LONDON–The most expensive part of drug development is the beginning: screening millions of compounds, predicting how they will behave, and deciding which few deserve the decade and billions of dollars required to reach patients. Novo Nordisk, the Danish pharmaceutical giant, has decided to attack that stage with artificial intelligence, and on Aug. 10 it announced a strategic partnership with Amazon Web Services to build an AI innovation center in London focused on drug discovery.
The arrangement goes beyond the typical cloud contract. Novo Nordisk will house research teams alongside AWS engineers in a joint facility, working on AI models for target identification, compound design, and the prediction of clinical outcomes. The companies described the center as a co-development effort rather than a vendor relationship, a distinction that reflects how the pharmaceutical industry’s use of cloud computing has evolved.
Drug companies have bought computing power from cloud providers for years, using AWS, Microsoft Azure, and Google Cloud as storage and processing backbones. The new generation of partnerships is different: pharma companies are asking cloud providers to help build the AI itself, not just supply the machines it runs on. The Novo Nordisk deal is among the largest examples of that shift.
For AWS, the partnership is a foothold in one of the most valuable markets for AI: drug discovery. The cloud provider has been expanding its life-sciences business, and executives have said that healthcare and life sciences are among the fastest-growing segments of its portfolio. Winning Novo Nordisk’s AI infrastructure business positions AWS to capture the wave of pharmaceutical AI spending expected over the next several years.
For Novo Nordisk, the deal is part of a broader push into AI-led research. The company, best known for its obesity and diabetes treatments, has been building internal AI capabilities and forming partnerships across the technology industry. The London center is designed to bring large language models into the earliest stages of discovery, where decisions about which targets to pursue and which molecules to test are made.
The science is evolving quickly. AI models trained on chemical and biological data can now propose molecular structures with predicted properties, and some companies have reported that AI-designed compounds reached clinical trials faster than conventionally developed ones. The technology is not yet a proven replacement for the lab, but it has moved from curiosity to core tool in much of the industry.
The partnership’s location carries its own logic. London has become a hub for AI and for biomedical research, with a dense cluster of universities, startups, and established pharmaceutical operations. The joint center will draw on that talent pool, and both companies said they plan to hire researchers and engineers locally.
The financial terms were not disclosed. People familiar with the arrangement said the deal is structured as a multi-year commitment, with AWS providing compute, model development support, and engineering capacity while Novo Nordisk contributes its data, domain expertise, and research teams. Such partnerships typically run into the hundreds of millions of dollars over their terms.
The competition is crowded. Novo Nordisk’s rivals, including large pharmaceutical companies and a wave of AI-native drug startups, have struck similar deals with cloud providers and specialized AI companies. The question facing all of them is the same: whether AI genuinely accelerates drug discovery enough to justify the investment, and which organizational structure captures that value.
Early evidence is encouraging but incomplete. Several AI-designed drugs have entered human trials, and partnerships between pharma and technology companies have produced an expanding pipeline of candidates. But drug development fails at every stage, and the industry’s leaders caution that AI improves the odds of success rather than guaranteeing it.
For AWS, the deal is also a defensive move. Microsoft has partnered with major health systems and pharmaceutical companies, and Google Cloud has its own life-sciences initiatives, including AI models for protein structure. The cloud market for drug discovery is becoming a three-way contest, with each provider assembling partnerships to secure the workloads of the industry’s biggest spenders.
The Novo Nordisk center is expected to begin operations in phases, with early projects focused on models for target identification and molecule design, according to people familiar with the plans. The companies said they will publish results of the collaboration over time, a commitment that reflects the scientific community’s demand for transparency in AI-driven discovery.
The deeper significance of the deal is what it says about the pharmaceutical industry’s direction. A company that built its fortune on traditional R&D, hiring chemists and biologists and running clinical trials for decades, is now betting that software will decide which drugs get made. The London center is a small piece of that bet, but it is a clear one: the next blockbuster, Novo Nordisk is saying, may be found by an algorithm before it is found by a scientist.
