Radical Numerics raises $50 million to teach AI the language of biology and defend against threats

Exclusive: The researchers who built AI-generated DNA just raised $50 million to reinvent biology | Fortune

Eric Nguyen was a perpetual student, much to his parents' chagrin. After receiving a master's of engineering at Cornell and doing a stint training AI to interpret and understand the visual world, Nguyen enrolled in a Stanford bioengineering PhD program specifically to find a problem worth fighting for. "I basically went back to the PhD to look for purpose," he told Fortune. "I wanted to find something that I thought I could contribute to, that if I didn't work on it, nobody else would." He found it in DNA. Nguyen's startup Radical Numerics emerged from stealth with a $50 million seed round led by Emergence Capital, Fortune learned exclusively. Obvious Ventures, Triatomic Capital, Factory, and First Spark Ventures also participated. Patrick Collison, the CEO of Stripe and cofounder of the Arc Institute, backed the company at pre-seed. Radical Numerics teaches AI to read, write, and reason in the language of biology -- not just DNA, but RNA, proteins, and every other molecule that makes living systems work, all at once, in a single model. The company's founding team -- Nguyen, Michael Poli (chief AI scientist), Stefano Massaroli (president), and Armin Thomas (chief technology officer) -- are among the researchers who created the field of generative genomics. Three of the four previously built core technology at Liquid AI, an MIT-spinout designing new AI model designs. Together they built Evo and Evo 2, the first AI models capable of generating DNA at scale, trained on the genomes of more than 100,000 species. Last September, researchers using Evo's open-source weights produced the world's first fully AI-designed functional virus (it was harmless to humans). That milestone is what pushed the team to build a company. "It still wasn't being picked up in the way we thought it would," Nguyen said of the academic work. "So we basically said: we have to show the recipe." The overall AI drug discovery market is projected to reach $25 billion by 2035, and competitor Ginkgo Bioworks recently signed a five-year AI platform deal with Google Cloud. But most AI biology companies today are single-modality like Isomorphic Labs for proteins or Inceptive for RNA (which just signed a deal with Alnylam potentially worth $2 billion). Radical Numerics is instead betting that the bottleneck in drug development is about understanding how they behave inside an entire biological system. "Getting the drug made won't be the bottleneck forever," Nguyen said. "You have to understand the whole system." The company has two early commercial partnerships: one applying its multimodal model to pancreatic and multi-cancer detection, and one with a national laboratory to detect and characterize pathogens, including AI-generated ones. The revenue model is still taking shape but is a mix of API licensing, fine-tuned proprietary models for pharma partners, and milestone payments. "No one has figured out the right business model for how AI companies commercialize in life sciences," Nguyen argued. "If anybody says they have a formula, they're just full of it." There's a catch baked into the entire enterprise. The same models that could accelerate cancer diagnostics could also lower the barrier to designing biological weapons, and Radical Numerics knows it better than anyone, because its own open-source work enabled that first AI-designed genome. "The defense side is sorely losing the race," Nguyen said. The company brought on Andrew Weber, former U.S. assistant secretary of defense for nuclear, chemical and biological programs, as an advisor, and is partnering with a national lab specifically to build AI-powered pathogen detection. Future model releases won't automatically be open-source. Ninety-eight percent of the human genome is still not understood. Nguyen is betting the same technology that could one day explain it could also protect against those who might exploit it. That's either the best argument for building Radical Numerics -- or the most urgent reason to hope it works.

Radical Numerics launches with $50M to build general biological intelligence

Radical Numerics launches with $50M to build general biological intelligence Radical Numerics Inc., an artificial intelligence research lab building what it calls general biological intelligence, launched today with $50 million in funding to scale its models and hire frontier AI talent. The company is the work of the team behind generative genomics, the field built around Evo, the first AI model able to read and write DNA at scale and the largest fully open-source AI project in any domain. Evo and its successor Evo 2 made the cover of Science and Nature and external scientists later used the model to generate the first complete AI-designed genome, a bacteriophage that infects bacteria and is harmless to humans. Radical Numerics is building a class of models that learn directly from biological data across DNA, RNA, proteins and beyond, folding the separate strands of biology into a single system. The company argues that multimodal models reasoning across every dimension of biology at once can open paths to cancer diagnostics, drug target identification and biosecurity that single-modality models cannot reach. Alongside the launch, the company previewed Omnii, its next-generation genomic language model. Radical Numerics says early results show Omnii setting a new state of the art in identifying causal regulatory variants and transferring zero-shot to experimental settings. Without specific training, the model recovers experimentally validated functional variants at genetic locations tied to Alzheimer's disease and it also leads on detecting AI-generated or AI-manipulated pathogens. That second capability points to the company's dual mandate: advancing biological design for human health while building the defenses to protect it. Radical Numerics is working with a cancer diagnostics company on pancreatic and multi-cancer detection and with a national lab to detect and characterize pathogens, whether they occur naturally or are engineered by AI. "Evo showed that AI can generate DNA and whole genomes, the next generation of models will go further with the ability to control function and eventually create entirely new forms of life," said Eric Nguyen, chief executive of Radical Numerics, in the company's announcement. "The same models that can help cure disease may also lower the barrier to designing harmful biology. These forces are inseparable. Biology will be the most consequential application of AI." Nguyen founded the company with Chief AI Scientist Michael Poli, President Stefano Massaroli and Chief Technology Officer Armin Thomas. Three of the four previously worked on model design at Liquid AI Inc.. Its scientific advisers include Microsoft Corp. Chief Scientific Officer Eric Horvitz, Stanford's Chris Ré, Harvard geneticist George Church and Andrew Weber, a former U.S. assistant secretary of defense for nuclear, chemical and biological defense programs. Emergence Capital led the seed round. Obvious Ventures, Triatomic Capital Private LP, Factory and First Spark Ventures also took part, with Stripe Inc. co-founder Patrick Collison having backed the earlier pre-seed round.