Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →Novo Nordisk describes AI as a partner to its researchers: it helps search virtual molecules, guide molecular design, and make sense of clinical-trial data, while laboratory experiments and scientific teams remain part of the process. The company says its researchers assessed one billion virtual molecules and screened approximately 2,500 compounds in the lab; those are company-reported discovery figures, not proof that AI has produced an approved medicine.
Where AI fits in Novo Nordisk’s research workflow
Novo Nordisk’s 2024 Annual Report presents AI as part of a broader research workflow, not a substitute for laboratory science. Computational screening and predictive molecular design can help narrow candidates for experiments. Predictive pharmacology and knowledge mining can inform subsequent design cycles. In clinical research, data analysis can support patient stratification, trial design, site selection, and forecasting outcomes.
As an Amazon Associate I earn from qualifying purchases.
These are different tasks at different points in development. A model’s prediction or a pattern found in data can help researchers decide what to investigate; it does not by itself establish that a candidate is safe or effective in people.
What the company reports from molecule discovery
Novo Nordisk says its researchers combined AI with high-throughput experimentation to assess a very large virtual set, then screen a much smaller number of compounds in the laboratory. The annual report states: “By combining AI with high-throughput experimentation, we have assessed one billion virtual molecules via computer modelling and screened approximately 2,500 compounds in the lab.”
#1 Best Overall
The company also reports that this work led to a highly selective amylin compound that closely mimics the natural hormone, with 50–75% fewer design rounds. That is a discovery-stage result reported by Novo Nordisk. Fewer design rounds do not demonstrate shorter clinical development, regulatory approval, or patient benefit.
How AI is being used with clinical-trial data
The annual report says Novo Nordisk harmonized data from around 1,600 clinical trials, including SELECT and STEP. Harmonizing data means making information from different trials more consistent for analysis. The company says the resulting dataset is intended to support disease insights, patient stratification, and drug-target identification.
Rank #2
This is a way to generate and examine research questions across accumulated trial data. The reported number does not mean AI independently validated a treatment or that every trial dataset was used for every stated purpose.
Free tools Windows power users keep installed
One-click scans. No signup required.
Valo Health: using human data to pursue cardiometabolic targets
Novo Nordisk and Valo Health expanded their collaboration in an announcement dated 8 January 2025. The work covers obesity, type 2 diabetes, and cardiovascular disease, drawing on Valo’s human dataset and AI-powered computation. The companies reported that several targets had been identified and multiple small-molecule preclinical programmes were underway.
Rank #3
Novo Nordisk’s annual report describes an ambition to accelerate up to 20 cardiometabolic programmes through the collaboration. “Up to 20” is a development target, not a count of approved medicines or completed programmes. The announcement establishes the companies’ reported progress and plans, not independent evidence of clinical effectiveness.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How the technology partnerships differ
The partners have distinct roles, and their announcements describe different evidence stages. They should not be read as interchangeable AI products or as equivalent demonstrations of clinical outcomes.
| Partner | Role described | What has been reported | Status and source |
|---|---|---|---|
| Valo Health | Human data, target discovery, and preclinical drug programmes | Several targets identified and multiple small-molecule preclinical programmes underway in obesity, type 2 diabetes, and cardiovascular disease; up to 20 programmes is the stated ambition. | Company announcement, 8 January 2025, and Novo Nordisk’s 2024 Annual Report: Novo Nordisk–Valo collaboration. |
| Microsoft | Cloud, data science, and collaborative AI research | The 2022 collaboration covered scientific-literature analysis and cardiovascular risk modelling. A 2024 Microsoft customer story described an Azure-based AI platform and said initial cardiovascular risk models had been published. | Initial results as described by Microsoft, not an independent evaluation: Microsoft customer story. |
| NVIDIA and the Danish Centre for AI Innovation | Large-scale computing and AI model tools | NVIDIA named the Gefion supercomputer and its BioNeMo, NIM, NeMo, and Omniverse technologies. Proposed research use cases included predicting cellular responses, molecule design, and biomedical language models. | Use cases described in NVIDIA’s 11 June 2025 announcement: NVIDIA announcement. |
| OpenAI | General AI capabilities and potential workforce and operational uses | The announcement described pilots across R&D, manufacturing, and commercial operations, with governance and human oversight as requirements. Full integration by the end of 2026 was a target. | Planned integration, not a completed outcome, in Novo Nordisk’s 14 April 2026 announcement: OpenAI partnership announcement. |
| AWS | Cloud and AI services, including drug-discovery use cases | The announcement described a strategic partnership and a London co-innovation hub. | Newly announced partnership in Novo Nordisk’s 10 August 2026 announcement: AWS partnership announcement. |
The Microsoft collaboration was announced by Novo Nordisk in 2022 as a partnership spanning cloud, AI, and data science: Novo Nordisk’s Microsoft announcement. Microsoft’s later customer story said initial cardiovascular risk models were published. Novo Nordisk’s Senior Vice President of Global Drug Discovery, Karin Conde-Knape, told Microsoft that using AI to define patients’ risk of developing diseases could help inform better treatments and precision medicine. That statement describes the intended value of the work; it is not evidence that a particular AI model improved patient outcomes.
For the later partnerships, the distinction between a research use case and an established result matters. NVIDIA described areas the collaboration could investigate; OpenAI described pilots and a future integration goal; AWS described a newly announced partnership and hub. The announcements establish what the organisations say they are pursuing, not independent proof of impact.
Quick Recap
What these reports do—and do not—show
- Reported research figures: Novo Nordisk’s annual report provides the molecule-screening, design-round, and trial-data figures. They are company-reported measures, not independently audited estimates of AI’s contribution.
- Discovery is not clinical validation: A selective compound or a candidate advanced to preclinical work still needs further research and testing. The evidence cited here does not establish an approved AI-discovered therapy or a patient benefit attributable to AI.
- Partner announcements are not outcome studies: Vendor and company descriptions can document collaborations, early results, proposed use cases, pilots, and goals. They do not all establish completed deployments or equivalent effectiveness.
- No established end-to-end time saving: The cited sources do not quantify how much AI shortened the full journey from early discovery to an approved drug.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
