Pharmaceutical companies are adopting AI-powered predictive modelling to reduce the trial-and-error in formulation development, helping to select viable candidates earlier and cut both time and cost from a process that typically spans 12 to 18 years and costs around US$2.6 billion（約4200億円）. In January 2026, US and European regulators released joint guidance on AI use in drug development, establishing ten key principles to help ensure AI-generated recommendations are rigorously validated and transparently documented for regulatory confidence.
Summaries like this, in your inbox every morning.
Sign up free →What happened
AI predictive modelling is being used in drug formulation to narrow experimental space and reduce iteration cycles. According to GlobalData's Drugs database, just over 4,100 drugs have been developed or repurposed using AI, though most remain in discovery or preclinical stages rather than late-stage clinical development or market.
Why it matters
Drug development typically takes 12 to 18 years at an average cost of around US$2.6 billion（約4200億円）, with only about 10% of candidates succeeding. By identifying high-probability formulations early—when uncertainty is highest and decisions have the greatest downstream impact—AI has the potential to reduce costs, accelerate timelines, and lower the risk of costly late-stage failures. For development teams, this translates to speed (fewer iterations), reduced risk (better formulation confidence), and cost efficiency (lower API consumption).
What to watch
In January 2026, the European Medicines Agency and the US Food and Drug Administration released joint guidance on incorporating AI into drug development workflows, recommending ten key principles and emphasizing detailed record-keeping of data sources and processing steps to align with Good Practice requirements. A concrete example: in one programme with a poorly soluble compound, predictive modelling identified suitable polymer systems early, leading to a spray-dried dispersion with approximately an eight-fold increase in Cmax and a five-fold increase in AUC compared to the crystalline form.
Drug development has long been constrained by time, cost, and high failure rates—only about 10% of candidates make it through clinical development—creating pressure for tools that can de-risk early decisions. Traditional empirical approaches, while still essential, require iterative cycles that consume both time and expensive active pharmaceutical ingredients (API) without guaranteeing success. AI predictive modelling addresses this by analysing molecular data to identify viable formulations before costly experiments, thereby shifting decision-making to the earliest, highest-uncertainty stages where the downstream impact is greatest.
The regulatory environment has moved in parallel. The joint guidance from the EMA and FDA in January 2026 signals that regulators now expect AI integration to follow rigorous, transparent, and validated practices—not a free-for-all. This framing (detailed record-keeping, risk assessment, experimental validation) emphasises that AI outputs are a starting point for human review, not a replacement for it, and that confidence comes from traceability and evidence, not speed alone. In contexts like oncology, central nervous system disorders, immunology, and infectious diseases, where molecular complexity is high, AI's ability to navigate that complexity early could yield meaningful cost and timeline savings. However, realising these benefits depends on companies combining AI predictions with targeted experimental validation and maintaining the documentation standards regulators now expect.
No comments yet. Be the first to share your thoughts!
Log in to join the discussion
Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
1 minute a day. The AI essentials.
200+ sources · Email / LINE / Slack
