Big Pharma Teams Up With AI Titans as the Race for “AI Bio” Innovation Intensifies

Beaten Back by Mounjaro, Wegovy Bets on an OpenAI Alliance
Roche and Eli Lilly Bring Nvidia Supercomputers Online
AI Upends the Drug Industry as Bio-AI Convergence Accelerates

Global pharmaceutical companies are accelerating “AI alliances” that fuse their proprietary data with the technology of AI hyperscalers. Danish drugmaker Novo Nordisk has adopted OpenAI’s technology to speed up drug development, while U.S. pharmaceutical group Eli Lilly has forged partnerships with multiple AI companies and is integrating AI across all of its drug programs. As the use of AI rapidly expands from identifying drug candidates to clinical design and production-process optimization, the competitive landscape of the pharmaceutical industry is also being reordered around technology.

Novo Nordisk Reaches Strategic Partnership With OpenAI

On April 14, Novo Nordisk announced that it had entered into a partnership with OpenAI to apply AI across drug-candidate discovery, production-process optimization, supply-chain management and commercial operations. The company plans to begin with a pilot application in research and development and then extend the technology to production, distribution and other areas, with the goal of deploying it across all divisions by the end of this year.

With the partnership now in place, Novo Nordisk is expected to gain the ability to apply advanced AI capabilities to analyze complex datasets, identify promising drug candidates and shorten the time required to move from research to the point at which new medicines are actually delivered to patients. Safeguards have also been put in place to ensure ethical use and transparency in deploying the technology. The two companies said they would establish stringent data-protection systems and operational governance guidelines, while designing the process so that expert oversight is maintained throughout to prevent misuse.

Speaking that day, Novo Nordisk Chief Executive Officer Maziar Mike Doustdar said, “We believe this will help us discover new drug compounds for millions of people living with obesity and diabetes.” He added, “As we introduce AI, we will validate new and more efficient patterns of work.” OpenAI CEO Sam Altman said, “As AI begins reshaping industries and the life sciences, it can help people live better lives and enjoy greater longevity,” adding, “Through our collaboration with Novo Nordisk, we expect to support scientific discovery, smarter global management and the reshaping of the future of patient care in many ways.”

The collaboration is widely seen as an extension of Novo Nordisk’s “AI alliance” with Nvidia last year. In June last year, Nvidia signed an agreement with Novo Nordisk to work on building customized AI models for drug development. Centered on Gefion, Denmark’s national supercomputer, the two companies are pushing pharmaceutical R&D innovation by combining advanced computing technologies. Under a Gefion usage agreement between Novo Nordisk and DCAI, Denmark’s AI infrastructure operator, Nvidia participates as the technology partner. The companies plan to build a research environment that applies AI to drug-candidate generation, drug-response prediction and protein-structure simulation, while also developing customized workflows based on generative AI and agentic AI.

Full-Scale AI Adoption Across the Business as Leadership in Obesity Drugs Wavers

The reason Novo Nordisk is devoting such extensive resources to its AI strategy is to regain leadership in the obesity-treatment market. The company has already ceded part of its market share to Eli Lilly, which developed the GLP-1 obesity drug Mounjaro. According to pharmaceutical data analytics platform BRP Insight, Mounjaro’s market share stood at 70.3% in January, while Wegovy’s share plunged to 21.1%.

As recently as last year, the picture was reversed. In diabetes and obesity drugs, Novo Nordisk had held the upper hand. Both Novo Nordisk and Eli Lilly have sold insulin for diabetes treatment since 1923, but Novo Nordisk had demonstrated stronger specialization in the field. In 1978, Novo Nordisk became the first company in the world to successfully produce human insulin using E. coli, and in 1985 it launched the world’s first pen-shaped insulin injector, securing the top market share position. In 2014, it also launched Saxenda, the first obesity treatment in the world based on a GLP-1 analogue, and reinforced its market dominance in 2021 with Wegovy.

The turning point came late last year. In December, Novo Nordisk released clinical results for its next-generation obesity drug CagriSema, but the outcome fell short of market expectations and triggered a sharp drop in its share price. At the same time, Wegovy continued to suffer from supply shortages, while Eli Lilly’s orally administered obesity drug moved ahead in clinical development, undermining Novo Nordisk’s position. Wegovy is injected once a week, whereas Eli Lilly’s oral obesity drug candidate is taken once daily in pill form. Analysts say oral medication could significantly expand demand for obesity drugs because it carries lower administration costs and offers greater convenience. In a recent report, BMO Capital Markets said, “Eli Lilly appears likely to overtake Novo Nordisk in the diabetes and obesity drug market in the near term.”

