Major Pharma Companies Say AI Cuts Drug Development Timelines by Up to Half

nashnova research
今天发布阅读约 8 分钟

Seven of the world's largest drugmakers — including GSK, Novartis, and Eli Lilly — report that AI has cut early-stage drug development time by up to 50%, from molecule design through to clinical entry, marking the first large-scale, quantifiable pre-clinical evidence behind the AI-pharma narrative.

01

How much faster, exactly? What do the numbers show?

Novo Nordisk: the design-produce-test cycle dropped from roughly 8 rounds to 3, and each round shrank from 8 weeks to 4. This means → both the number of attempts and the time per attempt were halved — the combined effect is far more than "a bit faster."
Sanofi: AI-driven early-stage decisions on the asthma drug lunsekimig eliminated an entire dose-selection study, saving at least one year. In plain terms = it wasn't done faster — one whole step was skipped.
Novartis: AI identified 5 priority targets for kidney disease. Bristol Myers Squibb said a sickle-cell disease candidate could not have been developed without AI.
02

All seven are using AI — where exactly?

GSK, Bristol Myers Squibb, Novartis, Eli Lilly, Novo Nordisk, AstraZeneca, and Sanofi all say AI has moved beyond paperwork into molecule design, candidate screening, and safety-risk identification.
AstraZeneca executive director Sharon Barr: "Across the entire pipeline from discovery to development, there is no step where we are not using AI."
This reflects a shift: AI's role in pharma has moved from "efficiency tool" to a core capability that determines whether a drug can be made at all.
03

Anthropic's $1.5 trillion IPO — what are AI firms and pharma fighting over?

Anthropic listed on Nasdaq this week at a $1.5 trillion valuation. CEO Dario Amodei said: "We want fewer families to look back and wish the cure had come just a few years sooner." Anthropic has also opened its own lab and announced drug-development investments.
Big pharma broadly argues that decades of R&D experience and proprietary data are its edge over AI companies. This means → AI firms hold the algorithms; pharma holds the data and clinical know-how — each side believes it holds the decisive card.
Aliza Apple, head of Eli Lilly's TuneLab, was blunt: "If they can cure one disease first, then we can talk about the rest."
04

Where does clinical validation stand — how far from "actually working"?

As of March this year, researchers counted 129 AI-derived drug candidates across 69 companies, up modestly from roughly 117 eight months earlier. Most remain in early stages.
Waqas Haque, an oncology researcher at the University of Chicago, expects a visible jump in pre-clinical and clinical AI drug counts within one to three years — but adds: "I don't know whether these drugs will be more successful than non-AI drugs."
In plain terms = the time savings so far are in the lab. The real test is late-stage trials — whether AI-selected molecules fail less often than traditionally selected ones. That answer does not exist yet.

市场有风险,内容仅供研究参考,不构成投资建议。