Novo names AWS strategic AI partner as CagriSema decision nears

Novo Nordisk has named Amazon Web Services its preferred cloud provider and strategic AI partner, opening an AI co-innovation hub in London to accelerate drug discovery. The deal arrives as Novo shares sit about 66 percent below their 2024 peak after CagriSema failed non-inferiority against…

Novo names AWS strategic AI partner as CagriSema decision nears

Novo Nordisk has named Amazon Web Services its preferred cloud provider and strategic AI partner, and will open an AI co-innovation hub in London designed to accelerate early drug research and clinical development. The designation is a procurement commitment as well as a research statement: new AI and data workloads at Novo will default to AWS infrastructure, and the AWS model platform becomes the company's main route into foundation-model-based drug discovery. The expanded alliance is the company's most explicit answer yet to a prolonged share price decline: it pairs the maker of Wegovy and Ozempic with Amazon's cloud unit at a moment when both companies are trying to show that generative AI can change pharmaceutical economics.

Financial terms were not disclosed, so the agreement cannot yet be classified as a procurement contract, a research collaboration, or both. AWS chief executive Matt Garman said the collaboration will "fundamentally change how medicines are discovered and developed." Novo Nordisk president and chief executive Mike Doustdar was blunter about the occasion: "We need to be faster and we need to learn from the failures and reinvent ourselves again."

The stock context explains why. Novo Nordisk shares closed at $47.26 on August 7, 2026, about 66 percent below the all-time closing high of $137.40 set on June 25, 2024, and about 5 percent below where they traded five years earlier. The decline tracked a string of clinical disappointments, each of which narrowed the company's claim to the next generation of obesity and cardiometabolic medicines.

The clinical setbacks that preceded the deal

The largest of those disappointments was CagriSema , a fixed-dose combination of the amylin analogue cagrilintide and semaglutide . In December 2024, the REDEFINE 1 efficacy trial reported 22.7 percent weight loss, below the roughly 25 percent investors had been led to expect. The gap was narrow in absolute terms, but its meaning was strategic: CagriSema was the asset designed to take the best-in-class narrative in obesity, and the miss handed that narrative to Eli Lilly before the head-to-head had even run.

In February 2026, the REDEFINE 4 head-to-head trial in 809 patients reported 23.0 percent weight reduction at 84 weeks against 25.5 percent for Eli Lilly's tirzepatide . CagriSema failed non-inferiority, and Novo shares closed down about 16 percent the day the results were released.

The setbacks accumulated. REIMAGINE 4 , reported this month in patients with type 2 diabetes, established non-inferiority to tirzepatide for weight reduction but failed to establish non-inferiority for HbA1c reduction. On July 31, the ZEUS phase 3 outcomes trial of ziltivekimab missed its primary major adverse cardiovascular events endpoint, with a hazard ratio of 0.99. In the second quarter, Novo booked a DKK 6.3 billion non-cash impairment after discontinuing monlunabant . The charge does not move cash; it records that the value of the discontinued program's assets was written down. Its size, however, signals an abandoned metabolic bet at a moment when the company could least afford one.

The August 5 earnings call showed where investor attention now sits. Nine sell-side analysts asked roughly 17 questions over one hour, and only two dealt with manufacturing supply and business development. The rest concerned commercial performance, and the disclosed numbers explain why: about 35 percent of U.S. injectable Wegovy use is now self-pay, up from 10 to 15 percent a year earlier. A cash-paying patient base is the most price-sensitive demand a drug company has, and its growth changes how much revenue Novo can assume from any given price point. It is also the demand channel that the set prices on Amazon Pharmacy 's Wegovy pill listing, $25 per month for commercially insured customers and $149 cash, are explicitly designed to capture.

What the AWS partnership actually contains

The centerpiece of the AWS arrangement is a co-innovation hub in King's Cross, London, staffed initially by roughly 40 employees. King's Cross sits inside the Knowledge Quarter that already hosts the Francis Crick Institute and the Alan Turing Institute, with the Wellcome Trust nearby, placing the hub within walking distance of major academic biomedical research. Novo first announced the digital innovation hub in early 2024; the AWS designation and the expanded scope are new. The two companies say the hub will compress the path from drug target to first human dose, the metric that matters most for peptide and antibody programs because it determines how many candidates a pipeline can evaluate in a given time.

The scale of the existing AI work is substantial. Novo says it has generated more than 2,500 use cases built on Amazon Bedrock and Bedrock AgentCore , its agent orchestration layer, and that a Bedrock-based generative AI tool has reached more than 25,000 employees. A use case, in enterprise AI practice, is a deployed or in-development application rather than a validated workflow, and the company has not said how many have reached production. The most concrete metric in the announcement is in documentation: in a project using Anthropic and MongoDB, the time required to create clinical study reports fell from 12 weeks to 10 minutes. Study reports are the documents that transmit trial data to regulators, and their assembly from multiple source systems is a known bottleneck in submission timelines. A reduction of that size does not remove the need for human review, but it removes weeks from the critical path between a completed trial and a regulatory submission.

The arrangement reaches beyond the cloud unit. Amazon chief executive Andy Jassy leads the parent company whose health businesses, alongside AWS, include Amazon Pharmacy and Amazon One Medical , and Amazon and Novo describe the relationship as spanning all three. The established track record between the two companies is in drug delivery, not drug discovery: Amazon Pharmacy began carrying the Wegovy pill in January and added oral Ozempic and Wegovy to same-day delivery in May. Doustdar described the purpose of the AI work as "accelerating discovery and developing responsible AI solutions that will create real progress for people living with serious chronic diseases" and, in a phrase that captures the portfolio logic, "Have more shots at goal."

