You’ve probably heard the story: OS/2 was a technically superior operating system—more stable, more secure, more forward-thinking than Windows. It failed because IBM couldn’t execute, or because Microsoft had a better product, or because the market just didn’t get it.
That story is wrong. And the truth is far more uncomfortable.
OS/2 died not because it was inferior, but because IBM made a strategic decision that handed Microsoft the very knife that would slit OS/2’s throat. The tragedy isn’t technical failure. It’s strategic suicide.
Here’s what really happened—and why every founder, product manager, and platform builder should feel a chill running down their spine.
The Partnership That Was Never a Partnership
In the mid-1980s, IBM needed an operating system for its next-generation Personal System/2 line. Microsoft had MS-DOS, but IBM wanted something more powerful, something that could own the future of computing. So they partnered with Microsoft to build OS/2 together.
Seems logical, right? Two giants, complementary strengths, shared vision.
But here’s the catch: Microsoft’s survival depended on OS/2 failing.
Microsoft was already building Windows 3.0—a product that would compete directly with OS/2. If OS/2 succeeded, Windows would be irrelevant. Microsoft would become a second-tier software vendor. Bill Gates understood this intimately. IBM? They apparently missed the memo.
“Partnerships work when both companies have aligned incentives,” says a former Microsoft executive quoted in the video. “We had an existential reason to make sure Windows won. IBM had… a PowerPoint slide.”
The result? Microsoft slowed progress on OS/2, prioritized Windows development, and eventually abandoned the joint project entirely. IBM was left with a codebase, a handful of loyal developers, and a burning platform with no ecosystem.
The Ecosystem Trap
IBM’s error wasn’t technical mediocrity. OS/2 2.0 was genuinely impressive—preemptive multitasking, a stable 32-bit kernel, advanced memory protection. It made Windows 3.1 look like a toy.
But IBM treated OS/2 like a product. Microsoft treated Windows like a platform.
That distinction matters. A product wins on features. A platform wins on network effects—developers, applications, hardware compatibility, user habits. IBM spent millions engineering the OS. Microsoft spent millions convincing developers to write for Windows, manufacturers to bundle it, and users to never switch.
Control the incentive, not the code. If you control the ecosystem, you don’t need the best product.
IBM could have invested in a developer program, lowered the cost of entry, or even acquired a competitor to seed the ecosystem. Instead, they trusted Microsoft to do the heavy lifting—while Microsoft quietly redirected that energy into killing their partner.
The Twist: What IBM Missed
The most painful part? IBM had the resources to win. In 1990, IBM had $69 billion in revenue. Microsoft had $1.8 billion. IBM could have outspent, outhired, outmaneuvered Microsoft on developer relations, marketing, and distribution. They didn’t.
Why? Because IBM saw OS/2 as a technical battle. Microsoft saw it as a strategic one. IBM optimized for engineering quality; Microsoft optimized for market control. One created a better OS. The other created a better market.
The lesson echoes today: Google+ was arguably better than Facebook in 2011. Slack is technically brilliant, yet Teams wins because it’s bundled. The best product doesn’t win—the one that owns the ecosystem does.
If you’re building a platform, don’t ask ‘Is this better?’ Ask ‘Does this align the incentives of everyone who matters?’
The Real Villain: Strategic Naivete
IBM’s real mistake wasn’t trusting Microsoft. It was entering a partnership without understanding the asymmetry of incentives. Microsoft had everything to gain from OS/2’s failure. IBM assumed goodwill. Goodwill doesn’t survive when your partner’s survival is on the line.
This is the hard truth that every startup, every product manager, every platform builder must internalize: If your success depends on a competitor’s cooperation, you’ve already lost.
OS/2 wasn’t killed by a better product. It was killed by a better strategy. And that strategy was executed by the company that knew exactly what game they were playing—while IBM was still reading the manual.
FAQ
Q: Was OS/2 actually technically superior to Windows 3.1?
A: Yes, by most accounts. OS/2 2.0 had preemptive multitasking, a 32-bit kernel, and better memory protection. Windows 3.1 was a 16-bit overlay on DOS. But technical superiority doesn't matter if you can't attract developers and users.
Q: What can a modern startup learn from OS/2's failure?
A: Don't enter a partnership where your partner has an existential incentive to see you fail. Align incentives before you align code. And never confuse engineering excellence with market victory—you need to control the ecosystem, not just the product.
Q: Could IBM have done anything differently to save OS/2?
A: They could have terminated the Microsoft partnership early, invested massively in their own developer ecosystem, or even acquired a competing platform like Digital Research's DR-DOS to gain leverage. Instead, they bet on goodwill and lost.