OpenAI’s Value‑Sharing Model for Drug Discovery: What It Means for Researchers and Companies

Overview

OpenAI’s CFO Sarah Frier sparked a heated debate at Davos by hinting that the company could take a share of intellectual property (IP) generated with its AI models, specifically in the realm of medical research and drug discovery. The comment went viral, leading many to fear that any use of ChatGPT might cost a slice of their future profits.

The Reality of the Announcement

• Scope limited to pharma – The value‑sharing proposal applies only to drug‑discovery projects that use OpenAI’s compute, not to general content creation or software development.
• How it would work – A biotech firm could license a drug discovered with OpenAI’s models and give OpenAI a percentage of the resulting revenue, similar to existing partnerships between pharma companies and data‑rich platforms.
• Why OpenAI wants this – Running millions of AI agents 24/7 for drug screening is extremely expensive. By taking equity or royalties, OpenAI offsets compute costs while locking in long‑term partners.

Precedents in the Industry

1. 23andMe & GSK (2018) – The genetics company partnered with GlaxoSmithKline, sharing genetic insights in exchange for a share of future medicines. 23andMe later filed for bankruptcy in 2025, showing the risk of IP‑sharing deals.
2. University policies – Institutions like Stanford claim ownership of any patentable invention conceived with more than incidental use of university resources. This mirrors the idea that heavy reliance on a platform can trigger IP claims.
3. Big‑tech rivals – Alphabet’s DeepMind, Anthropic, and Isomorphic Labs are already discussing data‑licensing and partnership models with biotech startups, indicating a broader industry trend.

Why Pharma Might Embrace the Model

• Speed matters – In drug discovery, faster identification of candidates can mean lives saved and earlier market entry.
• Compute costs – Massive AI workloads require specialized hardware; sharing future profits can be cheaper than raising large rounds of capital.
• Investor appeal – Startups can attract funding by offering a slice of future royalties rather than diluting equity.

Potential Risks and Concerns

• Loss of control – Companies may become dependent on OpenAI’s infrastructure, limiting flexibility.
• Economic stratification – If only a few entities control the compute power, they could dominate multiple industries, creating a permanent underclass of innovators without access.
• Future of AGI – Should artificial general intelligence emerge, the holder of the compute could autonomously discover drugs, software, or new mathematics, further consolidating power.

Practical Implications for Users

• General ChatGPT users – No need to worry about giving away IP for everyday tasks, coding, or content creation.
• Researchers and startups – Must read partnership agreements carefully; consider whether a royalty‑based model aligns with long‑term goals.
• Alternative routes – Companies can still opt to pay per‑compute or seek other AI providers if they wish to retain full ownership.

The Bigger Picture

OpenAI’s move reflects a shift from pure subscription revenue to hybrid models that blend compute provision with equity stakes. While this could accelerate drug discovery, it also raises questions about who ultimately controls the most powerful tools of the future.

Bottom Line

The value‑sharing proposal is industry‑specific, not a blanket claim over all AI‑generated ideas. It mirrors existing collaborations in biotech and academia, but it also highlights the growing power imbalance that could arise if compute resources remain concentrated in the hands of a few.

Sponsored by Dell Technologies – powerful AI workstations like the ProMax series are designed for exactly these high‑compute workloads.

OpenAI’s proposed royalty model is limited to drug‑discovery collaborations, offering a way to fund massive AI compute while accelerating medical breakthroughs. However, it underscores a broader trend: as AI compute becomes a strategic asset, control over it may dictate who can innovate, potentially widening the gap between well‑funded giants and independent creators.