AI-Driven Drug Repurposing Market (2026-2036)

The growth of this market is mainly driven by the reduced development cost and attrition risk of a molecule that already carries an established human safety package, by seven years of orphan exclusivity that makes rare disease the one indication class in which a repurposed drug can recover a development programme, by a regulatory pathway that allows an applicant to rely by reference on safety and effectiveness findings already made for the listed drug and to submit only the information supporting the new indication, and by the growing availability of linked real-world, transcriptomic and claims data against which a new indication hypothesis can be tested before any trial is funded. Computational screening of approved and shelved molecules against disease signatures costs a fraction of a discovery programme, so the method produces hypotheses at a rate the clinical system cannot absorb, and the market is therefore limited by which hypotheses can be funded rather than by how many can be generated.

However, factors such as the three-year exclusivity available for a new indication on an already approved drug, the absence of composition-of-matter protection on an off-patent molecule, off-label prescribing that captures the clinical benefit without supporting a commercial return, and the difficulty of financing a phase 3 programme on a molecule any manufacturer can make are expected to restrain the growth of this market to some extent. Three years does not recover the cost of the clinical investigations that earned it, so repurposing concentrates on rare disease indications where seven years applies and on molecules whose originator still holds exclusivity and can repurpose its own asset, and that narrowing rather than any computational limit is what caps the size of this market.

21 U.S.C. 355(c)(3)(iv), under which approval of an application for a change to an approved drug supported by reports of new clinical investigations essential to approval may not be made effective before the expiration of three years from the date of approval of the supplement. A company that funds clinical investigations to establish a new indication for an already approved drug obtains three years in which a competing application for that change cannot be made effective, against 21 U.S.C. 355(c)(3)(ii), under which a later application referring to a new chemical entity generally may not be submitted before the expiration of five years from the date of approval, and against 21 U.S.C. 360cc, under which approval of a drug designated for a rare disease or condition carries seven years of exclusive approval from the date the application is approved for the same approved use or indication within that rare disease, subject to exceptions where the holder cannot ensure sufficient quantities to meet patient needs or consents in writing, and under which, for drugs approved after 18 August 2017, a sponsor seeking exclusive approval for an otherwise identical drug must demonstrate clinical superiority through a significant therapeutic advantage in efficacy or safety or a major contribution to patient care. Three years is not long enough to recover a phase 3 programme on a molecule that any manufacturer can make, so the computational ability to generate a credible new indication hypothesis is not the limiting factor in this market. The limiting factor is whether an indication exists in which the exclusivity on offer covers the cost of proving it.

Rare disease is where that arithmetic works, which is why it is the fastest-growing therapeutic area here while oncology remains the largest by current spending. A designated orphan drug carries seven years of exclusive approval for the approved use within the rare disease, prices in rare disease are set against a small treated population rather than against the cost of the molecule, and the trials are smaller because the populations are smaller. A repurposed generic entering a rare indication can therefore sustain a commercial return that the same molecule could never sustain in a common indication, and the clinical superiority requirement that applies to otherwise identical drugs approved after 18 August 2017 protects the holder against a later identical entrant rather than obstructing the original repurposer.

21 CFR 314.54, which permits a 505(b)(2) application for a modification of a listed drug such as a new indication, where investigations other than bioavailability or bioequivalence studies are essential to approval, and which provides that such an application need contain only that information needed to support the modification of the listed drug, with the applicant relying by reference on the safety and effectiveness findings already made for that listed drug and identifying it by established name, proprietary name, dosage form, strength, route of administration, application holder and approved new drug application number is the route most repurposing programmes take, and the reliance it permits is what makes the economics work at all. An applicant does not repeat the non-clinical and clinical programme that established the safety of the molecule, it relies by reference on findings the agency has already made and submits only what supports the new indication, which removes the largest single cost and the largest single source of attrition from drug development. The identification requirements attached to that reliance also mean the pathway is available only where a suitable listed drug exists, so the method applies to approved molecules and to molecules whose sponsor holds the original application, and not to compounds that were never approved.

Shelved and discontinued candidates occupy a commercially distinct position from marketed drugs and are the fastest-growing molecule category. A compound that completed phase 1 and failed in phase 2 for efficacy carries a human safety dataset, an established manufacturing process and frequently remaining composition-of-matter protection, and the company that holds it has written it off. Acquiring or licensing such an asset gives a repurposing company the safety package that makes development cheap together with the patent protection that a generic molecule lacks, which resolves the exclusivity problem rather than working around it. The constraint is access, since originators have historically been reluctant to release failed assets and the data packages are rarely assembled for transfer.

Off-label prescribing separates clinical benefit from commercial return in a way that has no parallel in other therapeutic markets. Once a repurposing hypothesis is published, a clinician may prescribe the existing drug for the new use without waiting for a label, so patients obtain the benefit while no sponsor recovers the cost of establishing it, and the stronger the published evidence the faster that substitution occurs. The practical consequence is that a repurposing programme has to be funded on an expectation of label-dependent value, in reimbursement, in a specific formulation or in a rare disease price, rather than on the clinical finding itself, and programmes whose only asset is the finding have consistently failed to attract commercial capital.

The U.S. Food and Drug Administration draft guidance Considerations for the Use of Artificial Intelligence To Support Regulatory Decision-Making for Drug and Biological Products, published in January 2025 as a Draft Level 1 guidance carrying the notice not for implementation and non-binding recommendations, issued under docket FDA-2024-D-4689, which sets out a risk-based approach for establishing whether an artificial intelligence model is credible for a specified context of use is the relevant position on the method itself, and what it does not do matters. There is no distinct approval route for an indication identified computationally, so a repurposed drug is reviewed on the clinical evidence submitted for the new use and the artificial intelligence that generated the hypothesis is a research tool rather than a regulated article. The guidance addresses the separate question of when a model output may support a regulatory decision, and its risk-based framing ties credibility to a specified context of use rather than to the model in general, which matters where a sponsor intends to use model-derived evidence in the submission itself rather than only to select the indication.

Key Questions Answered in the Report