Mihir Rao

If biology needs AI, biology needs data.

Reductionism has prevailed in the life sciences because progress would have been intractable without it. Identifying reliable subsystems, discerning their interactions, and rationalizing interpretations of causality has brought us far. Reductionism also dictates which features of the world we choose to measure—what data we collect. This approach to thinking is not unique to the life sciences, with the financial services and technology industries having broadly applied systems thinking over the same period of time. Across these industries, the limitations of systems thinking have been clear. Even in its quantitative-first era, biology speaks in the language of probabilities and randomness—Latin for “my model of the world is informationally-incomplete.”

This is cause for concern and motivation because it indicates, at minimum, a partial incompatibility between our objectives and the information frameworks in which we hope to achieve them. Some diseases offer a tractable chain of causality along which we can engineer our objective of eradicating them. Other grander objectives encompass vastly more complex mechanics that have been pressure testing our models of life. This is not an argument against reductionist thinking, but rather the statement that we have yet to fully explore the phase space of even our current reductionist framework. As a result, it is simultaneously true that we have both made tremendous progress in understanding biology and that we have left much room for improvement. We have made meaningful progress on many surrogate objectives, created biological and clinical frameworks that enable these heuristics, and have designed our industry around this way of work. Better, more elegant solutions, however, are still out there.

The purpose of AI in the life sciences is not to discern knowledge beyond the scope of today’s reductionism, but rather to enable efficient coverage of the reductionist phase space that remains unexplored. If AI is the necessary path forward, then consider the scaling laws. We get much of the algorithms and compute for free from broader progress in research and technology, but biological data remains a bottleneck. Who are the life science industry’s data generators? While academic labs and hospitals offer unique data, their ability to support the quality and volume demands of early AI scaling is currently bottlenecked by limited infrastructure and tight regulation. In-house experimental capabilities can generate a lot of first-rate data, but biopharma has a bias to spending on external data vendors (CRO/CDMO) over high-capex internal builds. Exceptions include the new breed of techbios that are AI and computation-native and thus choose to invest in their own high-throughput data factories.

Whether or not data generators will horizontally expand across data modalities or remain highly focused in the long run is unclear. A conservative projection is one where increasing a company’s scope and volume of data generation lowers the likelihood of them successfully operating other business functions (pipeline, model training, etc.). More data, however, can offer a company higher scientific alpha. This tradeoff suggests the possibility of a spectrum of biodata generator archetypes in the coming years. On one end, they can operate as a standalone vendor, focusing purely on data generation. Moving toward the other end of the spectrum, data generators might exist within the scaffolding of a pipeline and/or model company. The latter is likely to be home to one or a few closely-related data types, while the former can choose to remain focused or horizontally-integrate across assays. In parallel, more companies will solve meta problems around the data generation process. Biodata management solutions will continue to expand the set of assays they support to keep pace with the science. Marketplaces will help match data vendors with data demand. Experimental infrastructure developers will continue to serve data generators with off-the-shelf solutions (they might even generate data themselves). In a landscape without a trivial future trajectory guided by fundamental realities of the science or industry mechanics, a strong few players can shape the market.

A relevant and significant unknown is the question of how biodata will be valued. Pricing structures may need to be developed de novo beyond traditional data licensing agreements or transactions between a data generator and model provider (in which case they can refer to the LLM data vendor playbook). The economics of biodata will inevitably impact the incentive structures around starting this kind of company, so pricing needs to make sense if we want biodata scaling to persist.

If you have an opinion on how things might play out, feel free to get in touch (mihir.rao@twosigmaventures.com). If you are actively building in this space, please do the same. I’d be happy to help talk through ideas and find the few key people that could help you drive this market.