Miles Wang, an OpenAI researcher who contributed to the company’s work applying artificial intelligence to biological and scientific discovery, is departing the ChatGPT maker to found a startup focused on developing AI models for drug discovery, according to multiple sources familiar with his plans. Several other OpenAI researchers are expected to join the new venture. The move signals a continued brain drain from frontier AI labs into biotechnology, where the promise of accelerating drug development has attracted both talent and substantial capital.
Funding Talks Place New Startup at $2 Billion Valuation
Wang is in discussions to raise approximately $200 million at a $2 billion valuation, according to two people with knowledge of the talks. Venture capital firm Lightspeed is in discussions to lead the funding round, sources said. The negotiations remain ongoing, no final agreement has been reached, and details could still change. Wang disputed the reported funding figures and the description of the company but declined to provide corrected numbers or further specifics. Lightspeed did not respond to a request for comment.
Even if the final terms shift, the scale of the proposed round underscores significant investor enthusiasm for applying AI to life sciences. A $200 million raise at a $2 billion valuation would place Wang’s startup among the most richly valued early-stage companies in the AI-drug discovery space, reflecting both the founder’s pedigree and the broader market appetite for AI-driven biotech.
What Is the New AI Drug Startup Working On?
Wang’s startup is expected to develop AI models designed to identify new therapeutic uses for existing drugs and, potentially, for drug candidates that previously failed in clinical trials, according to several sources. This approach, known as drug repurposing or repositioning, offers a faster path to revenue than de novo drug development because repurposed candidates have already undergone safety testing in humans and received FDA approval for their original indications. By applying advanced AI models to large datasets of molecular, clinical, and genomic information, the startup aims to uncover hidden connections between approved drugs and diseases they were not originally designed to treat.
The strategy carries lower risk and shorter timelines compared with traditional drug discovery, where candidates must clear Phase I, II, and III clinical trials — a process that typically takes a decade or more and costs billions of dollars. If successful, AI-driven repurposing could dramatically compress that timeline and reduce the capital required to bring new treatments to patients.
Wang’s Background and Trajectory at OpenAI
Wang joined OpenAI in 2024 after dropping out of Harvard University, where he was pursuing a bachelor’s degree in computer science. His decision to leave academia for the AI industry reflects a broader trend: investors have grown increasingly comfortable backing young founders who have not completed college, particularly those with demonstrated research capability and experience at frontier AI labs.
During his tenure at OpenAI, Wang co-authored research papers examining how AI models can automate and accelerate scientific discovery, including work published under the title “Accelerating Biological Research in the Wet Lab.” His research focused on evaluating whether large language models and other AI systems could perform tasks traditionally done by human researchers, such as designing experiments, analyzing biological data, and generating hypotheses. That work placed him at the intersection of AI and biology, a domain that has become one of the most active areas of investment and innovation in the technology sector.
The Competitive Landscape: AI Drug Discovery Heats Up
Wang’s move comes at a moment of rapid expansion in the AI-driven drug discovery sector. Just days before news of his startup emerged, Chai Discovery — a two-year-old company developing AI models that predict molecular interactions to identify new drugs — announced it had raised $400 million at a $3.8 billion valuation. Chai Discovery’s co-founder, Josh Meier, also previously worked as a researcher at OpenAI, underscoring the flow of talent from general-purpose AI labs into specialized biotech applications.
At an even larger scale, Isomorphic Labs, the Google DeepMind spinout led by Sir Demis Hassabis, raised a $2.1 billion Series B round in May 2026. Isomorphic Labs builds AI models specifically for drug discovery, leveraging DeepMind’s breakthroughs in protein structure prediction, most notably AlphaFold. The company’s massive fundraising round signaled that institutional investors see AI-powered biology not as a niche experiment but as a potentially transformative industry with blockbuster returns.
These three companies — Wang’s startup, Chai Discovery, and Isomorphic Labs — represent a new generation of AI-native drug discovery firms that aim to supplant or augment traditional pharmaceutical research and development. Their emergence reflects a conviction that AI models, trained on vast biological and chemical datasets, can identify drug candidates faster, cheaper, and with higher probability of success than conventional methods.
Why Investors Are Betting Big on AI for Life Sciences
The influx of capital into AI-drug discovery companies is driven by several converging factors. First, the cost of developing a new drug has risen steadily for decades, with some estimates placing the average cost at over $2 billion per approved therapy. Any technology that can meaningfully reduce that cost or accelerate the timeline represents enormous economic value. Second, the availability of large-scale biological data — from genomics, proteomics, clinical records, and high-throughput screening — has created the raw material needed to train sophisticated AI models. Third, breakthroughs in AI architecture, particularly transformers and diffusion models, have demonstrated that complex biological problems such as protein folding, molecular interaction prediction, and drug-target binding can be tackled computationally.
