Eli Lilly Is Building a Proprietary AI Data Flywheel Few Rivals Can Replicate

The long-term bull case for $LLY is not just its drug pipeline—it is the proprietary research data feeding its AI infrastructure.
Lilly is running more than 1,000 $NVDA B300 GPUs in what it describes as the pharmaceutical industry’s most powerful company-owned supercomputer. Its initial TuneLab dataset includes over 500,000 data points , more than 20 years of research , over $1 billion in investment , and hundreds of thousands of unique molecules.
The real advantage may be the data competitors cannot access: failed molecules, toxicity signals, abandoned hypotheses, and millions of historical experiments accumulated across decades.
Lilly’s acquisition strategy could strengthen that flywheel further. Its announced 2026 deals, valued at up to approximately $27 billion , bring additional proprietary research across cancer, gene therapy, vaccines, and other fields.
Each new experiment and acquisition can improve the central AI system, while every business unit benefits from the intelligence it produces—a model similar to the organisational data advantage Elon Musk is attempting to build across his companies.