原文 · 未翻译
AI is saving pharma billions in manufacturing and back-office work, just not in the lab
Key Points
According to Eli Lilly's digital chief, AI in pharma has barely moved the needle on actual drug discovery, and is instead paying off in the rest of the process.
Despite billions in investment, there's no proof AI is making clinical trials more successful. Pioneers like Recursion still don't have a single AI-developed drug on the market 13 years in.
So far, AI's wins have mostly come from manufacturing and back-office tasks. Analysts still expect savings of around $90 billion.
Eli Lilly's digital chief admits that, so far, AI is paying off in pharma everywhere except where the industry hyped it most: drug discovery.
Diogo Rau, Chief Information and Digital Officer at Eli Lilly, told the Wall Street Journal that the real benefits of AI haven't shown up in drug discovery at all, but in the rest of the process. It's a strikingly candid take, especially as Lilly pours money into billion-dollar partnerships with Nvidia and builds one of the most powerful supercomputers in the industry. Roche, GSK, AstraZeneca, and Merck have all signed their own billion-dollar deals with AI specialists in recent months.
Clinical trials tell a similar story. RBC analyst Trung Huynh isn't sold on the promised jump in success rates, saying there's no definitive proof yet that AI actually improves outcomes. Recursion Pharmaceuticals, one of the earliest pioneers in AI-driven drug discovery, still hasn't brought a single AI-developed drug to market nearly 13 years after its founding, and had to lay off 20 percent of its workforce last year. The company set out to crack the industry's notorious 90 percent failure rate in drug development.
Where AI is actually paying off
According to the WSJ, most of the AI payoff for drugmakers so far has come from streamlining back-office work and speeding up manufacturing, not from research breakthroughs. Lilly, for example, built a digital twin of its manufacturing process for tirzepatide (the active ingredient in Mounjaro and Zepbound) and used machine learning to find pressure and temperature combinations that cut production time and boosted output.
There are some early bright spots in drug discovery itself. Recursion designed an experimental cancer drug in 18 months, compared with an industry average of about four years. The catch is that human trials still take years. Even so, RBC estimates AI could save the US pharmaceutical industry around $90 billion over the next five years.
AI News Without the Hype – Curated by Humans