Ozempic maker adds Claude next to OpenAI's Rosalind

Novo already has one frontier lab inside its research: OpenAI's science model GPT-Rosalind, under a deal sealed in April. On Wednesday it added a second, announcing that it will use Anthropic's models and Claude Science to push scientific reasoning through R&D, Euronews reports.

At a glance

• Mike Doustdar, president and CEO of Novo, said joining forces with Anthropic will supercharge the R&D organisation and help the mission of bringing new health solutions to people with chronic diseases.
• Dario Amodei framed the stakes wide, saying AI's increasing capability brings the potential to compress a century's worth of biological and medical breakthroughs into a decade.
• The announcement names no first project, no timeline and no financial terms, while Anthropic's Bristol Myers Squibb deal from May 2026 at least listed the company functions Claude would cover.

If you have not followed Novo's AI buildup: in April it signed with OpenAI to use GPT-Rosalind for analysing complex datasets, identifying potential drugs and cutting the time between research and patient access. In 2024 the Novo Nordisk Foundation went in with Nvidia and Denmark's Export and Investment Fund to set up the Danish Centre for AI Innovation, which runs Gefion, the country's first AI-ready supercomputer.

Claude joins GPT-Rosalind inside the same R&D organisation

Doustdar framed the deal as more than a productivity play. AI tools, he said in the press release, open "completely new scientific opportunities" and help with reasoning about human biology and drug mechanics, on top of speeding up research work.

Amodei went to the widest framing available. Giving leading researchers access to safe, capable and trusted frontier models, he said, makes it possible to shorten research timelines, improve outcomes and find medicines that significantly improve human life.

According to The Wall Street Journal, the partnership will also be used to strengthen Novo's software development, so the scope reaches past the lab bench into the code the company writes for itself.

Novo will use Claude Science to advance scientific reasoning in R&D

Anthropic describes Claude Science as a workbench rather than a chat window. By its account, the tools a researcher normally juggles, PubMed, Jupyter, R, a terminal on the compute cluster, sit out on one bench instead of living in separate drawers.

You talk to a generalist coordinating agent that, per Anthropic, carries over 60 curated skills and connectors pre-configured for genomics, single-cell, proteomics, structural biology and cheminformatics. It can spin up specialist agents for narrow work, and, by the same description, a reviewer agent checks citations and calculations, flagging and correcting errors.

Anthropic also says every figure Claude Science generates carries the exact code and environment that produced it, a plain-language description of how it was made, and the full message history, so a result can be validated months later. Environments, again per Anthropic, get built on a laptop, on a cluster or on on-demand GPUs.

Bristol Myers Squibb took Claude company-wide in May 2026

That agreement was the broad kind: Claude across research, clinical development, manufacturing, commercial and corporate functions. Anthropic says it launched its life sciences effort last fall and has since been improving model capabilities and wiring Claude into the scientific ecosystem through MCP connections.

By Anthropic's own materials, Claude is already in use at Genentech, AstraZeneca, AbbVie, Cepheid and Evinova as well. Genentech's head of research and early development, quoted in those materials, says Claude is helping the company "reshape day-to-day scientific research and lay the foundation for the autonomous labs of the future".

Cepheid, a Danaher company, uses it for something narrower, per the same materials: drafting and refining regulatory documentation for diagnostics in development, including work tied to its Xpert Hemorrhagic Fever test in Ebola outbreak response.

The Novo announcement is thin exactly where a reader would push: no first programme, no money, nothing about how Claude and GPT-Rosalind divide work inside one research organisation. In our view that reads oddly next to the Bristol Myers Squibb deal, which at least came with the list of functions Claude would sit in.

Which programme gets Claude first

Claude Science, according to Anthropic, is in beta for Pro, Max, Team and Enterprise users and is still being refined on user feedback, so the workbench Novo's scientists sit down to may not be the one described today. The next concrete signal would be a named disease area or a first research project running on it. Novo has given no date for either.