The announcement was subtle, dropped during a panel at Davos, but the implications sent tremors through the startup ecosystem: OpenAI is exploring a "value sharing" model, demanding a piece of the intellectual property (IP) generated by customers using its AI technology for scientific breakthroughs. This proposal moves beyond typical API usage fees and directly into profit-sharing, signaling a fundamental shift in how foundational AI labs intend to monetize their immense computational power.
OpenAI CFO Sarah Friar spoke at a panel moderated by The Information CEO Jessica Lessin at Davos, suggesting that in fields like drug discovery, OpenAI could take "a license to the drug that is discovered using OpenAI’s technology." This concept of trading compute for equity, or a piece of the eventual revenue stream, represents an aggressive attempt by OpenAI to capture the exponential financial upside of the innovations their models enable, rather than settling for linear revenue growth tied merely to token consumption. It is a pivot that reframes the generative AI provider not as a utility, but as a strategic, high-stakes investor.
The cost of running massive AI models for round-the-clock research is astronomical. Drug discovery, in particular, requires massive, specialized compute (often referred to as 'agents') to analyze vast datasets and accelerate clinical timelines. Traditionally, biotech firms lacking the internal resources would raise massive capital from venture capitalists or public investors, giving up equity to afford the computational power necessary to hire these agents. OpenAI's value-sharing model cuts out this middleman entirely. By offering compute directly in exchange for a profit-sharing stake, OpenAI transforms its AI capability into a form of capital, a non-dilutive (to cash) investment whose payoff is tied directly to the success of the discovery.
The initial reaction from many in the tech community was skepticism, questioning what company would ever agree to such a demand. Yet, the model is not without precedent, particularly in sectors where speed and resources are paramount. The video points to the 2018 partnership between pharmaceutical giant GSK and 23andMe, where genetic insights were leveraged for novel medicine development. Furthermore, aggressive IP capture in exchange for access to resources is already a normalized, albeit often controversial, practice in deep research environments. Elite academic institutions often employ stringent patent policies, such as those at Stanford, which claim ownership of "potentially patentable inventions created in the course of an individual’s university responsibilities, participation in a research project at Stanford, or with more than incidental use of Stanford resources." This institutional model sets a clear precedent for centralizing IP rights around the provider of core resources—in this case, computational power.
For drug discovery companies facing long timelines, immense regulatory hurdles, and high failure rates, the ability to accelerate research using cutting-edge, proprietary AI could outweigh the cost of surrendering a portion of future profits. The alternative is slower growth, reliance on traditional, slower fundraising cycles, and potentially being outpaced by competitors who embrace AI-driven acceleration. In the high-stakes world of life sciences, moving quickly to cure a disease is often the best—and safest—bet, despite the equity cost.
This strategy is inextricably linked to the concept of Artificial General Intelligence (AGI) and the scarcity of compute. The race to secure high-value IP using scarce compute sets the stage for a dramatic concentration of wealth and capability. This concern is not limited to OpenAI; its rivals, including "Anthropic, Google DeepMind and Isomorphic Labs, an Alphabet subsidiary focusing on using AI for drug discovery, have also held discussions with early-stage biotechnology startups about data licensing or partnerships." The current scarcity of advanced GPUs and the massive capital required to train and run frontier models means that the entities controlling the compute infrastructure hold immense leverage. By leveraging this scarcity to acquire IP stakes, these foundational labs are positioning themselves to capture the full economic value of AGI-driven breakthroughs across multiple industries, solidifying their dominant position.
The immediate effect of this value-sharing model is clear: intellectual property is rapidly becoming the currency exchanged for access to the most powerful engines of discovery. For early-stage biotech companies lacking the massive capital reserves necessary to compete in the computational arms race, this agreement might be a strategic necessity to accelerate their work. The long-term implication, however, is the centralization of innovation and economic power around the few organizations capable of providing and controlling the world’s most advanced computational resources.
