Miles Wang departs from OpenAI to join a $2 billion AI pharmaceuticals startup.

      No name. No product. No confirmed funding. Yet, possibly a $2bn valuation. This outlines the latest venture in AI drug discovery.

      Miles Wang, a researcher at OpenAI, is set to start his own company, according to TechCrunch. He is currently in discussions to raise around $200m at a $2bn valuation, with Lightspeed potentially leading the investment. Several other OpenAI researchers intend to join him as well.

      However, there is a twist. Wang challenged both the funding figures and TechCrunch’s portrayal of the company, though he did not provide corrected specifics. Lightspeed chose not to comment. Negotiations are continuing, and the terms may evolve.

      Reviving discontinued drugs

      The specifics of what the startup will create remain unclear. Two sources informed TechCrunch that it may work on AI models aimed at discovering new applications for existing drugs, including those that did not succeed in previous trials.

      The rationale is financial. A drug that has already been approved for safety can be administered to patients and generate revenue much more quickly than one developed from scratch. While repurposing still requires proving the drug’s efficacy for its new application, it eliminates years of preliminary testing.

      Wang is an unconventional choice for leading a $2bn funding round. He transitioned from Harvard to OpenAI in March 2024 and is part of the reinforcement-learning team. According to him, his work encompasses alignment and safety, along with applications of AI in science, especially biology. He co-authored a lab study where GPT-5 proposed modifications to a cloning protocol while human researchers conducted the experiments. OpenAI later noted a 79-fold efficiency improvement in that specific setup, while emphasizing that no drug was developed.

      An AI-bio gold rush

      The excitement surrounding Wang reflects more on the overall market than on him personally. There is substantial investment flowing into AI applications in life sciences. Chai Discovery secured $400m at a $3.8bn valuation on the same day, with its co-founder also having links to OpenAI. Isomorphic Labs, a spin-off from Google DeepMind, raised $2.1bn in May.

      The trend is noticeable. Companies like ByteDance and Anthropic are pursuing similar goals, and OpenAI researchers continue to depart to start their own ventures.

      The challenge AI hasn’t tackled

      This enthusiasm overlooks the most difficult aspect. Creating a molecule is the straightforward part of the development pipeline, while successfully navigating human trials is far more complex.

      “Drug discovery is a ‘lighter’ aspect in the process of bringing a drug to market,” said Immunai CEO Noam Solomon to Calcalist, referring to Anthropic’s separate initiative rather than Wang’s. “The greater challenge is ensuring that drugs are effective in clinical trials.” These trials generally take about a decade, cost approximately $2.7bn, and have over a 90% failure rate.

      For now, Wang has no such concerns. He has an idea, a team beginning to take shape, and a market ready to invest nine figures even without a company name established.