The application of generative artificial intelligence to pharmaceutical research and development, from identifying promising target proteins to suggesting novel molecules and screening for toxicity.
Drug development is notoriously failure-prone. Only one in every ten drug candidates that enter human trials eventually reaches the market. Turning a promising molecule into a useful medicine typically takes ten to 15 years after its discovery. These economics mean that the cost of developing each successful drug is roughly $2.8bn.
AI-designed molecules show an 80-90% success rate in early-stage safety trials, compared with a historical average of just 40-65%. It will be years before it becomes clear whether success rates rise in later-stage trials too. But even if they do not, one model suggests that early-stage improvements alone could increase the success rate across the entire pipeline from 5-10% to 9-18%. McKinsey reckons that if AI is fully utilised by the pharma industry it could provide a boost worth $60bn-110bn annually.
By ingesting and analysing vast biological data sets, AI tools can identify promising target proteins and suggest novel molecules that could latch onto drug targets. They can sift through libraries of data to predict the potency and toxicity of candidates before a single test tube is touched. AI can also help with clinical trials, analysing health records to find the patients most likely to respond to novel treatments.
One projection suggests annual investment in AI drug discovery will rise from $3.8bn in 2025 to $15.2bn in 2030. In 2024 a dozen tie-up deals between pharma companies and AI firms were announced, with a combined value of $10bn, according to IQVIA, a health-intelligence company.
A new generation of AI-native biotech startups—particularly in America and China—is emerging. Pharma companies are increasingly forming alliances with AI-biotech firms, as well as with technology giants including Amazon, Google, Microsoft and Nvidia.
Isomorphic Labs, a spin-out from Google DeepMind, is trying to design entirely new therapeutic molecules from scratch inside a computer. It has contracts with Eli Lilly and Novartis. See protein design.
Nvidia has a generative-AI platform for drug discovery and is signing deals to offer design services to pharma companies. In October 2025 it teamed up with Eli Lilly to build the pharma industry's most powerful supercomputer.
Insilico Medicine, a biotech firm in Boston, was the first to apply transformer-based AI to drug discovery in 2019. Its drug rentosertib, for idiopathic pulmonary fibrosis, completed successful mid-stage clinical trials, having taken 18 months from idea to development candidate compared with the usual four and a half years. The firm has a pipeline of more than 40 AI-developed drugs for conditions including cancers and diseases of the bowels and kidneys.
GSK has developed Phenformer, a software tool trained to read genomes, linking genomic information with phenotypes to generate hypotheses about diseases. It also uses Cogito Forge, an AI agent-based system that can write its own code, gather datasets, draw conclusions and verify or falsify biological hypotheses through a literature search using three competing agents.
AstraZeneca says over 90% of its small-molecule discovery pipeline is now AI-assisted, sorting promising candidates from unpromising ones twice as fast as before.
Recursion, a firm in Salt Lake City, has built an AI "factory" in which millions of human cells are photographed undergoing chemical and genetic changes, allowing AIs to learn patterns connecting genes and molecular pathways.
Owkin, an AI biotech in New York, is training its model on a vast set of high-resolution molecular data from hospital patients. Its boss, Tom Clozel, argues that this work is moving towards true artificial general intelligence in biology.
Unlearn.AI, a digital-twins firm in San Francisco, creates synthetic patients to act as matched controls in clinical trials. Work published in 2025 suggested that this approach could reduce the control arm in an early Parkinson's disease trial by 38%, and by 23% in an Alzheimer's study.
OpenAI is working with Moderna, a pioneer of RNA vaccines, to speed the development of personalised cancer vaccines.
AlphaFold, developed by Google DeepMind, solved the problem of predicting protein structures. More complex puzzles, such as how cell membranes function, are likely to be cracked at some point. AIs are now being trained to model interactions between proteins and other molecules, to predict RNA folding and even to simulate cells.
As drug discovery becomes more efficient, governments will need to address potential bottlenecks in regulation and trials. America's Food and Drug Administration and the European Medicines Agency are themselves starting to use AI to screen the data they receive. If the costs and riskiness of innovation fall dramatically, patent terms—which typically provide ten to 15 years of market exclusivity—may need to become shorter. Companies such as OpenAI and Isomorphic Labs are training systems to reason and make discoveries in the life sciences, hoping these tools will become capable biologists. For now, drug firms have the advantage of a wealth of data and the context to understand it, so collaboration is the order of the day. But as AI models make biology more predictable, the balance of advantage may shift towards tech firms.
FORTUNE'S RULES TO LIVE BY: #2 Never goose a wolverine.