Hard Drive: Bill Gates and the Making of the Microsoft Empire (21 page)

“If I was awake, I was thinking about the project,” said O’Rear, who worked throughout December without even taking Christmas or New Year’s Day off just like many of the others at Microsoft who were on the project.

Though Seattle Computer had furnished Microsoft a copy of 86-DOS back in September, when Microsoft informed the company it had a possible OEM customer for the operating system, no licensing agreement had been signed by the end of 1980. That didn’t seem to worry anyone. “We had no hesitation to let them try it out,” said Tim Paterson. Six days into the new year, Microsoft and Seattle Computer finally signed an agreement giving Microsoft a nonexclusive right to market 86-DOS. This meant Seattle Computer could continue to license its operating system to other customers. The negotiations were handled by Paul Allen and Rod Brock, the owner of Seattle Computer. Although the agreement was signed by Gates, he and Brock never met or even talked. For each sublicense of 86-DOS, Microsoft agreed to pay Seattle Computer $10,000, plus an additional $5,000 if the source code were part of the sublicense. Seattle Computer also received $10,000 for signing the agreement.

“We came to an agreement fairly easily,” said Paterson, vice-president of Seattle Computer. “We even called Digital Research to see what they sold their stuff for. We got a feel for what the prices were like.”

Of course, no one at Seattle Computer knew that Microsoft’s unnamed customer for the operating system was IBM, with revenues approaching nearly thirty billion dollars.

One important clause in the contract stated: “Nothing in this licensing agreement shall require Microsoft to identify its customer to Seattle Computer Products.”

Recalled Brock: “That seemed strange to us, but we agreed to go along.”

Microsoft ended up paying Seattle Computer a total of $25,000 under terms of the agreement, because it turned around and sublicensed 86-DOS and the source code to only one customer, IBM.

A source at Microsoft who was privy to the negotiations with IBM for the operating system said Microsoft licensed the first version of DOS to Big Blue for only $15,000. Microsoft also received royalties as part of the license, although the royalty arrangement has always been a closely guarded secret at IBM and Microsoft. “We were an aggressive company,” the Microsoft employee said. “Our strategy was, we would make our money on the languages. Remember, we already had the deal for all the languages, and Digital Research was supposed to have the deal for CP/M. And when it looked like we might lose the language deal because IBM didn’t have an operating system, we simply were going to solve the problem. And we solved that problem for about $15,000. But I can assure you it cost us more than that to make the delivery. And I can also assure you we made money on the BASIC. We made money on the licenses we already had, and we made sure that we got the operating system deal. Not so much to make money, and not so much to set the world standard, but simply because we couldn’t close our language deals without it.”

For a while, O’Rear wondered if he would ever get 86-DOS running on the hardware he was using in the small, stuffy, win- dowless room at Microsoft. The technical problems with both the software and hardware seemed endless. The prototype machines delivered to Microsoft just after Thanksgiving were just that—rough drafts. They didn’t work very well. O’Rear continuously fired off memos to his contacts in Boca Raton about equipment problems, complaining about the difficulty of meeting the January 12 date for the delivery of DOS and the BASIC due to breakdowns in the hardware IBM had provided.

Microsoft didn’t make the January 12 deadline. It was not until February that O’Rear finally got 86-DOS to run on the prototype. He still vividly remembers the moment. “It was like the middle of the night. It was one of the most joyous moments of my life, to finally after all the preparation and work, and back and forth, to have that operating system boot up and tell you that it’s ready to accept a command. That was an exciting moment.”

The IBM team in Boca Raton wanted several changes in the operating system, and Allen asked Paterson to help with these. The changes were all fairly minor. For example, IBM wanted one feature that remains on DOS today—the date and time stamp. Another change involved the so-called prompt that DOS left on the screen when the system came up. The prompt that Paterson had designed for 86-DOS was the drive letter followed by a colon. Neat and simple. But IBM wanted the CP/M prompt, which was the drive letter followed by a colon followed by the “greater than” sign used in mathematics.

