In February 2024, the US Food and Drug Administration approved a drug called Rentosertib for the treatment of idiopathic pulmonary fibrosis — a devastating lung disease with few effective treatments. The approval was notable not primarily for the drug itself but for how it was found. Rentosertib was identified by an AI system developed by Insilico Medicine, a drug discovery company that had used machine learning to predict that the molecule would be effective, design its structure computationally, and suggest a synthesis pathway — all steps that traditionally required years of laboratory chemistry and biological screening.
From initial AI-assisted identification to FDA approval took approximately seven years — still a long time by ordinary standards, but dramatically faster than the typical drug discovery timeline of twelve to fifteen years. And the pipeline behind it is substantial: dozens of AI-discovered or AI-optimized drug candidates are in clinical trials across therapeutic areas ranging from cancer to Alzheimer's to rare genetic diseases. The FDA approval of Rentosertib is not an endpoint. It is evidence that a fundamentally new approach to drug discovery is working.
Why Drug Discovery Has Always Been Slow
Traditional drug discovery is a process of searching an almost incomprehensibly large space. The number of possible small molecules with drug-like properties — what chemists call chemical space — is estimated at ten to the sixtieth power, or roughly ten trillion trillion trillion trillion trillion molecules. Laboratory screening can test perhaps a few million per year against a target of interest, with heroic effort. The odds of finding a molecule that is both effective and safe in this space by anything resembling random search are astronomically bad.
The process that has historically worked is expert-guided search — medicinal chemists with deep knowledge of how molecular structure relates to biological activity making educated guesses about which directions in chemical space to explore, testing those guesses, and iterating based on results. This process is effective but slow: it typically takes two to five years to identify a candidate molecule worth advancing to clinical trials, and only one in ten candidates that enter clinical trials ultimately receives approval. The total time from initial research to approved drug averages twelve to fifteen years, and the total cost averages over a billion dollars per approved drug.
What AI Changes About the Search
AI approaches to drug discovery are fundamentally different from this expert-guided random walk. Machine learning models trained on databases of known molecular structures and their measured biological activities can predict, with useful accuracy, how novel molecules will behave — their binding affinity for specific proteins, their toxicity profile, their solubility, their metabolic stability — before any laboratory testing occurs. This transforms the search from a primarily experimental process to a primarily computational one, with laboratory work focused on the most promising candidates rather than distributed across a broader exploratory space.
AlphaFold, DeepMind protein structure prediction system released in 2021, was the breakthrough that made much of this possible. Proteins are the molecular machines of biology — enzymes, receptors, ion channels — and drug molecules work by interacting with specific proteins in specific ways. Predicting what a drug-protein interaction will do requires knowing the three-dimensional structure of the target protein. For decades, determining protein structure experimentally was the rate-limiting step in structural biology — extraordinarily expensive, slow, and limited to proteins amenable to experimental structure determination.
AlphaFold predicted the three-dimensional structure of essentially every known protein with accuracy comparable to experimental methods, and did so freely available to researchers worldwide. The result was a sudden, dramatic expansion in the number of protein targets with known structures — and therefore in the number of targets amenable to structure-based drug design, where the three-dimensional shape of the protein guides the design of molecules that will fit into it like a key into a lock.
The Companies Building This
The AI drug discovery space has attracted significant investment and a diverse ecosystem of companies pursuing different approaches. Insilico Medicine, which produced Rentosertib, uses generative AI to design novel molecular structures. Recursion Pharmaceuticals combines robotic laboratory automation with machine learning, running massive biological experiments and using AI to extract patterns from the resulting data. Schrödinger applies physics-based simulation augmented with machine learning to predict molecular properties with high accuracy. Exscientia — acquired by Recursion — was among the earliest companies to advance AI-designed drug candidates to clinical trials.
The major pharmaceutical companies — Pfizer, Roche, AstraZeneca, Novo Nordisk — are all integrating AI deeply into their drug discovery pipelines, either through internal capability building or partnerships with AI-focused companies. The competitive pressure to adopt AI is intensifying as early results accumulate: companies that can find and optimize candidates faster will reach clinical trials sooner, generating earlier evidence of efficacy and eventually earlier approval and revenue.
Clinical Trials Are Where AI Faces Its Hardest Test
The drug discovery phase — identifying and optimizing a candidate molecule — is where AI has demonstrated the most dramatic acceleration. The clinical trials phase, where the molecule is tested in humans for safety and efficacy, has proven harder to accelerate with AI, and this is where most drug candidates still fail.
AI can help with clinical trial design — identifying patient populations most likely to respond to a treatment, optimizing dosing protocols, designing adaptive trial structures that can adjust based on interim results. AI can help with patient recruitment, one of the most time-consuming aspects of clinical research, by identifying eligible patients in electronic health records and clinical databases. And AI can analyze the data generated by clinical trials faster and with more sophisticated statistical methods than traditional approaches.
But the fundamental rate limiter on clinical trials is not analysis speed — it is the time required for biology to respond and for safety signals to emerge. You cannot run a two-year cancer survival trial in six months regardless of how good your AI is. The biological timescales of disease and treatment response are irreducible constraints that computational speed cannot compress. AI can make everything around the clinical trial faster and smarter, but the trial itself still takes the time it takes.
The Diseases That Matter Most
The areas receiving the most AI drug discovery attention reflect both scientific opportunity and commercial incentive. Cancer — the most lucrative drug market — is attracting enormous AI research effort. Alzheimer's and Parkinson's — diseases with large patient populations, no effective disease-modifying treatments, and desperate need for innovation — are high-priority targets. Rare diseases, where the small patient populations make traditional economics of drug discovery unfavorable, are a particularly promising area for AI: if AI can dramatically reduce discovery and development costs, drugs for rare diseases that would be economically nonviable under traditional development economics become feasible.
Antibiotic resistance — one of the most significant public health threats of the coming decades — is receiving growing AI research attention. The pipeline of new antibiotics has been nearly empty for decades because the economics of antibiotic development are poor: a course of antibiotics costs thirty dollars and is used for a week, compared to a cancer therapy that might cost tens of thousands of dollars per month. AI that can compress antibiotic discovery timelines and costs may make it economically viable to invest in this neglected area before antibiotic resistance becomes a full-scale global crisis.
The Rentosertib approval is a beginning, not a conclusion. The real test of AI drug discovery will be the performance of the clinical trial pipeline over the next five to ten years — whether AI-discovered candidates succeed at rates significantly higher than historically expected, and whether the combination of faster discovery and higher success rates translates to dramatically better outcomes for patients. The scientific and commercial bets being placed on these questions are enormous. The next decade will tell us whether they were right.