The trend line shows clinical development costs plummeting at Moore's Law speeds. We are crossing a threshold that will fundamentally restructure the pharmaceutical industry.
In 2019, bringing a novel small molecule from hit to Phase II required $2.6 billion in average total cost. By 2024, AI-first biotechs like Recursion and Insilico reduced this to ~$200M through computational pre-filtering. That is a 13x reduction in 5 years.
By my models, we are entering the steepest part of the cost curve. Three exponential trends are converging:
Synthetic accessibility prediction: AlphaFold3 + generative chemistry models eliminate 95% of unsynthesizable compounds before they enter the wet lab. Failed synthesis attempts drop from thousands to dozens per program.
Virtual patient trials: Digital twins trained on multi-omic patient cohorts predict Phase II failure with 85% accuracy. Failed candidates are killed in silico, not in $50M clinical trials.
Automated medicinal chemistry: Robotic synthesis platforms execute 10,000+ analogs per month, with AI optimizing each iteration cycle in real-time. The design-make-test cycle compresses from months to days.
Applying the exponential trend: by Q2 2027, the total cost to advance a computationally-discovered small molecule from virtual hit to Phase II proof-of-concept will fall below $10 million for at least 3 distinct therapeutic programs.
This is not incremental efficiency — it is a phase transition. Drug discovery becomes accessible to university labs, patient foundations, and seed-stage startups. The R&D monopoly held by Bi





























