Miles Van, former OpenAI researcher, has launched an AI-powered drug discovery startup with a $2 billion valuation. The venture focuses on repurposing existing medications, potentially cutting drug development timelines from 10-12 years to 3-5 years while reducing R&D costs by 30-50%. Industry sources indicate advanced partnership talks with major pharmaceutical companies.
At a glance: Key facts
- Founder Miles Van led OpenAI's scientific discovery acceleration research
- Specializes in AI-driven drug repurposing for approved medications
- $2B valuation targeting $200M Series A funding
- Lightspeed Venture Partners likely lead investor
- Team includes OpenAI/DeepMind alumni and FDA regulatory experts
- Processes 10,000 molecular interactions daily vs. traditional 100
- Potential to slash drug development costs by $1.6B per treatment
- Market context: Competitors Chai Discovery ($3.8B) and Isomorphic Labs ($2.1B)
The brains behind the operation
Van joined OpenAI in 2024 after Harvard CS studies, pioneering AI applications for biological discovery. His work included molecular interaction prediction algorithms and biological data analysis. Recognized as MIT Technology Review's "35 Innovators Under 35" in 2025, Van's approach blends deep learning with computational biology.
Dream team assembly
The startup combines:
- ML experts from OpenAI/DeepMind
- Pharma-industry bioinformaticians
- Clinical pharmacologists
- FDA regulatory specialists
Tech stack and methodology
The platform analyzes FDA-approved drugs for new therapeutic applications using:
- Graph neural networks for molecular analysis
- Transformers processing scientific literature
- Diffusion models generating molecular configurations
- Federated learning for secure medical data handling
Five core applications:
- Therapeutic effect discovery for existing drugs
- Drug-target interaction prediction
- Clinical trial patient pre-screening
- Pharmacokinetic/toxicology modeling
- Regulatory documentation automation
Investment landscape
AI pharma funding hits record highs:
| Company | Founded | Latest Round | Valuation | Focus |
|---|---|---|---|---|
| Chai Discovery | 2024 | $400M | $3.8B | Oncology |
| Isomorphic Labs | 2021 | $2.1B | N/A | Protein interactions |
| Van's startup | 2026 | $200M (target) | $2B | Drug repurposing |
McKinsey projects AI could generate $60-110B annual pharmaceutical value by 2030.
Challenges and risks
- 85% current prediction accuracy ceiling
- Pharma industry integration hurdles
- FDA AI validation requirements
- Algorithmic IP protection complexities
- $500k+/month cloud computing costs
Regulatory gray areas
- AI validation standards under development
- Liability for algorithmic errors
- Results reproducibility requirements
Industry disruption potential
Successful drug repurposing examples:
- Thalidomide (morning sickness → myeloma)
- Sildenafil (hypertension → ED)
- Rituximab (lymphoma → arthritis)
Questions & answers
What's Van's AI background?
Van pioneered transfer learning for biological tasks at OpenAI, developing protein-drug interaction prediction systems comparable to AlphaFold.
Who leads the AI drug discovery space?
Chai Discovery ($3.8B) and Isomorphic Labs ($2.1B) dominate, with 50+ niche players emerging since 2026.
Why are investors bullish?
3-5x average ROI potential from accelerated development cycles and personalized medicine demand.
First product timeline?
2-3 years for repurposed drugs, 5-7 years for novel molecules, likely in oncology/neurodegenerative fields.