AI Drug Discovery 2026: Insilico, Isomorphic Labs & the First Pipeline Wins
The first wave of AI-designed drugs is in late-stage trials. Here's who's winning, what's working, and where AI still falls short.

The pharmaceutical industry's 10-year, $2 billion-per-drug development cycle is finally cracking — and AI drug discovery is the reason. In 2026, the first wave of AI-designed molecules entered Phase 3 trials, biotech IPOs are roaring back, and DeepMind's Isomorphic Labs just signed its third pharma mega-deal. Here's the state of AI drug discovery in 2026.
The Headline Wins
- Insilico Medicine's INS018_055 (idiopathic pulmonary fibrosis) is the first fully AI-discovered, AI-designed drug in Phase 2b — readout expected late 2026.
- Recursion + Exscientia (merged in 2024) has 10 AI-designed candidates in clinical trials.
- Isomorphic Labs (Alphabet) signed deals with Novartis ($1.2B) and Eli Lilly ($1.7B) — first IND filings expected 2026.
- AlphaFold 3 now predicts protein-ligand and protein-DNA interactions with near-experimental accuracy.
How AI Compresses the Timeline
Traditional discovery: 4-6 years from target to clinical candidate. AI-native pipelines (Insilico, Iambic, Genesis) are doing it in under 18 months. The savings stack across:
- Target identification — transformer models on multi-omics data find novel disease drivers.
- Generative chemistry — diffusion models design molecules with desired properties.
- Lab automation — robotic wet labs validate 10,000 compounds per week.
- Trial design — synthetic control arms cut Phase 2 size by 30%.
The Foundation Models for Biology
Every big pharma now has a "BioGPT" stack. ESM-3 from EvolutionaryScale predicts protein function. Chai-1 rivals AlphaFold 3 with an open license. Nvidia's BioNeMo is the cloud layer. See our multimodal AI deep-dive for the underlying tech.
Where AI Still Struggles
- Toxicity prediction — still the #1 cause of clinical failure; models help but don't solve it.
- Novel mechanisms — AI is great at optimizing known targets, weaker at finding truly new biology.
- Clinical trials — recruiting human patients is still the bottleneck.
Investment Landscape
AI biotech funding hit $18B in 2025, up 60% YoY. Names to watch: Isomorphic (private), Recursion (RXRX), Schrödinger (SDGR), Relay Therapeutics (RLAY), Genesis Therapeutics (private), and Iambic (private).
FAQ
When will the first AI-designed drug be FDA approved?
Most likely 2027-2028. Insilico's pipeline is furthest along; Isomorphic's first IND filings in 2026 won't yield approval until 2030+.
Is AI drug discovery hype or real?
Real, but timeline-compressed marketing is overblown. Real wins are in preclinical speed — clinical timelines still take years because of human biology, not AI.
Conclusion
2026 is the year AI drug discovery moves from "interesting demos" to "shipping medicines." The companies that combine great AI with great wet-lab execution will define the next decade of pharma. Explore more in our AI healthcare 2026 trend and AI Trends category.






