Eli Lilly has launched TuneLab, an artificial intelligence platform that gives biotech partners access to models trained on more than $1 billion of proprietary research data. The company is providing the platform at no cost, but every new dataset contributed by partners improves Lilly’s models. This structure allows Lilly to expand its data advantage and strengthen relationships with emerging biotechs, even as it supports broader innovation in drug discovery.
Chief Scientific Officer Daniel Skovronsky called TuneLab “an equalizer so that smaller companies can access some of the same AI capabilities used every day by Lilly scientists”. For many small firms, building high quality datasets is too expensive. Nisha Nanda, who leads Lilly’s Catalyze360 program, said TuneLab helps by “compressing decades of learning into instantly accessible intelligence”.
The company estimates that the first release of models includes data valued at over $1 billion. These datasets cover drug disposition, safety, and preclinical research. They also include failed experiments, which Thomas Fuchs, Lilly’s first Chief AI Officer, described as “enormously valuable” because they provide unique training signals for machine learning. Analysts at Jefferies estimate that the industry’s annual spending on AI could reach $40 billion by 2040.
TuneLab is hosted by a third party and uses federated learning. This means Lilly’s AI models are sent to a partner’s systems for training, and only encrypted model updates return to Lilly. Neither side has to share raw proprietary data. Lilly says this structure creates a loop where each partner contributes new data that improves the models for everyone.
TuneLab is also part of Catalyze360, a larger support program that includes venture funding through Lilly Ventures, laboratory space at Gateway Labs, and development expertise via ExploR&D.
Early partners include Circle Pharma, which is using TuneLab in cancer research, and Insitro, led by Daphne Koller, which is building predictive models for small molecule behavior. Firefly Bio and Superluminal Medicines are also participating.
For Fuchs, TuneLab is also about culture. “Machine learning and AI are really the art of failure,” he said in a recent interview with NYSE. “You try to model something, you fail, and you repeat this loop until you have a better understanding of the problem. That’s also the scientific loop”. He said the company needs to bring chemists, biologists, and executives into this way of working.
He also warned that the industry must stay grounded. “There is enormous fluff out there, a lot of snake oil,” Fuchs said. “You fight that with a sober outlook, with transparency about what AI can and cannot do. Models will not replace roles, they will replace tasks. And they must be safe, effective, and equitable for all patients”.
Lilly is also expanding AI into other areas. In manufacturing, the company is using digital twins of production lines and computer vision for quality control. In clinical development, AI is helping with trial enrollment and efficiency. For patients, Lilly is testing AI chat systems and telehealth services to improve engagement.
The FDA’s push to reduce animal testing gives further support to these efforts. The FDA Modernization Act 2.0 of 2022 encouraged alternatives to animal studies, and AI-driven modeling is one approach that could meet that goal.
Whether TuneLab will deliver faster breakthroughs remains uncertain. Fuchs said fully AI-designed drug candidates entering clinical trials are likely still a decade away. But improvements are already visible in toxicity modeling, trial recruitment, and production. By offering TuneLab at no cost, Lilly strengthens ties with biotechs while building a stronger data advantage for itself. Fuchs said his long-term vision is for Lilly to become an “AI-driven medicine company” that uses these tools to bring new treatments to patients more quickly.
