Radical Numerics Raises $50M to Build Multimodal AI for Biological Design
Key Takeaways
- ▸Radical Numerics raised $50M from Emergence Capital and others to commercialize multimodal AI models that understand the entire biological system, not just single molecules
- ▸The founding team created Evo and Evo 2, the first AI models to generate DNA sequences at scale from 100,000+ species, trained at MIT's Liquid AI
- ▸The company has early commercial partnerships in cancer detection and national laboratory pathogen detection, with revenue models blending API licensing, proprietary fine-tuned models, and milestone payments
Summary
Radical Numerics, founded by researchers who created the Evo AI models for DNA generation, has closed a $50 million seed round led by Emergence Capital to commercialize AI-powered biology. The startup teaches artificial intelligence to read, write, and reason across the entire biological system—DNA, RNA, proteins, and molecules—simultaneously in a single model, a multimodal approach that sets it apart from single-modality competitors like Isomorphic Labs and Inceptive.
The founding team, including CEO Eric Nguyen, Chief AI Scientist Michael Poli, and three other researchers from MIT spinout Liquid AI, built the foundational Evo and Evo 2 models that were trained on genomes from over 100,000 species. The company has already secured two early commercial partnerships: one for pancreatic and multi-cancer detection, and another with a national laboratory for pathogen detection and characterization. The broader AI drug discovery market is projected to reach $25 billion by 2035, with competitors like Ginkgo Bioworks establishing major platform deals with cloud providers.
Radical Numerics confronts a critical biosecurity challenge head-on. Nguyen acknowledged that the same open-source Evo models his team released enabled researchers to design the world's first fully AI-generated functional virus last September—demonstrating both the technology's power and its dual-use risks. The company has brought on Andrew Weber, a former U.S. assistant secretary of defense for nuclear, chemical, and biological defense, as an advisor and is pivoting away from automatic open-source releases for future models to mitigate biosecurity risks.
- Radical Numerics explicitly addresses biosecurity risks from its own technology, hiring former defense officials and reconsidering open-source releases to prevent dual-use bioweapon development
