Lilly and Novo Show How AI Is Rewiring Big Pharma

Eli Lilly signed a $2.75 billion partnership with Hong Kong-based AI drug discovery firm Insilico Medicine, giving it exclusive rights to develop and commercialize drugs built on Insilico’s Pharma.ai platform. As reported by PYMNTS on Monday (March 30), the platform has already produced at least 28 drug candidates using generative AI, nearly half of which have entered clinical testing.

Novo Nordisk, meanwhile, is running AI agents inside its live clinical trials to cut weeks or months from development timelines. Together, the moves signal that AI in pharma is no longer a productivity tool layered on top of existing processes. It is becoming the process itself.

Eli Lilly’s AI strategy started on the factory floor. Facing a federally declared shortage of its GLP-1 drugs that lasted from late 2022 through 2024, the company deployed digital twin technology, a virtual model of its manufacturing plant fed by real-time data, to identify production improvements it could test digitally before implementing them physically. It also used computer vision to photograph each autoinjector from multiple angles in rapid succession, catching defects that manual inspection would miss.

The results showed up in revenue. Diogo Rau, Lilly’s chief information and digital officer, told Forbes the company produced more GLP-1 product last year than it could have without AI, at volumes large enough to affect earnings. Mounjaro sales doubled year over year to $23 billion, while Zepbound revenue grew from $4.9 billion to $13.5 billion, with both drugs together accounting for more than half of Lilly’s $65 billion in total revenue.

The Insilico deal pushes AI further upstream into molecule design, including oral drug candidates that would offer a commercial advantage over injectable GLP-1s.

“By deploying frontier AI technologies that scale from biomarkers to life models, world models of human and animal life, we can identify multi-purpose targets driving multiple diseases at the same time,” said Alex Zhavoronkov, founder and CEO of Insilico.

Novo Nordisk Uses Agents to Accelerate Clinical Trials

Novo Nordisk, which has generated nearly $100 billion in cumulative sales from Ozempic and Wegovy, is focused on compressing the clinical trial phase.

The company spent roughly a year building AI agents trained on its own internal data and publicly available competitor disclosures, using software from German firm Celonis, which has raised $1.6 billion from investors including Accel, as The Information reported.

Those agents, drawing on models from Anthropic, OpenAI and other providers, run inside active trials. They detect protocol gaps and incomplete data, alert trial leadership to risks before delays materialize and have begun handling analytical work previously outsourced to several hundred external contractors.

Novo also partnered with Nvidia to access the Gefion sovereign AI supercomputer in Denmark, as announced by Nvidia, to run large-scale drug discovery workloads. Stephanie Bova, Novo’s digital transformation officer, framed the time-to-market question in terms any CFO would recognize.

“A week of time saved in getting to market can mean tens of millions if not hundreds of millions of dollars in peak revenue impact,” she told The Information. “It’s a race.”

Not every AI tool earns a full rollout. Novo restricted access to Found Data, an internal research tool powered by Anthropic’s Claude that lets scientists scan decades of prior trial results for overlooked patterns, because running costs outpaced the value it delivered at scale. Bova was direct about the lesson.

“There’s been this mad dash to have AI use cases without thinking about whether it’s the best use of your time and money to do it that way,” she said.

The GLP-1 race now provides the most concrete evidence available that AI deployed vertically, across discovery, manufacturing and clinical execution, generates competitive separation that AI used as a horizontal productivity layer cannot replicate.

For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.