IBM Placed Three Big Bets. One Paid, One Got Buried, One Is Still Unsettled.

IBM likes to tell its own story as a trilogy. Three big bets, placed by a hundred-year-old company refusing to die quietly: it bought the open-source company Red Hat for about $34 billion¹, it poured a fortune into Watson to revolutionize medicine, and it built the quantum computers that would carry computing past the limits of the machines that came before. Told that way, it sounds like one coherent leap into the future. It isn't. Strip the narrative away and you find three bets that didn't land the same way at all - one that compounds, one that got buried, and one that's still arguing with the referee.

The official story is that IBM made three coherent bets on the future. The real story is that it made one hit, one retreat, and one open question - and learned three very different lessons about what an adjacency actually costs.

The bet that pays: buying the plumbing everyone else stood on

Red Hat is the bet that worked, and it worked for a boring, durable reason: IBM didn't buy a product, it bought a position. Red Hat sells the enterprise version of Linux and the tooling that runs containers across whoever's cloud you happen to be using - the unglamorous plumbing that large companies standardize on and then almost never rip out. When the deal closed on July 9, 2019 at $190.00 a share, Red Hat was already a $3.4 billion business growing 15 percent a year.¹² IBM expected it to add roughly two points of compound annual revenue growth over five years² - and unlike most acquisition promises, this one kept showing up in the filings. The Red Hat segment was up 10 percent in the fourth quarter of 2022 and 9 percent in the third quarter of 2023.³ Double-digit growth that outlasts the press release is rare. This one did.

The mechanism behind the win is that Red Hat sat in the right adjacency - close enough to IBM's enterprise customers to be sold through the same relationships, but defended by switching costs IBM didn't have to manufacture. Companies that build their software on Red Hat are stuck to it the way a city is stuck to its water mains. IBM bought the mains.

The bet that got buried: a moonshot that never made money

Watson Health is the cautionary half of the same instinct. Launched in April 2015, it was sold as a moonshot that would revolutionize cancer care, and IBM backed the rhetoric with reportedly more than $4 billion in acquisitions - Truven, Merge, Phytel, Explorys - stitched together into a health-data empire.⁶ The empire never turned a profit. A 2017 STAT News investigation found the system bogged down in interoperability problems and questions about whether it was anywhere near ready for the clinic it promised to transform.⁶ The gap between the marketing and the machine was the whole problem: Watson was sold as a doctor and delivered as a database that struggled to talk to the hospitals it was meant to serve.

The retreat was quiet. In January 2022, IBM signed a deal to hand the healthcare data and analytics assets to the private-equity firm Francisco Partners⁴ - the corporate equivalent of moving a once-celebrated experiment into the garage. IBM never disclosed what it sold for; Bloomberg reported, citing unnamed sources, that the price was more than $1 billion.⁵ Even taking that figure at face value, it's a fraction of what went in. The same adjacency reflex that bought a fortress in Red Hat bought a money pit in healthcare - because in medicine, proximity to IBM's customers meant nothing without proximity to the messy reality of clinical data.

The bet still in dispute: when 'utility' isn't 'advantage'

The third bet hasn't resolved yet - and IBM has been more careful with its language than the headlines suggest. In June 2023, IBM published a paper in Nature, with UC Berkeley co-authors, reporting that its 127-qubit Eagle processor produced accurate results on a physics model at 100-plus qubits, beyond the reach of leading brute-force classical simulation. IBM called this 'quantum utility.'⁷ It was widely read as the day quantum beat classical computing. That reading didn't survive the summer.

Within weeks, independent academic groups published papers showing that classical tensor-network and approximate methods could closely simulate the same experiment on ordinary hardware - Tindall and colleagues, Begušić and Chan, Liao and colleagues, all in 2023.⁸ The result wasn't that IBM cheated; it was that the classical competition was stronger than the framing implied. The honest distinction, which IBM itself now draws, is between 'quantum utility' - useful results that compete with brute-force simulation - and 'quantum advantage,' outperforming every known classical method. The first is a real milestone. The second hasn't happened.

Isn't three bets the right number to be placing?

The fair objection is that this is the portfolio working exactly as designed. A company placing bets on the future should expect one to land, one to fail, and one to stay uncertain - that's not incoherence, it's how a venture portfolio is supposed to look. Fund three, kill the loser, double down on the winner, hold the option on the maybe. By that logic, selling Watson Health was discipline, not defeat, and the quantum work is a long-dated option you're not supposed to mark to market yet.

The objection holds, with one correction: the bets were never priced like options. Watson Health wasn't a small, hedged stake - it was a multi-billion-dollar acquisition spree marketed to the public as a cure for cancer, which is a very expensive way to learn that a database isn't a doctor.⁶ And the quantum claim wasn't sold as a hopeful early result; it was packaged as a milestone past classical computing, then walked back to 'utility' once the classical world answered.⁸ The strategy may be sound. The storytelling outran it - and overmarketing a bet doesn't make it more likely to pay; it just makes the retreat more expensive when the bill comes.

IBM's three bets aren't a trilogy. They're a lesson in what adjacency costs, told three ways. Red Hat shows that the best expansion is buying something customers can't put down. Watson Health shows that you can spend a fortune and a decade discovering that nearness to your customers isn't the same as nearness to their problem. And quantum shows that the gap between 'we did something useful' and 'we beat everyone else' is exactly the gap a press release loves to close and a referee will pry back open. One hit, one retreat, one open question - and the only thing that made them look like a single coherent leap was the company telling the story.