(Author’s note: All facts in this piece come from open sources — The Wall Street Journal, STAT News, IEEE Spectrum, and others. Names and events are real. Our goal is to learn from others’ mistakes, not to throw stones at a giant. Great failures teach better than modest successes.)

Case History: The Dream of a Digital Hippocrates

In the early 2010s, IBM was on top of the world. Its AI Watson had humiliated the best players on the TV show Jeopardy!, and the company decided to direct that power at the hardest target of all — medicine. The idea was grand: to create a digital mind that would analyze millions of medical documents, diagnose diseases more accurately than doctors, and — no less — cure cancer.

IBM spared no expense for this dream. Billions were spent acquiring companies with medical data, including Truven Health Analytics for $2.6 billion. Total investments were estimated at tens of billions. The world was promised a revolution. The market held its breath.

Fatal Error: Selling a Hammer Instead of Building a House

The failure of IBM Watson Health is a classic case of a company falling in love with its technology while forgetting about the customer. IBM wasn’t selling a product — it was selling a “hammer,” a complex, powerful, universal AI from which, in theory, “anything could be built.”

As IEEE Spectrum aptly noted, it’s one thing to win a quiz game with clear rules, and quite another to deal with the chaos of real-world medical data. The Watson team tried to sell hospitals a complex and expensive “engine,” instead of offering them a simple, comprehensible “car” that solved one specific problem. Doctors didn’t want a “cognitive computing platform.” They wanted a tool that would help them quickly analyze MRI scans or select treatment protocols without disrupting their workflow. That tool was never provided.

Chain Reaction: The Market’s Immune System

Reality soon struck back.

• Rejection by doctors: A STAT News investigation revealed shocking facts: the system produced “unsafe and incorrect” cancer treatment recommendations. In practice, doctors found that the AI either gave obvious advice or made mistakes. The product didn’t help — it hindered.
• Data uprising: The AI was powerless against the “dirty” data of real medical records. Formats, doctors’ notes, missing test results — the very chaos that humans intuitively handle proved an insurmountable obstacle for the machine.
• Financial bleeding: For years the project burned through billions without generating profit. According to The Wall Street Journal, by the time of its sale the business generated about $1 billion in revenue but remained unprofitable.
• Capitulation: In early 2022, IBM surrendered. The once-revolutionary division, into which colossal sums had been poured, was quietly sold to a private equity firm. The price was undisclosed, but insiders spoke of just over $1 billion — a drop in the bucket compared to the investment.

The Autopsy Results: Verdict

Diagnosis: Death occurred from acute Go-to-Market deficiency. The patient overestimated its technology and ignored the needs of the end user.

It’s important to understand: this was not a failure of Watson technology itself. IBM’s AI platform continues to exist and evolve. The failure was specific to one business unit — Watson Health — and its Go-to-Market strategy. IBM had a brilliant “brain,” but never managed to create the “arms and legs” — a viable product that solved real doctors’ problems. Their story is a monument to the idea that even the smartest technology in the world is useless without the right product and a clear go-to-market strategy.