The promise of AI drug discovery is about to meet its hardest test: the clinic. As of early 2026 there are more than 200 AI-discovered drugs in clinical development — yet not a single one has been approved by the FDA. The pipeline is full; the proof is still pending.
The state of the pipeline
The numbers show real momentum: of the 200-plus AI-discovered candidates, roughly 94 are in Phase 1, 56 in Phase 2 and 15 in Phase 3, with zero approvals so far. That distribution is exactly what you would expect from a young field — lots of early-stage shots on goal, a thinning cohort advancing, and the ultimate verdict still years away.
The candidates to watch
Several programs are carrying the field’s hopes. Insilico Medicine’s Rentosertib is widely cited as the first end-to-end AI-designed drug, moving from concept to Phase 1 dosing in just 30 months. Zovegalisib, a first-in-class inhibitor, has earned FDA Breakthrough Therapy Designation and is in Phase 3 for metastatic breast cancer. Absci’s AI-designed antibody ABS-201 is in early trials, and Generate:Biomedicines has an anti-TSLP antibody in Phase 3 for asthma.
Does AI actually speed things up?
Early evidence says yes, at the front end. A Boston Consulting Group study of 20 AI-focused pharma companies found that five of 15 AI-assisted candidates reached clinical trials in under four years, versus the historical average of five to six. Compressing the discovery-to-clinic timeline is meaningful — but it is the cheap, fast part of the journey. The expensive, slow part is proving a drug safe and effective in humans, where AI’s advantage is far less certain.
Why approvals matter most
Until an AI-discovered drug clears the FDA, the field’s central claim — that AI can produce better medicines, not just faster candidates — remains unproven. A candidate that is designed quickly but fails in Phase 3 has saved time on the wrong drug. One industry founder argues 2026 is the year AI drug discovery moves from research breakthrough to real deployment; the trials now reading out will decide whether that is true.
The bottom line
AI has unquestionably accelerated how quickly drug candidates are designed and pushed into trials. What it has not yet done is deliver an approved medicine. With 15 candidates in Phase 3, the next couple of years will reveal whether AI drug discovery is a genuine revolution in medicine — or a very efficient way of filling the early pipeline.
