CellType

CellType

The agentic drug company. We simulate human biology.

Winter 2026ActiveHealthcareDrug Discovery and DeliveryAI-powered Drug DiscoveryArtificial IntelligenceMachine LearningBiotechTherapeuticsNew York, NY, USA
CellType is building an agentic drug company. AI agents that run the full drug discovery pipeline on top of biological foundation models that simulate human biology. Our core technology, developed with Google DeepMind, has already discovered and validated a new cancer treatment signal. We’re working with Top 10 pharma.

Verdict

High Signal
Market Opportunity
Drug discovery is a multi-hundred-billion dollar industry with pharma spending $50B+ annually on R&D. ICP is clearly Top 10 pharma and biotech companies needing preclinical de-risking and trial acceleration. The pain point — predicting human drug response before first-in-human trials — is enormous given ~90% clinical failure rates. B2B enterprise contracts in this space are typically $1M+ ACV.
High Signal
Founder Signal
David van Dijk is a Yale Associate Professor with 11,000+ citations, published in Cell/Nature/NeurIPS/ICML, and co-developed the core C2S technology with Google DeepMind — genuinely world-class scientific credentials with real shipped research. Ivan Vrkic co-authored Cell2Sentence (ICML 2024), led foundation model training at a biotech startup, and built SCADA software for CERN's LHC at Siemens — strong technical depth for a CTO. The tech was invented by the founders themselves, not licensed or copied.
Medium Signal
Competition
Competitors include Recursion Pharmaceuticals (public, $1B+ market cap), Insilico Medicine, Exscientia, and Genentech's internal AI efforts, plus foundation model plays like Insitro and BioMap. CellType differentiates through the Cell2Sentence approach (language model native, not graph/GNN-based), a validated Google DeepMind collaboration, and demonstrated SOTA benchmarks — but the 'AI for drug discovery' space is crowded and well-funded. The proprietary 27B model trained on 1B+ transcriptomic tokens is a meaningful technical moat if it holds.
Medium Signal
Product
CLI tool is live and open-sourced, benchmarks show state-of-the-art performance on single-cell tasks with a 27B parameter model (C2S-Scale), and the website describes a full product stack across target de-risking, toxicity, clinical acceleration, and biomarker discovery. However, no named paying customer logos, no pricing page, no confirmed revenue — 'working with Top 10 pharma' is unverified and no specific deal is named. Still early-stage commercialization despite strong technical foundation.
OverallA Tier

CellType has one of the stronger technical foundations in the W26 batch: a Yale professor with 11K+ citations who literally invented the core IP, a Google DeepMind co-development that generated a lab-validated cancer hypothesis covered by Sundar Pichai, and SOTA benchmarks on single-cell tasks. The CLI is live and open-sourced, showing real product shipping instincts. The key risk is commercialization — 'working with Top 10 pharma' needs to convert to signed contracts with real revenue, and the drug discovery AI space is brutally competitive with well-capitalized incumbents like Recursion. If they can close even one enterprise pharma deal with disclosed terms, this flips to S-tier.

Active Founders

David van Dijk
David van Dijk
Co-Founder & CEO

Co-Founder & CEO of CellType (W26). Yale Professor, 11,000+ citations. Built biological foundation models with Google. Building the agentic drug company.

Ivan Vrkic
Ivan Vrkic
Co-Founder

Co-Founder @ CellType (W26). ML @ Yale & EPFL. Building the next-generation agentic drug company. Co-developed our core technology Cell2Sentence at Yale and published at ICML. Previously led large-scale foundation model training at a biotech startup and built software to control CERN's Large Hadron Collider.

CellType
CellType
TierA Tier
BatchWinter 2026
Team Size2
StatusActive
LocationNew York, NY, USA