Miles Wang exits OpenAI to launch $2B AI drug discovery startup
The AMW Read
New top-tier AI researcher exits to launch a well-funded vertical biotechnology startup; fits the AI spinout pattern with structural implications for the biotech segment.
Miles Wang exits OpenAI to launch $2B AI drug discovery startup
OpenAI researcher Miles Wang is leaving the company to start a new venture focused on AI-driven drug discovery, with plans to raise approximately $200 million at a $2 billion valuation, according to sources. Lightspeed is said to be leading the round. Wang, who joined OpenAI in 2024 after dropping out of Harvard, co-authored research on using AI to accelerate scientific discovery. Several other OpenAI researchers are expected to join him. The startup may focus on repurposing existing drugs and those that failed clinical trials, which could dramatically shorten the path to revenue.
Why it matters: This launch extends the established pattern of top-tier AI talent spinning out vertical applications, mirroring the Chai Discovery and Isomorphic Labs playbook. It also signals that the capital-compression arc in AI biotech is accelerating, with investors willing to bet billions on AI-first drug discovery—a segment that was mostly academic five years ago. The focus on drug repurposing addresses a key structural challenge: the long time-to-revenue in pharma. By targeting FDA-approved compounds, this startup could deliver faster commercial returns than traditional biotech, making it more capital-efficient and less dependent on speculative biology breakthroughs. This strategic choice may represent a deliberate pivot from the brute-force molecule generation approach that consumed billions at players like Recursion and Insilico.
Grounded expert take: The $2 billion pre-money valuation for a pre-product startup underscores the market's conviction that AI-native models, trained on proprietary biological data, can create a durable data moat—especially if they are initially validated through repurposed drugs with known safety profiles. However, the real test will be whether these models can consistently outperform traditional computational chemistry and wet-lab screening. If Wang's team succeeds even modestly, it will validate the 'conviction in vertical AI' thesis and likely trigger a wave of similar spinouts from Frontier AI labs, further blurring the line between artificial general intelligence research and applied life sciences.