Novo Nordisk and Anthropic announced a collaboration on September 16, 2026, to test Claude Science in selected pharmaceutical research-and-development workflows. The companies also plan to use Anthropic’s frontier models for Novo’s AI-driven software development. The initiative is designed as a governed pilot, with human oversight and data-governance measures included in the collaboration.
What the initial pilot is supposed to do
Claude Science is planned for selected R&D workflows and scientific problems identified by Novo Nordisk’s scientists and computational teams. The stated focus is drug discovery, biological reasoning, and the development of new medicines.
That scope is narrower than a company-wide rollout. Novo Nordisk says the technology will first be tested where the two companies expect the greatest joint impact, rather than deployed across every research operation at once.
Mike Doustdar, president and CEO of Novo Nordisk, said AI could help increase R&D productivity, shorten the path from research to a marketed product, and support new ways of understanding human biology and drug mechanics. Those are the collaboration’s intended benefits; the announcement does not report a medicine, clinical benefit, or measured acceleration resulting from Claude Science.
Dario Amodei, co-founder and CEO of Anthropic, described a much broader ambition: AI could potentially compress a century of biological and medical breakthroughs into a decade. That is a forecast about the possible long-term effect of increasingly capable AI, not an outcome from this pilot.
The agreement also reaches software development
The collaboration is not limited to laboratory and drug-discovery work. Novo Nordisk also plans to use Anthropic’s frontier models to strengthen AI-driven software development and help scale AI across the company.
That second track matters because pharmaceutical AI depends on more than scientific models alone. Research teams also need software and data systems that can support analysis, automation, and collaboration. The announcement places that infrastructure work alongside the planned Claude Science testing.
Governance and human oversight are part of the plan
Novo Nordisk says the collaboration includes robust data governance and human oversight aligned with its ethical and compliance standards. For pharmaceutical research, those controls are central: the work can involve sensitive scientific information, and AI-generated reasoning still sits inside a process led and reviewed by people.
The companies have not presented the pilot as an autonomous drug-development system. Its announced role is to support selected research workflows and software development while Novo Nordisk retains oversight of how the work is used.
Novo already had experience with Anthropic models
The new collaboration builds on an earlier Novo Nordisk workflow involving Anthropic’s Claude 3.5, MongoDB, and AWS. In an April 2025 presentation, Louise Skov, Novo Nordisk’s head of content digitalization, said the tools were used to generate and structure clinical-study-report documents.
Skov said that process moved from weeks to minutes and was 90% faster. That figure describes the earlier document-production workflow; it is not a result from the Claude Science R&D pilot.
The earlier use also shows the distinction between the two efforts. Claude 3.5 supported document production, while the new collaboration is aimed at selected scientific R&D problems, biological reasoning, and AI-driven software development.
What the announcement means for pharmaceutical AI
Novo Nordisk’s plan combines a targeted research trial with a broader effort to expand AI software inside the company. The immediate development is the planned testing of Claude Science—not a new drug, a regulatory decision, or a demonstrated change in patient care.
The next substantive milestone will be the results of the selected workflows as Novo Nordisk and Anthropic put the collaboration into practice.
