Novo Nordisk is betting big that AI can radically compress the maddeningly slow timeline of drug development. The pharmaceutical giant is expanding its partnership with AWS, establishing a co-innovation hub in London where its scientists will sit shoulder-to-shoulder with AWS engineers to build AI agents that handle target identification, therapy design, and complex research workflows.
The goal is brutally pragmatic: shorten the path from a promising biological target to a first human dose. The new hub is designed to eliminate the tedious handoffs between computational analysis and lab work. To do this, Novo Nordisk plans to use Amazon Bio Discovery, a service that strings together more than 40 biological AI models. AI agents will coordinate these models, generating and ranking potential drug candidates before sending the best contenders for physical synthesis. The experimental results then loop back into the computational workflow, creating a tight feedback cycle that mimics a scientific flywheel.
This isn’t the company’s first AI rodeo. An existing internal platform built on Amazon Bedrock already supports 25,000 employees across 2,500 different chatbot use cases, handling over 26,000 prompts a month. One particularly striking deployment uses Anthropic’s Claude 3.5 to draft clinical documentation. A process that previously required up to 50 people and 15 weeks was slashed to minutes for a team of just three reviewers. That’s not an incremental gain; it’s a wholesale restructuring of how documentation gets produced.
The new agreement pushes generative AI deeper into core science, using Bedrock to build applications that connect clinical, genomic, and imaging data. The logic is straightforward: insights from early-stage research should directly inform how clinical trials are designed. AWS’s Dan Sheeran framed it as moving from studying bottlenecks to actually removing them. The lingering question is how the notoriously conservative world of regulatory approval will greet drug candidates designed by an orchestrated swarm of AI agents.
