Recursion
Pioneering AI-driven drug discovery to decode biology and radically improve lives
Enterprise·Technical·Powered by Proprietary (Recursion OS)
Key strengths
Massive proprietary biological dataset (>50 petabytes) spanning phenomics, transcriptomics, proteomics, and moreAutomated wet lab with robotics and computer vision generating millions of cell experiments per weekRecursion OS — a continuously improving AI drug discovery and development platformBioHive-2 supercomputer built with NVIDIA for biopharma-scale computeActive clinical pipeline targeting oncology and rare diseases with first-in-class potential
Enterprise pricing
Salt Lake City, USA
Founded 2013
No ratings yet
Technical Overview: Recursion OS Platform
Recursion's drug discovery infrastructure is built on three pillars: data, models, and compute.
Data Layer
- >50 petabytes of proprietary, fit-for-purpose biological and chemical data
- Modalities include: phenomics (cell imaging), transcriptomics, proteomics, ADME assays, and de-identified patient data
- Generated continuously via an automated wet lab running millions of cell experiments per week using robotics and computer vision
Model Layer
- Intelligent ML models trained end-to-end on the proprietary dataset
- Applied to: target identification, molecule design, and optimization of drug candidates
- Continuously updated via a closed feedback loop between wet lab experiments and model outputs
Compute Layer
- BioHive-2: Built in partnership with NVIDIA, positioned as biopharma's most powerful supercomputer
- Designed to handle petabyte-scale biological data processing and large model training workloads
Partnership Integration
- Pharmaceutical partnerships: collaborative drug discovery programs
- Technology partnerships: computational and infrastructure co-development
- Data partnerships: enhancing dataset scope and generative AI capabilities for pharma
Note: Recursion OS is a proprietary internal platform. External API access is not publicly documented; technical integration is handled through formal partnership agreements.