Isomorphic Labs released IsoDDE on 10 February 2026, a unified drug-design engine that combines structure prediction, ligand binding, affinity prediction, and antibody interaction modelling into a single pipeline. IsoDDE roughly doubles AlphaFold 3's accuracy on the hardest ligand-binding cases (less than 20 percent sequence similarity to training data). Isomorphic targets first-in-human trials for an AI-designed candidate by end of 2026. This page consolidates the disclosed state of AI drug discovery: AlphaFold deployment, the competing labs, the pharma partnerships, and the regulatory readiness for AI-derived therapeutics.
Key Findings
- IsoDDE released 10 February 2026 unifies structure prediction, ligand binding, affinity prediction, and antibody-antigen interaction modelling. Reported accuracy on less-than-20-percent-similarity ligand cases: approximately 50 percent, compared to approximately 23.3 percent for AlphaFold 3.
- AlphaFold structure prediction serves more than 3 million researchers across more than 190 countries via the public AlphaFold Database; cumulative database entries exceed 230 million predicted structures.
- Isomorphic Labs has active partnerships with Eli Lilly and Novartis (announced 2024) plus a growing roster of mid-cap pharma partners. The targeted first-in-human trial for an IsoDDE-designed candidate is on track for end of 2026.
- Competing labs include Recursion (NYSE: RXRX), Insitro, Genesis Therapeutics, Atomwise, Cradle, and the in-house programmes at Eli Lilly, Pfizer, Roche, and AstraZeneca.
- The FDA approved its first AI-derived therapeutic in late 2025 (a small-molecule Insilico Medicine candidate for idiopathic pulmonary fibrosis entered Phase 2 in 2024); the regulatory framework for AI-derived candidates remains workload-comparable to traditionally derived candidates.
AI Drug Discovery Capability Comparison (May 2026)
| System | Lab | Capability | Public Availability |
|---|---|---|---|
| AlphaFold 3 | DeepMind / Isomorphic | Protein, RNA, ligand structure prediction | Public via AlphaFold Server |
| IsoDDE | Isomorphic Labs | Unified drug-design engine | Partner-only |
| ESM-3 | EvolutionaryScale | Protein language model with generation | Open weights with restrictions |
| Recursion's BioHive | Recursion (RXRX) | Phenotypic screening platform | Partner platform |
| Genesis Therapeutics GEMS | Genesis Therapeutics | Molecular design | Partner-only |
| Insilico Pharma.AI | Insilico Medicine | End-to-end drug discovery | Internal platform |
| Cradle | Cradle | Protein engineering platform | SaaS |
| RFdiffusion | Baker Lab UW | Protein design via diffusion | Open source |
Pharma Partnerships
| AI Lab | Pharma Partner | Year | Deal Size |
|---|---|---|---|
| Isomorphic Labs | Eli Lilly | 2024 | $45M upfront + milestones up to $1.7B |
| Isomorphic Labs | Novartis | 2024 | $37.5M upfront + milestones up to $1.2B |
| Recursion | Bayer | 2020 | ~$30M upfront + milestones |
| Recursion | Roche / Genentech | 2021 | $150M upfront + milestones up to $12B |
| Insilico Medicine | Sanofi | 2022 | $21.5M upfront + milestones up to $1.2B |
| BenevolentAI | AstraZeneca | 2019, expanded 2022 | Multiple disease areas |
| Exscientia | BMS | 2021 | $50M upfront + milestones up to $1.2B |
AI-Derived Clinical Pipeline (Public Status)
| Candidate | Originator | Indication | Phase |
|---|---|---|---|
| INS018_055 | Insilico Medicine | Idiopathic Pulmonary Fibrosis | Phase 2 |
| EXS-21546 | Exscientia | Solid tumours | Phase 1 (paused) |
| REC-2282 | Recursion | Cerebral Cavernous Malformation | Phase 2 |
| REC-994 | Recursion | NF2-related schwannomatosis | Phase 2 |
| IsoDDE candidate (undisclosed) | Isomorphic Labs | Undisclosed oncology | Pre-clinical, FIH targeted late 2026 |
Strategic Context
Three trends define 2026 AI drug discovery. First, the unified-engine thesis is winning: IsoDDE and adjacent platforms move from single-task models (structure prediction, docking, ADMET) to integrated pipelines that combine multiple tasks under a shared learned representation. Second, the pharma partnership model is consolidating: most major pharma companies now have one or more AI lab partnerships, and the unit economics are shifting from upfront licensing to milestone-heavy structures. Third, the first wave of AI-derived candidates is approaching clinical readout: by 2027-2028 several AI-derived candidates will have produced human-trial data, which will set the empirical baseline for what AI-derived really means.
Brand Visibility Implications
AI drug discovery is a high-citation topic in life-sciences journalism and increasingly in AI assistant queries from biotech investors, pharma BD teams, and academic researchers. Brands selling adjacent products (AI for clinical trials, AI for regulatory submissions, AI for medicinal chemistry, AI for diagnostic imaging) face strong AI-mediated discovery surface for queries about AI-derived drug pipelines, AI biotech investing, and AI for life sciences.
Methodology
Capability and partnership data compiled from Isomorphic Labs, DeepMind, Recursion, Insilico, and partner company disclosures, plus reporting from Fierce Biotech, Endpoints News, and peer-reviewed publications. Pipeline status sourced from ClinicalTrials.gov and company press releases. Updated quarterly.
How Presenc AI Helps
Presenc AI monitors brand-mention rates on AI drug discovery queries across ChatGPT, Claude, Gemini, and Perplexity. For brands in biotech AI, pharma AI tooling, and adjacent life-sciences AI categories, this is the operational visibility into a discovery surface heavily influenced by analyst, journalist, and academic citations.