“It made me want to throw up,” said Paterson of the request for CP/M prompts. But he obliged. Paterson was working blind as he made the requested changes in his operating system. He did not have a prototype computer. He did not even know one existed.

Seattle Computer picked up an occasional hint that Microsoft’s unnamed customer might be IBM. One day in early spring of 1981, Brock received a call from someone who said he was with IBM, and he had a question about the operating system. Brock knew his company had not licensed 86-DOS to IBM, so he asked the caller where he was located. The caller immediately hung up. Brock later mentioned to a sales rep who called on Seattle Computer from time to time that Microsoft was dealing with an OEM who did not want to be identified. The computer sales rep told Brock he had heard a similar story from a friend at Intel. The chip maker also had an agreement with an OEM who wanted to remain anonymous.

Given the scope of the PC project and the number of people at IBM who were either working on the computer or knew about it, word probably
should
have leaked out about what was going on at Microsoft and at the Entry Level Systems facility in Boca Raton. But other than rumors, specific details didn’t get out, at least not until near the end.

The original group of 13 engineers assigned to Project Chess grew to several hundred. Programmers at Microsoft joked that this was IBM’s smallest project and Microsoft’s biggest, yet IBM had more people writing requirements for the computer than Microsoft had writing code. O’Rear often felt overwhelmed by the number of people he had to deal with in Boca Raton. He had the authority to say “yes” to anything requested of him by the IBM team. But only Gates could say “no.”

“If they said we have to have this or we have to have that,

I could immediately give approval,” O’Rear said. “Otherwise, it had to go through Bill. I was dealing with a lot of people. IBM had this mainframe program where they outlined every little point and conversation that we would have, and I would get calls from people doing all different kinds of things, from contract administration to looking at the technical details to deciding where we were in the schedule to writing documentation. All these calls would be from different people. And I m trying to write all this stuff and make it work on the PC. Just to take two or three or four phone calls a day, and do all this stuff, and follow up on several of these phone calls, looking into each little aspect, or trying to educate them towards the project, it was a lot of work, it was a lot of stress.”

Everyone involved in the project at Microsoft was under incredible strain. One by one, deadlines for various stages of the project slipped by because of technical problems. Most programmers writing code for the PC worked seven-day weeks, frequently pulling all-nighters. Gates did not leave the office for days at a time, unless he had business in Boca Raton to talk about the design of the computer with Estridge or others.

Although he didn’t write much code himself, Gates reviewed most of the software code being written by his programmers for the PC. And he also helped with technical problems, as did Allen. Both suggested or ordered changes in the code when they found something they didn’t like or thought could be improved.

Bradley, the IBM software engineer, recalled a trip he made to Microsoft to bring a new power supply for one of the broken- down prototypes. The problem was fixed on a Saturday, but Bradley was told by his office to remain in Bellevue until Monday so he could pick up a new version of BASIC that Microsoft was supposed to have finished by then. But on Monday, Gates told Bradley it was not yet ready and to come back Tuesday. Around five the next morning, Bradley got off the elevator at Microsoft’s offices, walked down the hallway, and found Gates sprawled on the floor of a back office, going over a huge computer printout with a red pen, marking changes he wanted. He had been up all night debugging the BASIC.

Although the pace of work was unrelenting during the spring of 1981, Allen and a couple programmers took a quick break and flew to Florida in mid-April for the maiden launch of the Space Shuttle. “This was the first mission, and it was a big deal,” said Charles Simonyi, who had to talk an exhausted Allen into making the trip. But they almost didn’t get to go. Gates had scheduled a company meeting for Friday, April 11, the day Columbia was supposed to blast off. But a computer software glitch at NASA postponed the flight until Sunday, so Simonyi, Allen, and Marc McDonald flew down on Saturday. They rented a car in Miami and drove all night up the Florida coast to Cape Kennedy. The Columbia, carrying astronauts John Young and Robert Crippen, blasted into orbit at seven o’clock in the morning Eastern Daylight Time. The three space enthusiasts from Microsoft then drove back to Miami and flew home.

