IBM Didn't Reinvent Itself Three Times. It Did the Same Thing Three Times.

In 1990, IBM posted revenues of nearly $69 billion and net earnings of $6 billion - the most profitable run in its history.² Two years later it announced a net loss of about $5 billion, the largest any U.S. company had ever recorded in a single year.¹ The same machine that printed money had, almost overnight, become the machine that burned it. The board fired the CEO and went looking for an outsider. What it found - and what it has kept finding ever since - was not a new idea. It was the same old reflex.

The official story is that IBM is a master of reinvention: mainframes to services, services to cloud, each transformation a feat of strategic vision. The truer story is that IBM has run one play three times - ride a captive installed base until it cracks, then pivot under duress - and called the panic foresight.

The mainframe was a tollbooth that forgot it was a tollbooth

Strip away the engineering romance and the early-1990s IBM was a beautifully simple business. Mainframes were the source of almost half its revenues and 70 to 80% of its profits, and the proprietary technology behind them produced gross margins above 50%.⁴ Once a corporation built its operations on an IBM mainframe, leaving meant rewriting everything - so IBM didn't just sell hardware, it sold a base it could meter for years. That is a wonderful position to hold and a terrible position to depend on, because the day a cheaper architecture arrives, the captive base doesn't drift away politely. It collapses. Revenue grew unabated through 1990 and then fell off a cliff, taking CEO John Akers with it.² The lock-in that looked like a moat turned out to be a single point of failure.

Gerstner didn't have a vision. He had a customer's grudge.

Into the wreckage walked Louis Gerstner - the first outsider CEO in IBM's history, fresh from running RJR Nabisco and, before that, eleven years as a senior executive at American Express.³ He is often described as a man with no technology background, and that misses the point entirely. Gerstner had spent over a decade as one of IBM's biggest customers, on the buying side of the very lock-in IBM was selling. His insight was not a leap of imagination; it was the obvious complaint of every CIO who had ever been over-charged for a box and left to wire it together alone. Customers didn't want more hardware. They wanted someone to make the whole tangle work. So Gerstner pointed IBM at services and integration - selling the glue rather than the parts. It was the right pivot. It was also a pivot that only happened because the company had already lost five billion dollars and run out of alternatives.

Then services became the new mainframe

Here is where the pattern reveals itself. The services business Gerstner built became, over two decades, exactly what the mainframe had been - a large, sticky, slowly fossilizing base. And it began to decay the same way. The managed infrastructure services arm fell from $21.8 billion in revenue in 2018 to $20.28 billion in 2019, then to $19.35 billion in 2020.⁸ Running data centers for other companies was becoming a commodity, undercut by public-cloud providers the way commodity servers had once undercut the mainframe. IBM's response rhymed with 1993: buy a future and cut the past. It acquired Red Hat for approximately $34 billion at $190 per share, closing in July 2019,⁵ and on November 3, 2021 it spun off the declining services business as Kyndryl, distributing 80.1% of the new company's shares to its own stockholders.⁶ A clean amputation of the limb that was no longer growing.

By 2022 hybrid cloud was $22.4 billion in revenue, growing 11% as reported and representing 37% of IBM's total.⁷ The story sells itself as the third great reinvention. But notice the structure underneath: a new captive platform, a high-margin position customers find hard to leave, a base IBM will ride for as long as it can. The script is identical. Only the product changed.

Isn't surviving a hundred years the whole point?

The fair objection is that this reading is too cynical. Most companies that dominate one technology era simply die when it ends - the graveyard is full of firms that owned a category and could not let go of it. IBM is still here, larger than almost any contemporary, having survived a loss that should have killed it. Surely reacting well to crisis is itself a kind of genius? It is - and that is precisely the more honest claim than 'visionary reinvention.' The skill IBM has actually demonstrated is not seeing the future early; it is recognizing, faster than its peers, when a profit engine has died, and being willing to cannibalize it. That is rare and valuable. But it is also reactive by definition. A company that pivots only after a $5 billion loss or three straight years of decline is not steering ahead of the curve - it is braking hard at the edge of it, repeatedly, and surviving on reflexes. The danger of that skill is the same each time: the captive base feels safe right up to the moment it isn't, and reflexes only work if the crash comes slowly enough to react to.

IBM is celebrated as the company that reinvented itself again and again. The more useful truth is that it did the same thing again and again - found a base it could lock customers into, charged a fat margin until something cheaper arrived, then turned, hard, only when the floor gave way. Three times now the turn has worked. Hybrid cloud is base number four, and somewhere out there is the architecture that will commoditize it too. When it arrives, IBM will likely survive the way it always has: not by seeing it coming, but by being very good at the brake.