Kildall briefly returned to UW and finished his doctorate in computer science in 1972, then resumed teaching at NPS. He published a paper that introduced the theory of data-flow analysis used today in optimizing compilers, and he continued to experiment with microcomputers and the emerging technology of floppy disks. Intel lent him systems using the 8008 and 8080 processors, and in 1973 he developed the first high-level programming language for microprocessors, called PL/M. He created CP/M the same year to enable the 8080 to control a floppy drive, combining for the first time all the essential components of a computer at the microcomputer scale. He demonstrated CP/M to Intel, but Intel had little interest and chose to market PL/M instead.
CP/M's quick success took Kildall by surprise, and he was slow to update it for high density floppy disks and hard disks. After hardware manufacturers talked about creating a rival operating system, Kildall started a rush project to develop CP/M 2. By 1981, at the peak of its popularity, CP/M ran on 3000 different computer models and DRI had $5.4 million in yearly revenues.
Various reasons have been given for the two companies failing to reach an agreement. DRI, which had only a few products, might have been unwilling to license its main product to IBM for a one-time payment rather than its usual royalty-based plan. Dorothy might have believed that the company could not deliver CP/M-86 on IBM's proposed schedule, as the company was busy developing an implementation of the PL/I programming language for Data General. Or, the IBM representatives might have been annoyed that DRI had spent hours on what they considered a routine formality. According to Kildall, the IBM representatives took the same flight to Florida that night that he and Dorothy took for their vacation, and they negotiated further on the flight, reaching a handshake agreement. IBM lead negotiator Jack Sams insisted that he never met Gary, and an IBM colleague recalled that he said so at the time. He accepted that someone else in his group might have been on the same flight, but noted that he flew back to Seattle to talk with Microsoft again.
Sams related the story to Gates, who had already agreed to provide a BASIC interpreter and several other programs for the PC. Gates's impression of the story was that Gary capriciously "went flying," as he would later tell reporters. Sams left Gates with the task of finding a usable operating system, and a few weeks later he proposed using the CP/M clone 86-DOS from Seattle Computer Products (SCP). Paul Allen negotiated a licensing deal with SCP, had 86-DOS adapted for IBM's hardware, and IBM shipped it as PC-DOS.
Kildall obtained a copy of PC-DOS, examined it, and concluded that it infringed on CP/M. When he asked Gerry Davis what legal options were available, Davis told him that intellectual property law for software was not clear enough to sue. (According to Davis, the law later developed in what might have been DRI's favor.) Instead Kildall only threatened IBM with legal action, and IBM responded with a proposal to offer CP/M-86 as an option for the PC in return for a release of liability. Kildall accepted, believing that IBM's new system (like its previous personal computers) would not be a significant commercial success. When the IBM PC was introduced, IBM sold its operating system as an unbundled (but necessary) option. One of the operating system options was PC-DOS, priced at US$40. CP/M-86 shipped a few months later at $240, but sold poorly against DOS.
With the loss of the IBM deal, Gary and Dorothy found themselves under pressure to bring in more experienced management, and Gary's influence over the company waned. He worked in various experimental and research projects, such as a version of CP/M with multitasking and an implementation of the Logo programming language. He hoped that Logo, an educational dialect of LISP, would supplant BASIC in education, but it did not. After seeing a demonstration of the Apple Lisa, Kildall oversaw the creation of DRI's own graphical user interface, called Graphical Environment Manager (GEM) Desktop. Novell acquired DRI in 1991 in a deal that netted millions for Kildall.
Kildall also pursued computing-related projects outside DRI. In 1983 he started hosting a public television program on the side, called Computer Chronicles, that followed trends in personal computing. He started another company, KnowledgeSet, that adapted optical disk technology for computer use. In 1985 it released the first computer encyclopedia, Grolier's Academic American Encyclopedia. Kildall's final business venture, known as Prometheus Light and Sound and based in Austin, Texas, developed a home PBX system that integrated land-line telephones with mobile phones.
Although Kildall preferred to leave the IBM affair in the past and to be known for his work before and afterward, he continually faced comparisons between himself and Bill Gates as well as fading memories of his contributions. A legend grew around the fateful IBM-DRI meeting (encouraged by Gates and various journalists), suggesting that Kildall had irresponsibly taken the day off for a recreational flight, and he tired of constantly having to refute that story. In later years, he had occasional private expressions of bitterness over being upstaged by Microsoft.
Kildall was particularly annoyed when the University of Washington asked him, as a distinguished graduate, to attend their computer science program anniversary in 1992, but gave the keynote speech to college dropout Gates. In response he started writing his memoir, Computer Connections. The memoir, which he distributed only to a few friends, expressed his frustration that people did not seem to value elegance in software, and it said of Gates, "He is divisive. He is manipulative. He is a user. He has taken much from me and the industry." In an appendix he called DOS "plain and simple theft" because its first 26 system calls worked the same as CP/M's. Harold Evans used the memoir as a source for a chapter about Kildall in the 2004 book They Made America, concluding that Microsoft had robbed Kildall of his inventions. IBM veterans from the PC project disputed the book's description of events, and Microsoft described it as "one-sided and inaccurate."
Selling DRI to Novell had made Kildall a wealthy man, and he moved to the West Lake Hills suburb of Austin. His Austin house was a lakeside property, with stalls for several sports cars, plus a video studio in the basement. Kildall owned and flew his own Lear jet and had at least one boat on the lake. While in Austin he also participated in volunteer efforts to assist children with AIDS. In California, he owned a mansion with a panoramic ocean view in Pebble Beach.
At the time of Kildall's death, Bill Gates commented that he was "one of the original pioneers of the PC revolution" and "a very creative computer scientist who did excellent work."