitor but it was a competitor,” said Wood of CBASIC. “When Digital started marketing that, it upset things a little bit. What it did was add a little more fuel to our internal discussions whether we should be in the operating systems. There was a big market there. CP/M wasn’t all that sophisticated. Should we try and do something on our own?”

Kildall said there was never any promise on his part to keep out of languages. Digital Research, in fact, had developed several languages, such as BASIC and FORTRAN, which it had introduced in the business market. “There was an agreement in principal that Microsoft would do languages and we would do operating systems. But that was only because at the time, we were doing operating systems and they were doing languages. It wasn’t like we had divvied up the market place. My own personal expertise was in languages, so I certainly wasn’t going to give that up.”

CBASIC posed no market threat to Microsoft, Kildall said. Regardless of whether it was a real or perceived threat, it was at that time that Gates, in reaction, went to AT&T and licensed its UNIX operating system, which Microsoft sold at a discount under the name XENIX. “Bill was not happy,” recalled Steve Smith, Microsoft’s marketing manager who had joined the company in 1979. “Digital was now in our languages, and it couldn’t have been more than a month or two before we were introducing XENIX. It was as simple as that. When they came into our market, we went after their
market....
We knew that eventually we would be competing with all the software companies. But when they were the ones that were aggressively moving into our markets, we tended to react aggressively.”

Kildall said he did not lose any sleep when he learned Gates had licensed an operating system. “XENIX was not significant,” Kildall said. “It was not any bother to me at all. It didn’t make any difference. It was like, so what. UNIX was not in the market place we were into. It was mostly for scientific work stations, as it is now. It was never a threat to us in our business markets.”