It’s not that IBM hasn’t invested boatloads in Watson; it has. But while six years and billions of dollars is a lot of time and money for a Silicon Valley startup, it’s a pretty normal expenditure in the world of medical trials, most of which fail.

What’s more, when it comes to artificial intelligence, IBM is competing with rivals like Amazon and Alphabet—companies that are still growing fast, that see no need to buy back their shares, and that similarly see no need to hype their AI achievements before they’re really ready for the spotlight. Watson is great at publicity stunts—no article about it is complete without the requisite mention of the time it won Jeopardy (even, it seems, this one)—but AI is a very tough nut to crack, and one where the top computer scientists are in incredibly high demand. Those scientists aren’t going to the company with the best press, they’re going to the companies with the best AI, and the best science. Ask anybody in AI about Watson, and you’ll be told in no uncertain terms that they’re an also-ran in an incredibly competitive space.

The biggest hope for AI in oncology, then, is not that Watson is somehow going to get vastly better. Rather, it’s that Alphabet or someone else has been quietly building a game-changing AI without over-hyping it in advance.