Simonyi was not working on Project Chess. He had recently been hired to take over development of Microsoft s applications. Although much of the company’s attention was focused on the IBM project, Microsoft could not neglect its other business. Deals with OEM customers in this country and in Japan continued. Programmers like Simonyi worked on various applications. In dealing with these other customers, Microsoft took advantage of its inside knowledge that IBM was going to introduce its own personal computer based on Intel’s 8088 chip, according to a manager who was working for Microsoft at that time. “We would highly advise some of our customers to
choose
the 16-bit processor,” he said. “Nobody really knew that we were working on the IBM contract.”

On May 1, Tim Paterson went to work for Microsoft, where he learned for the first time who the customer for his operating system was. He had asked Allen about a job a few weeks earlier. Paterson decided to leave Seattle Computer because Brock could not make up his mind whether to sell the company’s products by mail order or through dealers. Brock was thinking of going back to a mail-order business, and Paterson did not want to work for what he figured would soon be a mom-and-pop operation. At Microsoft, Paterson joined O’Rear on the operating system. By the end of June, DOS was pretty much finished.

The company was growing rapidly, in part because of so many new employees hired to help with the IBM project. By June, the number of Microsoft employees had more than doubled from the previous year, to about 70.

One programmer hired in June, Richard Leeds, thought he was joining Microsoft to work on something else until he came to work the first morning, signed the nondisclosure agreement, and was told he would be helping with Project Chess. Each Microsoft employee on the project had to sign the document. Leeds was surprised when he got his first look at the PC. It had a clear plastic keyboard, and he could see right through the keys into the workings underneath. “We called IBM the typewriter company, he said. But the joke was, here was a typewriter company that couldn’t come up with a usable keyboard.” Leeds was made project manager for COBOL, one of the languages that Microsoft was supposed to deliver to IBM. His job was to convert Microsoft’s 8-bit version of COBOL to a version that could run on the 16-bit chip.

“It was very hectic around there,” Leeds recalled. “Everybody was real driven, real proud to be kids working on the next machine from IBM. ... It was nonstop work. I was working 65- plus hours a week and I had a girlfriend, and there were complaints that I wasn’t working hard enough. They wanted 80. There
were
times that I worked 80.”

When the COBOL work was done, Leeds received only a fourteen percent bonus instead of the promised fifteen percent.

Leeds had a habit of collecting pens that had been chewed obsessively by Gates. In only a couple of months, its grew to be a very large collection. Gates worried incessantly about the project. He knew that IBM had a habit of spending vast amounts of money on research projects that never saw the light of day. IBM would conclude a product could not be marketed, and then put the lid on it, buried forever in its giant bureaucracy. Up until the final days before the PC was announced, Gates was haunted by the thought that IBM would cancel the project. Had his nightmare scenario come true, Microsoft would have been hurt significantly because Gates had thrown so much of the company’s resources into the project.

“There was always the fear that IBM would decide not to announce this, that someone would say, ‘Nice effort guys, but we don t want to go into this personal computer business,’ ” said O’Rear.

Gates got a particularly bad case of the jitters when the industry journal
InfoWorld,
in its June 8 issue, reported in alarming detail about the top secret PC project in Boca Raton. Gates worried that such stories would blow IBM’s cover and cause it to abandon the effort.

The
InfoWorld
article was entitled “IBM to Pounce on Micro Market.” It was datelined Boca Raton. “A reliable source within IBM’s Entry Level Systems group in Boca Raton has provided
InfoWorld
with exclusive details on IBM’s new personal computer,” the article said. “The system is scheduled to be announced in New York in mid July 1981. The central processor for this new system will be the 16-bit Intel 8088.”

It went on to describe the computer’s memory size, monitor, and keyboard. The article even talked about the operating system. “IBM gave some thought to using CP/M as the disk operating system for the personal computer, but this would have been an incredible departure from normal IBM product development strategies,”
InfoWorld
reported. “Instead, the operating system for this new computer will be similar to CP/M in many respects. The designers didn’t strive for compatibility, just similarity.”