Watson’s fall shows AI is a vertical play

Watson Health falls short of expectations

In early 2022, IBM signed an agreement with Francisco Partners—a global investment firm that specializes in partnering with technology-enabled businesses. The private equity firm will take over various healthcare assets of Watson Health. Among them are healthcare database and software units, including Health Insights, the Market Scan collection of proprietary anonymized claims data, an evidence-based drug information service called Micromedex, imaging software services, and more.

The financial scope of the transaction has not been made public, but according to Bloomberg, the price is said to be over $1 billion.

In terms of strategy and profits, and respectively losses, the move makes sense: The Watson Health division has seen a decline in recent quarters. It reportedly brought in $1 billion in revenue annually but has failed to make a profit. The latest quarterly figures for the parent division “Cognitive Applications” didn't look particularly remarkable either. Plus, the problems come at a time when competitors like Oracle celebrate success in the healthcare market and BigTech consistently demands their claims on the market. 

Consequently, selling the Watson Health division has been in talks for months, and so the impending farewell is less emotional than you might think.

Watson Health meets the harsh reality

It’s always bitter when reality eats your dream. 

IBM has historically been very good at talking bluntly about technology-enabled disruption in healthcare. They always seemed on the verge of achieving something really big. Watson was supposed to bring the revolution. They wanted to integrate AI, analytics, and data to create advanced and augmented intelligence services for hospitals, insurance companies, and pharmaceutical companies. A great vision, but in the end, it seems as though it was nothing more than a fancy party trick. 

Granted, they have accomplished quite a bit in recent years and have celebrated some successes with their unit. Today, many drug manufacturers, government agencies, hospitals, and insurers are using Watson Health's data analytics and cloud services. A billion in sales is no coincidence. Nevertheless, IBM has never really been able to establish a foothold in the healthcare market.

One reason for this is that Watson Health always focused more on collecting data and analyzing academic papers, and less on consulting with real physicians who deal with patients every day. 

It's as if IBM has been demonized by its big Jeopardy! marketing stunt from a couple years ago: Just give Watson enough data, and any answer will be one click away. Sure, this top-down selling point went over well with accountants and administrators. But that's not how artificial intelligence is used effectively in healthcare. 

Good AI use cases are characterized by being problem-oriented—which, by the way, is why startups are having so much success right now; they often focus on one problem rather than a one-size-fits-all solution. AI works particularly well when it can take advantage of computing power in narrowly-defined areas.  

This is why IBM Watson works well when playing Jeopardy! but fails colossally in cancer research. If you play Jeopardy!, the rules are known, there are mountains of hard drives with historical data, and the feedback (right or wrong) produces simple patterns. In cancer research, however, the questions must be found before the answers can be addressed. 

Watson Health didn’t walk the talk but its medical assets are in good hands

Rather than transforming healthcare, Watson Health will likely go down in history as an object lesson in how not to introduce and deliver on big technology promises.

IBM should have taken the time to understand how physicians and other healthcare professionals could have worked with Watson, and to understand their goals, instead of optimizing for their Jeopardy! marketing vision. 

Maybe they could have done that. The success of customer-oriented development is no longer a secret. But the market has not made it easy for IBM in recent years. The pandemic has accelerated everything. It has shown the importance of being problem-focused and outcome-oriented in order to be at the top of the game. And for that, you need targeted and smart investments.

Something IBM is currently unable to provide. The mothership is undergoing a turnaround and has to pick its battles. The company wants to become more competitive in the cloud computing industry. In marketing terms, this means that IBM wants to focus on a “platform-based hybrid cloud and AI strategy,” according to IBM Senior Vice President Tom Rosamilia. 

So it's only logical for IBM to sell their medical assets. There is no reason to even worry about the assets sold. On the contrary. With Francisco Partners, they are probably better off because the PE firm can better support them with smart and targeted investments, which IBM is currently unable to do.