Drug discovery (2024)

It's widely known that developing a single drug can cost billions of dollars and take over a decade.

In 2023, Deloitte estimated that the top 20 pharma companies had increased their total R&D spending by 4.5%, reaching $145.5bn, up from $139.2bn the previous year. But the average R&D cost to develop a drug candidate from discovery to launch remains stagnant at $2.28bn. And around 90% of drug candidates still fail to pass clinical trials.

Can AI help reduce costs and improve efficiencies here? This is the question that a growing number of startups are grappling with.

For years now, AI has been touted as a way to significantly speed up the development of therapies by providing fresh insights from large datasets, predicting the success probabilities of clinical trials, and identifying target molecules. Funding for drug discovery startups in Europe peaked in 2021 when the sector soaked up $2.7bn.

Startups have raised $0.9bn from investors this year, meaning they’re likely on track to surpass the $1.3bn secured last year. And the average cheque size for Series A-C startups has increased over the past year, indicating growing investor optimism in the sector.

When we talk about AI being let loose in the industry, this covers generative AI (GenAI) but also machine learning (ML) models. Startups using ML methods to design therapies are gaining prominence. For example, UK-based LabGenius raised a £35m Series B in May in a round led by M Ventures. Founder James Field says GenAI can only answer questions about the early stages of the drug discovery process, while ML models are needed to help discover more complex molecular structures.

Another notable fundraise this year was Paris-based Aqemia’s €30m Series A extension. The company uses algorithms to invent new molecules and hopes that at least one candidate will enter clinical trials in 2025. The company recently announced a $140m partnership with French drug giant Sanofi.

Data indicates that AI-discovered molecules perform well in clinical trials, achieving a higher success rate than historical averages. One analysis reports an 80–90% success rate for AI-derived molecules in Phase I trials which then declines to around 40% during Phase II trials. While the research is still in the early stages, AI and ML tools show promise for drug discovery startups, and investor interest in the sector appears to be increasing.