Drug repurposing, the approach Wang’s startup is expected to pursue, offers a particularly attractive risk profile. Because repurposed drugs have already been approved for human use, they bypass the most costly and failure-prone stages of drug development: preclinical safety testing and Phase I trials. The primary challenge becomes identifying which existing drugs might work for which new indications — a pattern-matching problem well suited to AI. Several repurposed drugs have already reached the market, including thalidomide for leprosy and multiple myeloma, and sildenafil for pulmonary arterial hypertension, but systematic AI-driven repurposing could dramatically expand the pipeline.
How AI Models Are Transforming Drug Discovery
The technical approaches underlying these startups vary, but they share a common core: using machine learning to model biological systems. Chai Discovery focuses on predicting molecular interactions — understanding how drugs bind to proteins, how proteins interact with each other, and how small molecules affect cellular pathways. Isomorphic Labs builds on AlphaFold’s protein structure predictions to design novel therapeutics. Wang’s startup, based on available information, appears to be targeting the repurposing space, which requires AI models capable of analyzing disparate datasets — electronic health records, genetic databases, published literature, and molecular structures — to identify promising candidate drugs for diseases with unmet need.
The models typically use graph neural networks to represent molecular structures, transformers to process sequential biological data such as gene sequences, and reinforcement learning or generative models to propose novel drug candidates. The field has matured rapidly: where earlier efforts struggled with data quality and model interpretability, current systems can make predictions that are increasingly reliable and actionable.
Implications for OpenAI and the Broader AI Talent Market
Wang’s departure continues a pattern of researchers leaving frontier AI labs to found or join biotechnology startups. The trend raises questions about how OpenAI and similar organizations can retain talent with domain expertise that spans both AI and the sciences. As AI-drug discovery companies raise increasingly large funding rounds, they can offer compensation packages and equity stakes that compete with — or exceed — what even leading AI labs can provide. For researchers motivated by the prospect of directly improving human health, the pull of a biotech startup may be especially strong.
The exodus also suggests that AI talent sees greater upside in applying existing models to high-value vertical problems than in continuing to push the frontier of general-purpose AI. Drug discovery is a trillion-dollar industry with clear, measurable outcomes: a drug that gets approved saves lives and generates revenue. That clarity of purpose and economic reward may be more compelling than incrementally improving benchmark scores on general reasoning tasks.
Drug Repurposing: A Faster, Lower-Risk Path to Market
The strategic logic of drug repurposing deserves closer examination. Of the thousands of drug candidates that enter clinical trials each year, only a small fraction ultimately receive FDA approval. Most fail due to safety concerns or lack of efficacy. Repurposing eliminates the safety uncertainty by starting with compounds that have already demonstrated acceptable safety profiles in humans. The remaining challenge is efficacy: proving that an existing drug works for a new disease. That still requires clinical trials, but the trials can be smaller, shorter, and less expensive because the safety data already exists.
AI models can accelerate the identification of repurposing candidates by mining real-world evidence, such as electronic health records that show unexpected correlations between prescribed drugs and improved outcomes for secondary conditions. They can also analyze molecular data to find mechanistic explanations for observed effects, providing a biological rationale that strengthens the case for clinical testing. Several academic studies have already demonstrated AI-driven repurposing in areas including oncology, neurology, and infectious disease, but no company has yet scaled the approach to industrial levels.
Wang’s startup will need to demonstrate that its AI models can consistently identify repurposing candidates that succeed in clinical trials — a high bar that will require not only technical excellence but also partnerships with pharmaceutical companies, contract research organizations, and regulatory agencies.
The Road Ahead: Opportunities and Challenges
Wang’s startup enters a field that is growing rapidly but also facing significant hurdles. The AI models used in drug discovery must contend with noisy, incomplete, and heterogeneous data. Biological systems are profoundly complex, and even the most sophisticated models can miss critical interactions or produce false positives. Regulatory pathways for AI-discovered drugs remain unclear in some jurisdictions, and pharmaceutical companies have been slow to adopt computational approaches for core R&D decisions.
Nevertheless, the pace of investment and the caliber of talent entering the space suggest that AI-driven drug discovery is not a passing trend. With major players like Isomorphic Labs, Chai Discovery, and now Wang’s unnamed startup all backed by hundreds of millions of dollars, the sector has reached a critical mass of capital and capability. The next few years will reveal whether the promise of AI-accelerated drug development translates into approved therapies reaching patients. For now, the convergence of machine learning and molecular biology represents one of the most consequential developments in both technology and medicine — and Miles Wang’s move from OpenAI to the forefront of that convergence marks another milestone in its evolution.
