The promises are grandiose: Anthropic’s chief executive said artificial intelligence could cure most human diseases within a decade. Another tech exec joked that AI researchers have started smoking cigarettes because they are so confident AI will cure lung cancer. Pharma executives don’t buy it.

“That’s complete BS,” said Jeff Leiden, executive chairman of Vertex Pharmaceuticals, recalling a recent conversation with a tech leader who claimed AI would cure 90% of human disease in the next seven years.

Vertex has used roughly 50 AI companies to comb through its data set of 10,000 drug compounds, Leiden said. Not one managed to generate a working drug.

“It just doesn’t work by itself,” Leiden said, “and any company that says they can do that—at least in our experience, we’ve looked at most of them—it’s not true.”Executives leading the companies that discover new drugs say AI isn’t going to cure cancer anytime soon. They are dimming the hype around the technology’s use in drug research, saying human biology is too complicated even for superintelligence.

Drugmakers are certainly using AI—even to to create new drugs. It is helping researchers pick targets, fine-tune experimental medicines and identify which patients would be good candidates for the treatments. Leiden says it is useful as a scientist’s copilot.

Yet in the race to integrate AI into medicine, a sharp divide has emerged between Silicon Valley’s promises and what pharma execs say are the harsh realities of drugmaking. There are still so many unknowns with biology that data analysis can’t fix.

“If we are in the industry long enough, we quickly learn to be humble,” said AstraZeneca Chief Executive Pascal Soriot. “Disease is complicated, biology is complicated. To say AI is going to cure cancer is maybe a little bit of a jump.”The British drugmaker recently inked a deal with AI biotech Insitro worth $40 million up front to discover new heart and weight-loss drugs, and signed an agreement with Oracle to help streamline the process for recruiting patients to clinical trials. A recently opened research and development site in Cambridge, Mass., uses AI to help speed up experiments.

McKinsey senior partner Alex Devereson, who focuses on AI in drug development, said the technology can design new experimental medicines quicker than before, but that doesn’t mean it is coming up with workable drugs. Around a quarter of mid-stage to late-stage failures in drug development can be traced to safety problems because drugs have “off-target” issues, he said, meaning they do unintended things within the body that lead to unwanted or even harmful side effects. Current AI models struggle to predict this.

“Investment is flowing to the problems that are easiest to solve, not the problems that are most important,” Devereson said.

Venture capital and private-equity firms invested $8.4 billion globally into AI drug discovery in 2025 alone, up from roughly $4 billion two years prior, according to data from McKinsey. Yet only a few AI-native biotechs—such as Insilico Medicine and Generate Biomedicines—have gotten drugs into late-stage patient testing and the Food and Drug Administration hasn’t cleared any of them yet.

Still, Silicon Valley’s largest players see drug development as a lucrative opportunity prime for disruption. “I think it will actually be possible to cure most human disease in ~5-10 years, as crazy as it may sound,” Anthropic CEO Dario Amodei posted on X in September.

Anthropic bought a stealth-mode biotech earlier this year and recently bought a one-person biotech running an almost completely AI-enabled operation. OpenAI announced it bought an AI drug-discovery biotech last month.

Most pharma companies have partnerships with AI developers, too. Eli Lilly and Nvidia said in January they would invest up to $1 billion over five years to research AI’s role in such things as drug discovery and robotics. Bristol Myers Squibb, Novo Nordisk and Sanofi have all announced partnerships with OpenAI or Anthropic.

Research chiefs including George Yancopoulos—Regeneron’s chief scientific officer and holder of more than 1,000 patents—see the potential for AI to assist in the lab but don’t see it as a revolution.

“AI is not going to just miraculously come in and tell us how to do things and cut things from 15 years to five years,” Yancopoulos said.

Regulators have recently proposed new frameworks to govern AI use in regulatory submissions and trial data analysis, but that doesn’t help accelerate some of the thornier parts of the process such as finding enough of the right patients and monitoring them throughout studies.

As the technology evolves, it is likely to become a more useful and potentially more powerful tool in drug development.

A team at Germany-based Bayer’s gene therapy unit recently used AI to comb through 100 million patient records in a matter of days, looking at genetic markers and patient histories to find the five best diseases prime for their gene-editing technology.

“Is that AI as such curing a disease? It isn’t,” said Sebastian Guth, Bayer’s chief operating officer. “The magic happens when you marry the brilliance of human scientists with AI.”