A researcher at OpenAI, Miles Wang, is reportedly in talks with investors to launch a new AI drug discovery startup that would be valued at approximately $2 billion, according to reporting from TechCrunch. The discussions are still early-stage, but the valuation figure alone signals how aggressively capital is chasing the intersection of large language models and pharmaceutical research.

Why This Deal Matters

Drug discovery is notoriously expensive and slow — it typically costs over $1 billion and takes 10–15 years to bring a single drug from target identification to market approval. AI promises to compress that timeline dramatically by predicting protein structures, identifying viable molecular candidates, and modeling drug-target interactions at scale.

Wang's background at OpenAI — one of the leading labs in frontier model research — would give a new venture immediate credibility in both the AI and biotech investor communities. Spinning out of a top lab with pre-built institutional trust is a well-worn path in deep tech, and the $2B pre-launch valuation reflects that premium.

The Broader Investor Thesis

This isn't happening in isolation. The biotech-AI crossover has attracted serious capital over the past two years:

  • Isomorphic Labs, DeepMind's drug discovery spinout, has partnered with Eli Lilly and Novartis in deals worth up to $3 billion
  • Xaira Therapeutics launched in 2024 with $1 billion in funding
  • Recursion Pharmaceuticals has built a multi-billion-dollar public company on AI-driven drug pipelines

Investors are betting that the same scaling dynamics driving progress in language models can be applied to biological systems — that more compute, more data, and better architectures will eventually crack problems that have stumped traditional pharma R&D for decades.

What This Means for Founders and Operators

For startup founders watching this space, the Wang story reinforces a few trends worth tracking:

  • Talent spinouts from foundation model labs are commanding outsized valuations even before a product exists. Pedigree is functioning as a proxy for product-market fit in early funding rounds.
  • Vertical AI — models trained and deployed for specific, high-stakes domains like drug discovery, legal, or finance — continues to attract premium multiples compared to horizontal AI tooling.
  • The life sciences sector is increasingly receptive to AI-native companies, not just AI features bolted onto existing workflows. This creates genuine wedge opportunities for startups willing to build deep domain expertise alongside model capability.

For marketers and growth teams at AI startups, this kind of headline also shapes the broader narrative around AI's societal value — drug discovery is a domain where even skeptical audiences accept that AI could deliver transformational outcomes, making it a useful reference point when arguing for AI's legitimacy in high-stakes decisions.

Still Early

It's worth noting that talks are ongoing and no deal has closed. Valuations floated during early fundraising discussions don't always survive contact with term sheets. But the fact that a pre-product, pre-revenue company spun from a researcher's work at OpenAI is being discussed at $2 billion is itself a data point about where investor sentiment sits in mid-2025.