IBM and the PC mistake
Letting Microsoft own DOS in 1980
Description
In July 1980, a small team of IBM executives flew to Bellevue, Washington, to meet with a twenty-four-year-old programmer named Bill Gates, who ran a forty-employee software company called Microsoft. IBM was, at the time, the most powerful computer company in the world. The company had revenue exceeding $26 billion, employed over 340,000 people, controlled the mainframe and minicomputer markets that defined American business computing, and had been the dominant force in the information-technology industry for over two decades. Microsoft, in the same year, had revenue under $10 million and was best known for producing programming-language tools for the small hobbyist computers that were then beginning to appear in American homes. The meeting between the two companies was meant to be brief. IBM was developing a personal computer for the consumer and small-business market and needed an operating system. The team had been instructed to license one from Microsoft and to return to Boca Raton, Florida, where the project was being developed.
The deal that emerged from the Bellevue meeting and the negotiations that followed across the summer of 1980 became, in retrospect, one of the most consequential agreements in the history of American business. IBM agreed to license MS-DOS, the operating system Microsoft would assemble for the IBM Personal Computer, on terms that allowed Microsoft to retain the rights to sell the same operating system to other manufacturers. The clause was small and, at the time, did not seem to matter. IBM was the only major company building personal computers in any quantity. Other manufacturers were small operations that nobody at IBM thought of as competition. The expectation was that the IBM Personal Computer would dominate the personal-computer market the way IBM mainframes had dominated mainframes, and that Microsoft’s operating-system rights to other manufacturers would be effectively worthless.
The expectation was wrong in ways that would unfold across the following fifteen years. Other manufacturers Compaq, Dell, Gateway, dozens of smaller producers began reverse-engineering the IBM PC design and producing compatible machines that ran MS-DOS. The compatibles undercut IBM on price. The cumulative effect, by the early 1990s, was that IBM had lost the personal-computer market it had effectively created. Microsoft, which collected a license fee on every MS-DOS installation across every IBM-compatible manufacturer, became one of the most valuable companies in the world. The license clause that nobody at IBM had thought to refuse in 1980 turned out to be the structural feature that decided where the value of the personal-computer industry would accumulate. The mistake had a price tag eventually measured in hundreds of billions of dollars.
The question we’re asking: what happened in the 1980 negotiation, why did IBM make the decisions it made, and what does the case reveal about how industry value migrates between companies?
What we’ll see: the IBM context in 1980, the Microsoft side of the negotiation, the strategic consequences across the 1980s, and what survives.
Table of contents
01A monopoly with a side project
IBM in 1980 was organized around mainframe computing. The System/360 line had been the dominant business-computing platform globally since 1964. Revenue came from long-term contracts with corporate customers who leased IBM hardware and used IBM software. The model was profitable, predictable, and protected by substantial switching costs.
The personal computer was a side project. The microprocessor-based small computers that had appeared in the late 1970s the Apple II, the Commodore PET, the Tandy machines had not been taken seriously by IBM’s senior management. The volumes were small, the revenue per machine a fraction of mainframe hardware, the customer base (hobbyists, small businesses, schools) not the kind IBM was organized to serve. The IBM PC project authorized in 1980 was a deliberate departure: unusual autonomy, a one-year timeline, and permission to source components from outside vendors.
02The license terms and why they mattered
The license negotiation across the summer of 1980 has been reconstructed from congressional testimony, antitrust court documents, and memoirs from both sides. The IBM team negotiated the operating system as a procurement contract — buy software per unit, integrate it, ship the combined product. The Microsoft team, advised by Gates’s lawyer father and by Steve Ballmer, had been thinking about the broader strategic implications in ways the IBM team had not.
The specific clause that mattered was the non-exclusive licensing right. IBM had asked for an exclusive license. Microsoft refused. The compromise was a non-exclusive license at a modest per-unit fee that allowed Microsoft to license MS-DOS to other manufacturers. From IBM’s perspective in 1980, the clause was a minor concession. The company did not anticipate that other manufacturers would compete with the IBM PC.
03The 1980s expansion and the shifting value
The personal-computer market grew at rates that nobody had predicted in 1980. The IBM PC launched in 1981 and was a substantial commercial success. The market for IBM-compatible PCs grew at over fifty percent annually through the mid-1980s. By 1990, the installed base of MS-DOS-compatible PCs exceeded fifty million machines worldwide, with hardware revenue running in the tens of billions of dollars annually. The market that had not existed in 1980 had become, by 1990, one of the major sectors of the global technology economy.
The value distribution within that market had shifted substantially. IBM, which had been the originator of the architecture, was no longer the dominant manufacturer. Compaq had overtaken IBM in personal-computer sales by the late 1980s. Dell, founded in 1984, had built a direct-to-customer sales model that was undercutting IBM on price across all major market segments. The hardware manufacturers as a category were competing on price, with margins compressed by the increasing standardization of the components they were assembling. The hardware business was becoming a commodity business.
04What the case actually shows
The IBM-Microsoft case has been one of the most-cited examples in business-school case studies for the past three decades. The standard reading is that IBM failed because it underestimated Microsoft. The standard reading is partly right and partly misleading. IBM did underestimate Microsoft, but the deeper structural feature was that IBM did not anticipate how value would accumulate in the personal-computer industry. The hardware was assumed to be where the value would stay. The operating system was assumed to be a component that supported the hardware. The actual outcome that the operating system would capture most of the value while the hardware would commoditize was visible in retrospect but was not obvious in 1980.
The structural lesson is about where industry value accumulates over time. The IBM team in 1980 was thinking about value as mainframe companies had thought about it for two decades: value was in the hardware, software was a service supporting it. The PC industry inverted that. Hardware became commodity. Software — the operating system, the applications, the network services captured the margins. The pattern would repeat across smartphones, across cloud infrastructure, across other digital categories.
05Conclusion
IBM continues to operate as a major technology services company focused on enterprise software, cloud services, and consulting. It has not recovered the dominant position it occupied in the late 1970s. Microsoft has been one of the most consistently successful companies in the global economy across the four decades since the deal, with a market capitalization that has periodically exceeded three trillion dollars.
Related books
