from "Idea Man" - memoir by Paul Allen:

    That August, a three-piece-suited contingent led by Jack Sams
    approached us about Project Chess, the code name for what would
    become IBM’s PC. After we talked them out of an 8-bit machine
    and won them over to the Intel 8086 (or as it turned out, the
    cheaper but virtually identical 8088), they wanted everything in
    our 16- bit cupboard, including FORTRAN, COBOL, and Pascal.
    Aside from BASIC, none of these products were even close to being
    ready for the 8086 platform. It would take a wild scramble to get
    them all done on IBM’s tight timetable.

    Then, in late September, Sams asked us if we could provide
    a 16-bit operating system. We referred him to Digital Research,
    which we’d heard was far along in building one. Bill called Gary
    Kildall and said, “I’m sending some people over to you, and I want
    you to be good to them, because you and I are both going to make
    a lot of money on this deal.” He didn’t mention IBM by name be
    cause the company insisted on maximum discretion and secrecy.
    We’d had to sign a nondisclosure agreement before they’d even sit
    down with us.

    As Kildall himself later acknowledged, he was off flying on
    business when the Project Chess group arrived. His wife, who was also
    his business partner, refused to sign the nondisclosure and offered
    a Digital Research document instead. That was something you did
    not do with IBM. Sams came back to us and said, “I don’t think
    we can work with those guys—it would take our legal department
    six months to clear the paperwork. Do you have any other ideas?
    Could you handle this on your own?”

    After the fact, there would be endless rumors about Microsoft’s
    dealings with Digital Research. Kildall theorized that IBM chose to
    work with us because we were willing to license an operating system
    for a flat fee, while Kildall insisted on a per-copy royalty. But
    I had a front-row seat, and this is what happened: We tried to do
    Digital Research a favor, and they blew it. They dropped the ball.
    I vividly remember how furious Bill was at what had transpired.
    He couldn’t believe that Kildall had blown this golden chance and
    placed the whole project in jeopardy.

    Bill called an emergency meeting with me and Kay Nishi. What
    could we do to resuscitate the deal? There was silence for a
    moment, and then I said, “There’s another operating system that
    might work. I don’t know how good it is, but I think I can get it
    for a reasonable price.” I told them the story of Tim Patterson and
    Seattle Computer Products, which began shipping its 8086 machine
    earlier that year but had found sparse commercial interest.
    The missing link was an operating system. Kildall had promised a
    CP/M-86 by the first of the year, but he hadn’t delivered; his
    company lacked the typical start-up’s urgency. No one knew when his
    16-bit software would make it to market.

    Tim Patterson had gotten frustrated waiting. Our BASIC-86
    was fine for writing programs, but his customers couldn’t run a
    word processor or other applications on top of it. So Tim had
    cobbled together a provisional 16-bit operating system to help his
    company sell a few computers until Kildall came through. (As Tim
    later said, “We would have been perfectly happy having somebody
    else do the operating system. If [Digital Research] had delivered
    in December of ’79, there wouldn’t be anything but CP/M in this
    world today.”) He called the program QDOS, for Quick and Dirty
    Operating System, which he’d managed to cram into 6K of code.
    Once it was mostly done, he changed the name to 86-DOS.

    [text deleted]

    Bill was less enthusiastic. He didn’t know Tim Patterson, and
    we’d be betting our deal with IBM—the most critical one we’d
    ever have— on an unknown quantity once called Quick and Dirty.
    But Bill realized that we might lose the whole contract unless we
    came up with something, and he went along.

    I called Rod Brock, who owned Seattle Computer Products, to
    work out a licensing agreement. We settled on $10,000, plus a royalty
    of $15,000 for every company that licensed the software—a
    total of $25,000 for now, as we had only one customer. The next
    day, a Microsoft delegation (Bill, Steve, and Bob O’Rear) met with
    IBM in Boca Raton and proposed that Microsoft coordinate the
    overall software development process for the PC. Five weeks later,
    the contract was signed. IBM would pay us a total of $430,000:
    $75,000 for “adaptations, testing, and consultation”; $45,000 for
    the disk operating system (DOS); and $310,000 for an array of 16-
    bit language interpreters and compilers.

    Bill and I were willing to forgo per-copy royalties if we could
    freely license the DOS software to other manufacturers, our old
    strategy for Altair BASIC. Already enmeshed in antitrust litigation,
    IBM readily bought this nonexclusive arrangement. They’d later
    be slammed for giving away the store, but few people at the time
    discerned how quickly the industry was changing. And no one, including
    us, foresaw that the IBM deal would ultimately make
    Microsoft the largest tech company of its day, or that Bill and I would
    become wealthy beyond our imagining.