Lab as a Service Unlocks Pharma R&D
Sakana AI
The big shift is from selling a model once to becoming part of the customer’s research workflow every week. A model license is just software access, but a lab service can take in a problem, generate candidate experiments, run simulations or code, rank results, and return a draft report. That maps much more directly onto how pharma teams, materials groups, and corporate R&D budgets are actually spent, which is on repeated research cycles and program support, not just on access to a base model.
-
The AI Scientist already spans most of the research loop, from idea generation to code, experiments, result summaries, visualizations, and full manuscript drafts. That makes it easier to package as an outsourced discovery engine, not just a model endpoint, especially in domains where customers care about better experiments more than about which model produced them.
-
There is a clear precedent for this market shape in scientific software. Schrödinger sells software into biopharma and also earns drug discovery revenue through collaborations that support target analysis, lead identification, and lead optimization. In 2024 it reported $180.4 million of software revenue, plus a separate collaboration driven drug discovery business. That shows how research tooling can expand into larger, higher value service relationships.
-
Sakana already sells through partnership led enterprise workflows rather than mass self serve APIs. Its multi year MUFG deal combines technology and ongoing support around a concrete workflow, document automation for banking. A lab service for pharma or materials science would use the same commercial logic, but tied to bigger R&D budgets and longer programs.
The next step is for AI vendors to move up the stack from model providers to research operators. If Sakana can show that its systems shorten experiment cycles and improve hit rates in a few real programs, it can evolve from selling specialized models into owning recurring budgets for discovery, optimization, and internal R&D automation across large enterprises.