Novo Nordisk Doubles Down on AI Drug Discovery With Anthropic Deal

Novo Nordisk, the Danish company behind Ozempic and Wegovy, is in the middle of the most consequential identity shift in its century-long history: it wants to become, in its own words, the world's most AI-driven healthcare company7. In the span of a single year, the drugmaker has struck two major partnerships with frontier AI labs — first with OpenAI in April 2026, then with Anthropic in September 2026 — with the explicit aim of reengineering how it discovers and develops medicines8. The moves signal that the company that rode the GLP-1 wave to pharma superstardom now believes its future depends on algorithms as much as on molecules.

Two deals, one strategy

The OpenAI partnership, announced April 14, 2026, is deliberately sweeping. It covers not just drug discovery but manufacturing, supply chain, distribution, and commercial operations, with pilot programs launching immediately and full integration targeted for the end of 202691117. CEO Mike Doustdar framed the ambition in terms of scale: integrating AI into everyday work gives the company the ability to analyze datasets at a scale that was previously impossible, identify patterns researchers could not see, and test hypotheses faster than ever614. On the science side, the partnership is meant to help identify promising drug candidates and shorten the notoriously long path from research bench to patient12.

The Anthropic deal, announced in mid-September, is narrower and more pointed at the lab itself. Novo will use Anthropic's Claude models and a product called Claude Science to advance scientific reasoning inside its R&D organization, tackling drug-discovery challenges identified by its own scientists and computational teams, supporting biological reasoning, and accelerating the discovery of new medicines410. Claude Science is designed to help researchers analyze literature and execute multi-step research tasks, producing detailed artifacts that can be iteratively refined toward publication4. Doustdar said the collaboration would supercharge the R&D organization and help bring transformative health solutions to people living with chronic diseases10. Anthropic's CEO Dario Amodei went further, arguing that AI's increasing capability carries the potential to compress a century's worth of biological and medical breakthroughs into a decade10.

Reading the two deals together, the picture is clear: OpenAI is the enterprise-wide transformation layer — discovery through distribution — while Anthropic is the focused scientific-reasoning instrument placed directly into researchers' workflows. Coverage of both arrangements is consistent on that division of labor, though some outlets emphasize the commercial and manufacturing scope of the OpenAI deal while others foreground the discovery science12.

The elephant in the room

The urgency behind these deals is not subtle. At Novo's investor day on Monday, September 21, Doustdar himself called semaglutide's looming loss of exclusivity the "elephant in the room," acknowledging bluntly that the patent cliff on the company's twin blockbusters is what is on most people's minds — and rightfully so1. Investors reacted harshly, sending the company's American depositary shares down nearly 8 percent in their steepest one-day fall in more than six months1.

That context explains the timing. Novo has fallen behind Eli Lilly in the immensely lucrative weight-loss drug market, as Reuters put it, and the competitive pressure spans everything from next-generation injectables to the oral obesity-drug battle that intensified after the FDA approved Novo's oral weight-loss product in late 202596. When your two flagship products face both a fierce rival today and generic erosion in the near future, compressing the drug-discovery cycle stops being a nice-to-have and becomes an existential requirement. The company's earlier, quieter use of AI in early-stage obesity and diabetes research — disclosed on a quarterly earnings call and reported in 2024 — has now escalated into board-level strategy35.

Notably, the company has been careful about the workforce question. Reuters reported that Doustdar said the OpenAI partnership aims to boost staff productivity rather than cut jobs, and both companies emphasized structured upskilling and AI literacy across Novo's global workforce as a core deliverable161118. That framing matters for a company headquartered in a labor market as sensitive to industrial restructuring as Denmark's.

What AI drug discovery actually promises here

Strip away the press-release language, and the real bet is about time and target selection. Drug development traditionally takes a decade or more and costs billions per approved medicine, with the overwhelming majority of candidates failing. The coverage of Novo's partnerships describes a three-part theory of change. First, AI can surface non-obvious drug targets and mechanistic signals buried in complex omics and clinical datasets, accelerating candidate nomination and decision-making. Second, AI can compress the path from research to marketed product through faster hypothesis testing49. Third — and this is the part Doustdar keeps returning to — AI tools can offer what he calls completely new scientific opportunities, aiding reasoning and understanding of human biology and drug mechanics in ways that might not just speed up existing processes but open questions nobody was asking410.

That third claim deserves scrutiny. Productivity gains — faster literature review, better dataset analysis, smarter supply chains — are plausible and already visible across the industry. But "completely new scientific opportunities" is a stronger claim, and Novo has not publicly released any number of AI-discovered candidates currently in its obesity pipeline3. The industry-wide context suggests the industry-wide context: the total value of AI partnerships in pharma rose roughly 120 percent year over year between 2024 and 2025, meaning Novo is racing not just Lilly but every major drugmaker making similar bets. OpenAI, for its part, launched GPT-Rosalind, a reasoning model built to support research across biology, drug discovery, and translational medicine, around the same week as the Novo deal — a sign the lab sees pharma as a core vertical18.

The honest reading is that Novo is buying optionality across two competing AI stacks rather than betting everything on one model provider. That is a sensible hedge in a field where model capabilities shift quarter to quarter, and it mirrors how other large enterprises are approaching frontier AI procurement.

Why this matters beyond Novo

The Novo deals are a data point in one of the most consequential industrial questions of this decade: whether frontier AI models — chat-style reasoning systems like Claude and GPT, rather than the physics-based molecular simulation tools that dominated the first wave of AI drug discovery — can genuinely accelerate the discovery of medicines. Most of the industry's earlier AI-discovery efforts centered on structure prediction and generative chemistry. What Novo is testing with Claude Science is different: AI as a scientific collaborator that can reason through multi-step research problems, synthesize literature, and help scientists iterate toward conclusions45.

For patients, the stakes are concrete. Demand for obesity medicines has become one of the biggest stories in the pharmaceutical industry, and Novo and Lilly are racing to build deeper pipelines of next-generation treatments — pills, combination therapies, and drugs with fewer side effects or better muscle preservation35. If AI genuinely shortens the earliest stages of research, the beneficiaries are the millions of people living with obesity and diabetes who need options beyond the current generation of drugs, a point Doustdar made explicitly when the OpenAI deal was announced6.

There is also a competitive-geography angle. Reuters noted the openAI partnership comes as Novo aims to regain market share from Eli Lilly as the oral weight-loss battle heats up16. A Danish company enlisting two American AI labs as core R&D infrastructure is a small but telling marker of where the strategic technology in drug development now sits.

The verdict

My reading: these deals are less a guarantee of AI-discovered blockbusters than a rational, well-timed response to a dual squeeze — a patent cliff on one side and a faster, better-capitalized rival on the other. Doustdar's candor about semaglutide's loss of exclusivity1 tells you the company knows the clock is running. The OpenAI and Anthropic partnerships give Novo two of the best available tools for making the discovery engine faster and, potentially, smarter. Whether "a century of breakthroughs in a decade" — Amodei's framing10 — proves out will take exactly that kind of decade to verify. But among large pharmas, Novo has moved from the middle of the AI pack to its leading edge, and that alone reshapes how the next phase of the obesity-drug race will be fought.