The industry attributes Eli Lilly’s ability to produce faster and stronger clinical results in oral obesity-drug development to its open AI strategy. Over the past two to three years, Eli Lilly has been running around 100 drug-development projects simultaneously using AI platforms. To do so, it has signed a series of partnerships with outside companies including OpenAI, XtalPi, Genetic Leap and Atomwise. Through these alliances, the company is understood to have used AI to move faster and with greater precision across clinical design, candidate-drug identification based on real-time medical data, and even the search for pathways for new oral medicines.

AI-Led Drug Development Transformation Opens a $22.7 Billion Market by 2035

The convergence of biotech and AI is spreading rapidly across the industry. Earlier this month, AI company Anthropic acquired Coefficient Bio, a startup developing a drug-development platform, for $400 million. The platform being built by Coefficient Bio is designed to support the entire drug-development process with AI, including opportunity discovery, R&D planning and clinical-regulatory strategy management. Anthropic’s acquisition of a drug-development platform company is widely viewed as a strategic move to accelerate the buildout of a customized AI-tool ecosystem targeting the life sciences sector.

Swiss pharmaceutical giant Roche last month launched a “hybrid cloud AI factory” through a partnership with Nvidia. The AI factory refers to a hyperscale supercomputing platform, and under the agreement Roche will hold more than 3,500 of Nvidia’s Blackwell graphics processing units. That is the largest deployment publicly disclosed by any pharmaceutical company to date, and Nvidia’s high-performance GPUs are to be distributed across the United States and Europe.

In R&D, Roche will use Nvidia’s BioNeMo platform to strengthen its “Lab-in-the-Loop” model—an iterative R&D workflow in which AI proposes hypotheses, designs and predictions, automated experiments are run, and the resulting data are fed back in real time to continuously refine the model—and connect that framework to Roche’s own AI models. The company plans to validate vast numbers of hypotheses and accelerate discoveries that had previously been out of reach. In manufacturing, Roche will use digital twins—virtual replicas of production lines—powered by Nvidia’s Omniverse libraries. The goal is to create twin virtual factories in a 3D computing environment and identify the optimal production methods.

Eli Lilly also invested $1 billion in Nvidia in January this year to establish an “AI Collaborative Innovation Lab,” and last month began operating LillyPod, an Nvidia supercomputer composed of 1,016 GPUs. Elsewhere, AstraZeneca has been sharing the Cambridge-1 supercomputer in the United Kingdom with Nvidia and has trained AI on the structures of more than 1.4 billion compounds, while Amgen adopted Nvidia BioNeMo early on to build Freyja, a generative AI model for analyzing human datasets.

AI excels at identifying complex patterns within vast bodies of data. That strength allows it to handle large and intricate datasets in pharmaceuticals and biotech—including genomic data, protein structures, compounds and clinical data—at the same time while uncovering hidden correlations. The advent of generative AI has made it possible to directly design everything from text to compounds and proteins. As the technology advances into multimodal models capable of using increasingly diverse data types, it has also become possible to integrate and analyze genomics, images and clinical information together.

According to the Korea Biotechnology Industry Organization’s brief, Current Status and Outlook of the Global AI-Based Biotechnology Market, the AI-based biotechnology market is projected to reach $22.7 billion by 2035. From $3.5 billion in 2024, the market is expected to grow at a compound annual growth rate of 18.5% over the 11 years through 2035. By region, North America formed the largest market in 2024 with $1.5 billion, accounting for 42.6% of the total, followed by Europe at 28.2% and Asia-Pacific at 22.4%.

Asia-Pacific is expected to post the fastest growth going forward. The regional market is projected to expand from $800 million in 2024 to $5.7 billion by 2035, growing at a compound annual rate of 19.7%. Growth drivers include the expansion of AI-based drug development, policy support, rising foreign investment and an increase in the number of startups. North America and Europe are also expected to grow at compound annual rates of 17.9% and 18.9%, respectively, reaching $9.2 billion and $6.6 billion by 2035. Over the same period, the Latin American market and the Middle East and Africa market are projected to expand to $900 million and $400 million, respectively.