biotech.today

Insights

AI Drug Discovery by the Numbers

Data-driven insights across tools, companies, trials, and modalities — visualizing the state of AI in drug discovery.

1/6Tool landscape maturity

Most AI drug discovery tools remain in research stage — but the production tier is growing fast

Research
Research+
Production
Generative Design16Structure Prediction14Target Discovery12ADMET Prediction11Docking & Screening10Antibody Design10MD & Simulation7Protein LMs4Retrosynthesis3Single-Cell3AI Platforms1

Source: biotech.today tool database (60+ tools, Apr 2026) · practitioner-curated maturity assessment

Structure prediction leads with the most production-ready tools, anchored by AlphaFold2 and AutoDock Vina. Generative design and antibody engineering are maturing rapidly, with several tools graduating from research to production-grade in the past year. The amber Research+ tier represents tools with strong validation but pending broad industry adoption.

2/6Clinical pipeline by modality

Small molecules dominate AI-discovered clinical candidates, but gene editing is punching above its weight

Phase I
Phase I/II
Phase II
Phase III
Small Molecule8Gene Editing3Antibody2ADC2

Source: ClinicalTrials.gov, company disclosures · AI-discovered / AI-optimised programs only · biotech.today analysis

Of the AI-discovered or AI-designed drugs currently in clinical trials, small molecules account for the majority — unsurprising given the maturity of computational chemistry. However, gene editing therapies have achieved the most advanced stages, with Casgevy (CRISPR) already FDA-approved. ADCs are emerging as a key modality with multiple Phase III programs.

3/6Deal flow & capital allocation

2026 is already the biggest year for AI biotech deals, driven by mega-M&A

M&A
VC
Licensing
Partnership
IPO
$0M$5.5B$11.0B$16.5B$22.0BQ4 2025Q1 2026Q2 2026

Source: Press releases, SEC filings, Crunchbase, PitchBook · AI-focused biotech only · biotech.today analysis

Q1 2026 alone saw more than $30B in deal value, led by Novartis–Avidity ($12B) and the Lilly–Insilico licensing mega-deal ($2.75B). M&A dominates by dollar volume, but VC activity remains robust with multiple $100M+ rounds. The partnership category is expanding as pharma companies secure platform access deals rather than acquiring outright.

4/6Tool adoption landscape

Open-source tools cluster at Research+ maturity, while commercial platforms dominate production

Maturity →Accessibility →ResearchResearch+ProductionAlphaFold2 (Structure Prediction)ColabFold (Structure Prediction)Boltz-1 (Structure Prediction)Boltz-2 (Structure Prediction)Chai-1 (Structure Prediction)Protenix (Structure Prediction)ESMFold (Structure Prediction)OmegaFold (Structure Prediction)AlphaFold3 (Structure Prediction)OpenFold3 (Structure Prediction)IntelliFold-2 (Structure Prediction)DiffDock / DiffDock-L (Docking & Screening)GNINA (Docking & Screening)FlowDock (Docking & Screening)AutoDock Vina (Docking & Screening)Uni-Dock (Docking & Screening)GLIDE / Glide WS (Docking & Screening)GOLD (Docking & Screening)Deep Docking (Docking & Screening)ADMET-AI (ADMET Prediction)ADMETlab 3.0 (ADMET Prediction)SwissADME (ADMET Prediction)pkCSM (ADMET Prediction)Deep-PK (ADMET Prediction)ProTox 3.0 (ADMET Prediction)Chemprop v2 (ADMET Prediction)ADMET Predictor v12 (ADMET Prediction)QikProp (ADMET Prediction)RFdiffusion (Generative Design)RFdiffusion2 (Generative Design)ProteinMPNN (Generative Design)LigandMPNN (Generative Design)PocketFlow (Generative Design)REINVENT4 (Generative Design)SAFE-GPT (Generative Design)Chemistry42 (Generative Design)Chroma (Generative Design)ESM3 (Generative Design)DiffAb (Antibody Design)RFantibody (Antibody Design)

Source: biotech.today tool database · axis positions illustrative (based on maturity + openSource + related-tool count)

This bubble chart maps tools by maturity (x-axis) and accessibility (y-axis), with bubble size reflecting ecosystem impact. The densest cluster sits at Research+ / high accessibility — open-source tools like Boltz-1, DiffDock, and ESMFold that are freely available but not yet production-validated. Production tools tend to split between open (AlphaFold2) and commercial (Schrödinger).

5/6Category distribution

Structure prediction and generative design command the largest share of the tool ecosystem

16
14
12
11
10
10
Generative Design16(18%)
Structure Prediction14(15%)
Target Discovery12(13%)
ADMET Prediction11(12%)
Docking & Screening10(11%)
Antibody Design10(11%)
MD & Simulation7(8%)
Protein LMs4(4%)
Retrosynthesis3(3%)
Single-Cell3(3%)
AI Platforms1(1%)

91 tools tracked across 11 categories

Source: biotech.today tool database (Apr 2026)

The proportional breakdown reveals that protein structure prediction has the deepest tool bench, followed by generative molecular design and molecular docking. Newer categories like single-cell analysis and retrosynthesis are still nascent but growing. This distribution reflects where the most mature ML approaches have been applied in drug discovery.

6/6Modality readiness radar

Therapeutics lead on clinical validation; platforms lead on AI tool maturity

Therapeutics
Diagnostics
Platform
Enabling
Clinical ValidationAI Tool MaturityData AvailabilityRegulatory PathCommercial Adoption

Source: biotech.today modalities database · composite scores from tool maturity, clinical data, regulatory history

This radar chart compares aggregate readiness across five dimensions for each modality category. Therapeutics score highest on clinical validation and regulatory pathways, while AI platforms excel in tool maturity and data availability. Enabling technologies occupy a middle ground, with strong commercial adoption but limited direct clinical validation.