IBM Didn't Give Away the PC. It Mispriced One Clause and Lost an Industry.

On August 12, 1981, IBM put a $1,565 machine on the table and changed the shape of an industry it would spend the next twenty-three years slowly backing out of.² The IBM PC 5150 ran an Intel chip IBM didn't make, an operating system IBM didn't own, and a stack of parts anyone could buy off a shelf. It worked. It sold. And then it kept selling - for everyone but IBM. By the mid-1990s, the company that invented the standard had watched its share collapse from roughly 80% in 1982-83 to about 20% a decade later, overtaken by clone makers that did nothing but copy what IBM had published.⁹

The story everyone tells is that IBM, fat and arrogant, naively gave the PC away and got what it deserved. The truth is colder and more useful: IBM didn't give the PC away out of naivety — it made a deliberate speed-to-market trade, and then lost the platform on a single clause it underpriced. The blunder wasn't the open hardware. It was one line in a software license that nobody in the room thought would matter.

Open architecture was the price of shipping, not a gift

Pull up the actual decision and the naivety story collapses. IBM at the dawn of the 1980s was a company that designed everything itself, end to end - chips, systems, software, sales. A personal computer built that way would have taken years. IBM did not have years; it had a market being eaten by Apple and Commodore and a one-year mandate to ship.¹¹ So William Lowe walked into the Corporate Management Committee and told them the only way through: open architecture, non-IBM technology, non-IBM software, non-IBM sales, non-IBM service. He later said persuading the committee took half his presentation time, because for IBM this was heresy.³

Read that again as a tradeoff, not a mistake. Openness bought speed, and speed bought the market. The cost was that IBM ended up holding almost no proprietary ground: because none of the functional components were IBM's own design, the company obtained only a handful of minor patents - color monitor bytecoding, DMA access, the keyboard interface - and never even enforced them.⁷ IBM published the blueprint and kept the brand. For a while, the brand was enough. The danger was hiding somewhere else entirely.

The clause that cost more than the hardware ever could

Here is where the famous legend - Gary Kildall went flying, missed IBM, and blew the deal of the century - falls apart under its own weight. Accounts of the meeting genuinely conflict: IBM's own negotiator Jack Sams insisted he never met Kildall at all - a denial a colleague confirmed he made at the time - even as other accounts have Kildall returning to the office that same day.⁴¹⁰ What is not disputed is that no deal was signed, and the tidy "went flying and blew it" version is press-interview folklore.¹⁰ What actually happened is duller and more instructive: IBM and Digital Research couldn't agree on terms and a delivery schedule, so IBM turned to Microsoft - which didn't even own an operating system. Paul Allen found one: 86-DOS, the 'Quick and Dirty Operating System' from Seattle Computer Products, licensed it, had it bent to fit IBM's hardware, and IBM shipped it as PC DOS.⁵

Now the real fork. IBM took a non-exclusive license. Microsoft kept the right to sell the same operating system - as MS-DOS - to anyone with a compatible machine. This is the part the legend gets wrong twice over: IBM's lawyers were not ignorant of the clause. The evidence available suggests they saw it and concluded — whether from confidence in the IBM brand, the proprietary BIOS, or simple speed-to-market pressure — that non-exclusivity was an acceptable risk. Whatever the reasoning, it proved catastrophically wrong. Once clone makers legally reverse-engineered the BIOS — Columbia Data Products first in 1982, Compaq closely behind — every clone maker could buy the same DOS from Microsoft that IBM was buying¹⁵ - and the standard IBM created stopped requiring IBM at all. The hardware was always copyable. The operating system was the one piece IBM could have locked down, and it rented it out by the door.

Wasn't this just IBM's genius for reinvention?

The generous reading - the one IBM loyalists prefer - is that this was reinvention working as designed. IBM seeds an industry, the industry commoditizes, IBM gracefully moves up the value chain to fatter margins. And there's truth in it: by 2004 the PC division was an ~$11 billion business that analysts noted had 'sometimes lost money,' and IBM sold it to Lenovo to double down on the higher-margin services and software strategy Lou Gerstner had set in the 1990s.⁸ That sale was disciplined, not desperate. The exit was real strategy.

But notice what the reinvention story quietly excuses. IBM didn't choose to give up the PC platform; it lost it, watched its share crater, and only then decided the commodity business wasn't worth keeping. Reinvention was the escape hatch from a self-inflicted wound, not a master plan. And even the exit gets mythologized: the Lenovo deal is remembered as a clean $1.75 billion cash sale, when IBM's own 8-K shows the total consideration was approximately $1.75 billion comprising roughly $1.25 billion in cash and equity from Lenovo — not a straight cash payment — and IBM kept an 18.9% stake in the buyer.¹ Even when IBM left, it didn't quite leave. The pattern isn't serial genius. It's one platform mistake, escaped by exiting the commodity entirely.

IBM did not give away the PC. It made a smart, forced bet on speed, won the market it was chasing, and then handed the one irreplaceable layer to a partner for a flat fee. The hardware was always meant to be copied. The standard was supposed to belong to IBM. Instead it belonged to whoever could buy DOS from Microsoft and a BIOS from anyone - and that turned out to be everyone. The lesson outlived the machine: when you open a system, the value doesn't disappear. It just flows to whoever still owns the part you can't reverse-engineer. IBM published the road. Microsoft owned the toll.