The promises are huge. A leading AI executive has said artificial intelligence could cure most human disease within a decade. The people who actually discover drugs aren't buying it. "That's complete BS," said the executive chairman of one drugmaker, recalling a tech leader who claimed AI would cure 90% of human disease in seven years. A sharp divide has opened between Silicon Valley's promises and the harsh realities of making medicine.
The drugmakers aren't anti-AI. They use it, even to design new molecules and to pick which patients fit a trial. One executive calls it a useful co-pilot for a scientist. But a co-pilot isn't a cure. His company ran about 50 AI companies against its data set of 10,000 drug compounds. Not one managed to generate a working drug. "It just doesn't work by itself," he said. Another chief executive put it plainly: disease is complicated, biology is complicated, and saying AI will cure cancer is a jump.
Where the money is going
The money hasn't waited for proof. Venture and private-equity firms invested $8.4 billion into AI drug discovery in 2025, up from about $4 billion two years earlier. Yet only a few AI-native biotechs have gotten drugs into late-stage patient testing, and regulators have cleared none of them. A McKinsey expert noted the trap: investment is flowing to the problems that are easiest for AI to solve, not the ones that matter most. About a quarter of mid-to-late-stage drug failures trace to safety problems, where a drug does unintended things in the body, and current AI models struggle to predict exactly that.
The science may get there. The point is the gap between a confident promise and a cleared drug, and how much money is riding on faith across that gap.
Our read
This is an Investments and Risk story, module IN04, about the distance between a story and a product. AI drug discovery is a thrilling pitch, and $8.4 billion flowed to it last year on the strength of the pitch. The people closest to the work, the ones with the most to gain if it worked, are the most skeptical. When the insiders are more cautious than the promoters, that tells you the valuation is running on hope, not results. Zero cleared drugs is a number worth sitting with.
This is really a lesson about any investment sold on a dazzling future. The bigger and more certain the promise, the more carefully you should ask where the proof is. Here's the check this week. Take any holding, or any pitch you've been given, that leans on a big future story, AI anything, a cure, a disruption. Ask one question: what has it actually delivered so far, in revenue or results, not projections? If the answer is mostly promise, that's fine to own in a small, speculative slice you can afford to lose. It's not fine as a large position or anything close to your safe money. We don't chase single-story biotechs in our model books, and we didn't this week. The honest version of owning the future is to own a broad slice of it and size it so that a broken promise is a disappointment, not a disaster. If you want help separating the story stocks from the steady ones in your portfolio, that's fifteen minutes.