AI may speed molecule design and lab screening, but about 80% of drug‑development costs happen in clinical trials. Even perfect preclinical prediction saves weeks, doesn’t bridge animal‑to‑human translation, and won’t halve timelines without trial‑stage breakthroughs. Mega‑rounds for preclinical AI platforms may be mispricing where value is created.
— It resets expectations for AI‑in‑biotech by showing that without clinical‑stage innovation, AI won’t deliver the promised cost and time collapses.
Isegoria
2025.09.25
64% relevant
Mindstate used AI to select/design a candidate and reported Phase I safety/psychoactivity in 47 volunteers, but efficacy and most costs still lie in later clinical stages—illustrating that AI progress in preclinical design doesn’t by itself deliver the promised time/cost collapse.
Tyler Cowen
2025.09.11
60% relevant
His claim that AI won’t speed FDA approval and clinical trials 'nearly as much as it ideally might' aligns with the argument that most drug‑development cost/time sits in the clinical stage, limiting how much AI can accelerate real‑world medical impact.
David R. Henderson, Charles L. Hooper & Solomon S. Steiner
2025.09.10
70% relevant
The article claims the proof‑of‑efficacy step drives most costs, risk, and time in drug development—aligning with the idea that ~80% of costs occur in clinical trials, so reforms must address clinical stages rather than only preclinical tech.
Tyler Cowen
2025.09.03
100% relevant
The article cites Xaira’s $1B seed and notes that 80% of costs are clinical, limiting impact from preclinical AI gains.