IBM's next CEO killed the plan to break it into a dozen 'Baby Blues' in 1993

By early 1993, the plans to dismember IBM were already drawn. Under CEO John Akers, the company had spent 1991 and 1992 slicing itself into more than a dozen autonomous business units — a mainframe company, a storage company, a printer company, a PC company — each with its own balance sheet and a road toward one day trading as a stock of its own.³ The press had a name for the pieces before they existed: the Baby Blues, a knowing echo of the Baby Bells that AT&T had been broken into a decade earlier. The logic seemed unarguable. A company that had just posted the largest single-year loss in the history of American business¹ was clearly too big, too slow, and too tangled to save whole.

Then a man who had run a cigarette company and a credit-card company — and had never run a technology company — arrived in April 1993, read the breakup plan, and killed it.² That reversal, more than any product or any ad campaign, is the reason IBM still exists as one company. And the story usually told about it is upside down.

The tidy version says a visionary saw the future and refused to let a great American institution be carved apart. The truth is colder and more interesting: keeping IBM whole was not a sentimental act. It was a financial one. The mainframe was dying — but it was dying rich, and its cash was the only thing that could pay for the company that had to be built to replace it.

The breakup was the consensus answer, not the crazy one: everyone diligent agreed IBM should be carved up — which is exactly why it's worth asking what they missed

It is easy, thirty years on, to treat the breakup plan as an obvious error dodged by a genius. It was not. It was the responsible, well-reasoned answer, and Akers' team reached it honestly. The whole world was moving away from the mainframe. Cheap Unix servers and personal computers were pulling computing out of the glass-walled data center and spreading it across every desk, and IBM — a company built end to end on proprietary, high-margin mainframe systems⁵ — was too slow to turn.⁴ If your core product is being commoditized out from under you, the textbook move is to set the healthy pieces free before the sick core drags them down with it. Let each Baby Blue compete on its own, raise its own capital, live or die on its own results. Break the tangle.

The structure IBM had inherited made the argument stronger still. This was a company so decentralized that central management often didn't see a product group's consolidated results until year-end.⁵ From the inside, IBM already looked less like one company than a dozen quarreling ones sharing a logo. Formalizing the split just made official what already felt true. Every diligent instinct pointed the same way — which is precisely why the decision to go against it is worth understanding.

The mainframe was the checkbook, and the checkbook can't be spun off: the piece everyone wanted to protect from the sick core was the piece paying for the cure

Here is the move the breakup logic misses. When you split a company into independent units so each can be judged on its own merits, you don't just untangle the org chart — you sever the cash flows between them. And in IBM's case, those internal cash flows were the entire strategy. The mainframe and services businesses still threw off high-margin cash even as their market shrank. That cash was what could fund the newer, unprofitable, must-build units and pay for combining them into whole solutions no single Baby Blue could offer alone.² Gerstner's read was that IBM's ability to stay one integrated company — to move money from the fat, dying part to the thin, growing part — was its competitive advantage, not the liability everyone assumed.²

Spin the mainframe unit off as its own public company and it does the rational thing a public company does: it maximizes its own results. It stops writing checks to subsidize a services arm it no longer owns. The cross-subsidy vanishes on the day the shares are distributed. So the plan to protect the healthy pieces from the sick core had it backwards — the 'sick' core was the checkbook, and you cannot spin off the checkbook and keep spending from it.

Wasn't this just refusing to change?: the honest objection is that 'kept whole' can be a euphemism for 'refused to let go' — and here it partly was

The fair counter is that 'keep the company together' is exactly what a frightened incumbent always says, and that dressing incumbency up as strategy is the oldest trick in the corporate book. Plenty of dying giants have refused to break up out of pride and called it synergy. So why does IBM's version count as insight rather than denial? Because Gerstner didn't preserve the mix in order to protect it — he preserved the cash engine in order to redeploy it. The bet wasn't 'stay the same.' It was 'stay one company so we can afford to change.'

And the honest part of the counter holds: staying a single legal entity in 1993 did not freeze IBM in place, and the popular retelling that IBM was heroically 'kept whole' overstates it. In the years that followed, IBM shed major pieces anyway — its Federal Systems unit, and later its PC and x86 server businesses — divesting piecemeal once the crisis had passed.⁴ So the real decision wasn't 'never break up.' It was 'don't break up all at once, under duress, in a fire sale that severs the cross-subsidy before the new business exists to replace it.' Gerstner didn't refuse to shrink IBM. He refused to let it be dismembered on the market's schedule instead of his own.

The plan to break IBM into a dozen independent companies wasn't foolish. It was the smart, consensus, spreadsheet-clean answer — and it would have quietly killed the company by cutting the one wire that mattered: the flow of mainframe cash to everything that still had to be built. Gerstner's real insight wasn't that IBM was too grand to divide. It was that the shrinking business everyone wanted to protect the company from was, in fact, the company's checkbook. You can spin off almost anything in a crisis. You cannot spin off the thing paying for the crisis and expect the bills to keep getting paid.