Eli Lilly has launched Lilly TuneLab, an artificial intelligence (AI) and machine learning platform that gives biotechnology companies access to drug discovery models trained on years of Lilly’s research data.
The initial release includes proprietary datasets valued at more than $1 billion, covering drug disposition, safety, and preclinical results from hundreds of thousands of unique molecules. Lilly is starting with 18 models, including ones that predict small molecule properties and assess antibodies.
While access is free for selected biotech partners with no intellectual property requirements, the platform uses a federated learning system that allows biotechs to run Lilly’s AI models on their own proprietary data without handing that data over.
Through federated learning, each partner trains the model on its own data and sends back only the resulting model updates, which Lilly combines and redistributes as an improved version — allowing everyone to benefit without sharing raw data. The trradeoff is that participation effectively contributes to Lilly’s model training, giving the company stronger, more accurate systems over time.
Lilly built TuneLab with global technology providers and AI specialists and plans to extend it to include in vivo small molecule predictive models. TuneLab sits within Lilly Catalyze360, which also offers capital through Lilly Ventures, lab space at Lilly Gateway Labs, and development support via Lilly ExploR&D.
The company said the platform is meant to give smaller biotechs access to AI tools and datasets typically reserved for large drugmakers, estimating that more than 90 percent of participants will benefit from the models.
By opening its models in this way, Lilly is expanding access to AI-driven discovery and positioning its technology to become a standard across the sector.
What it means for biotechs
Technological advances have steadily pushed drug discovery forward, and today’s AI models are expected to accelerate that progress even further.
A clear precedent comes from research on cystic fibrosis (CF), where scientists used fluorescence screening combined with automated robotics to evaluate more than a million compounds, ultimately narrowing in on the molecules that became the foundation of the triple-drug therapy now transforming patient outcomes.
But today's datasets are increasingly large and complex, requiring extensive computing power, cloud infrastructure, and in-house AI expertise — resources many biotechs simply cannot afford.
TuneLab could help level the playing field. By providing access at no cost, Lilly lowers these barriers. Companies can test candidates against Lilly-trained models, refine hypotheses faster, and potentially reduce the number of compounds that fail before reaching animal or clinical studies.
Because Lilly’s models improve as more companies use TuneLab, widespread adoption could effectively make the platform a standard for evaluating early-stage candidates — giving Lilly a quiet but significant influence over discovery pipelines across the industry.
For biotechs, the calculus may be simple: Access to high-quality models today could mean faster, cheaper programs with higher odds of success. The risk is that the same tools that help them compete may also bind them closer to Lilly’s orbit in the long run.
AI has already shown measurable impact in early discovery: Preclinical costs are being cut by around 30 percent in some studies thanks to better compound prioritization, predictive toxicity models, and optimization of in vitro assays.
Frequently asked questions (FAQ):
(generated with AI assistance)
Who can access Lilly TuneLab?
Access is currently limited to selected biotech partners. Lilly says more than 90 percent of those chosen are expected to benefit from the models.
Do participating companies give up their proprietary data?
No. Biotechs run Lilly’s AI models locally on their own datasets. Only model updates — not raw data — are shared back with Lilly, which uses them to refine its systems.
Why is Lilly offering the models for free?
The company gains value as more biotechs use the platform. Widespread adoption strengthens the models and could make them a standard tool across the industry, indirectly boosting Lilly’s influence.