AWS is building for this market. The company says 19 of the top 20 pharmaceutical companies use AWS for generative AI and machine learning. On April 14, AWS launched Amazon Bio Discovery , a service offering access to more than 40 biological foundation models; the healthcare and life sciences unit runs under vice president Dan Sheeran. In June, AWS created its Forward Deployed Engineering organization with a $1 billion commitment, an operating model in which AWS engineers work inside customer organizations rather than delivering from a distance. Novo's expanded AWS agreement followed a partnership with OpenAI announced on April 14, the same day as the Bio Discovery launch, and roughly a week later Merck announced a $1 billion AI-focused partnership with Google built around the Google Co-Scientist AI platform. Novo had opened its AI era earlier: in 2022 it announced a drug discovery collaboration with Microsoft, and Denmark's Gefion supercomputer, based on an NVIDIA DGX SuperPOD and supported by the Novo Nordisk Foundation and the Export and Investment Fund of Denmark, gives the country a national AI compute asset outside any single commercial cloud. On the Novo side, enterprise IT and quality runs under executive vice president Thilde Hummel Bøgebjerg.

What the phase 3 trials can and cannot show

The evidence behind the CagriSema decision rests on phase 3 trials with different designs and different questions. REDEFINE 1 tested efficacy in a population with obesity or overweight. REDEFINE 4 tested non-inferiority against tirzepatide in 809 patients, with results measured at 84 weeks. REIMAGINE 4 compared the same two drugs in type 2 diabetes. ZEUS was an outcomes trial in a cardiovascular risk population the company has not further specified.

Non-inferiority designs answer a narrower question than superiority trials: whether a drug is unacceptably worse than the comparator, not whether it is better. The comparison requires a prespecified margin, and the margins for these trials have not been disclosed. Failing non-inferiority on weight in REDEFINE 4, at 23.0 percent versus 25.5 percent, means the trial could not exclude a meaningful disadvantage relative to tirzepatide. It does not prove inferiority; it means the data cannot rule it out within the chosen statistical standard. The split result in REIMAGINE 4 is harder for clinical positioning: non-inferior on weight, but a miss on HbA1c, the endpoint prescribers and payers in type 2 diabetes weigh first. If Novo seeks a type 2 diabetes indication, glycemic control is the registration endpoint that historically anchors diabetes drug approval, and a trial that fails non-inferiority on that endpoint leaves the label argument dependent on the rest of the evidence package.

ZEUS is the starkest result. A hazard ratio of 0.99 is statistically indistinguishable from no effect on the primary endpoint. The point estimate sits essentially on the null; what remains unknown is the confidence interval, which determines whether the result excludes a modest benefit, a modest harm, or both. Confidence intervals and full statistical details have not been reported for any of these trials, which limits how precisely the effect sizes can be judged.

The regulatory sequence: an approval, a pending decision, and a filing question

The regulatory frame for this franchise began on June 4, 2021, when the FDA approved Wegovy injection for obesity. That approval covered the injectable peptide. The oral version reached patients through a different channel: Amazon Pharmacy began carrying the Wegovy pill on January 9 at $25 per month for commercially insured customers and $149 as a cash-pay price.

The next milestone is an FDA decision on CagriSema expected late this year. That expectation is a forecast, not a confirmed event, and the indication under review has not been specified. Chief financial officer Karsten Munk Knudsen said he remains optimistic ahead of the decision. R&D chief Martin Holst Lange said CagriSema still has the potential to be the first GLP-1/amylin combination to reach market, a position that matters because the first drug in a new combination class sets the efficacy reference point that later entrants are measured against. A separate open question, raised during the August 5 call, is whether Novo will file the 50 mg dose of oral semaglutide.

Competition tightens the timeline. Eli Lilly launched oral orforglipron as Foundayo on April 9, placing an oral small-molecule GLP-1 agonist against Novo's oral peptide. In May, oral Ozempic and Wegovy were added to Amazon Pharmacy same-day delivery. The sequencing means that by the time the FDA rules on CagriSema, the oral small-molecule challenger will have been on the market for most of a year.

The biology: GLP-1, amylin, and the oral peptide problem

Semaglutide is a long-acting peptide agonist of the GLP-1 receptor. It potentiates glucose-dependent insulin secretion, suppresses glucagon, and acts on central circuits to reduce appetite and food intake. The glucose dependence of its insulinotropic effect is what gives the class its relatively low hypoglycemia risk compared with insulin. Cagrilintide is a long-acting analogue of amylin, the pancreatic peptide co-secreted with insulin that complements GLP-1 by slowing gastric emptying and reinforcing satiety through hindbrain pathways centered on the area postrema. The combination rationale is complementarity: two anorectic signals from different receptor systems in a single injection, with the…

Peptides referenced: Semaglutide, Tirzepatide, Cagrilintide, Amylin, Orforglipron, Glucagon, GLP-1.

Related reading: GLP-1 RA Plus Platinum: GI Toxicity Linked to Higher Mortality, Blow fly larva peptide active against MRSA with low cell toxicity, IndexBox Peptide Reagents Page Lists Other Markets, No Data, Khavinson bioregulator peptides: nuclear claims meet a clinical gap.