A Billion-Dollar Effort to Make Medicine Move Faster
The most intriguing AI trade in medicine may not be a machine that replaces a scientist. It may be a scientist who can finally stop waiting on the machine. Eli Lilly and Company (NYSE: LLY) and NVIDIA (NASDAQ: NVDA) are pursuing that opportunity through an AI co-innovation laboratory backed by up to $1 billion in combined investment over five years. Their ambition is to connect advanced computing, biological expertise and laboratory experimentation in a system that learns continuously, potentially making drug discovery faster and more productive. For many following the intersection of artificial intelligence and healthcare, that is a more substantial proposition than another chatbot with excellent bedside manners. The bullish case is straightforward: Lilly brings scientific knowledge, proprietary experimental data and the ability to develop and commercialize medicines. NVIDIA brings the computing infrastructure and AI technology needed to extract more value from that knowledge. If the combination improves the quality of research decisions, even without producing overnight cures, it could strengthen Lilly’s long-term earnings engine while expanding NVIDIA’s role in a demanding scientific market. That is the economic opportunity suggested by the collaboration, not a financial outcome the companies have already demonstrated.
Dario Amodei’s Big Idea Meets the Laboratory
In the video below, Anthropic (ANT) chief executive Dario Amodei describes a future in which sufficiently powerful AI, working alongside people, could compress a century of scientific and medical progress into a decade. He points to applications spanning drug design, cancer genetics and clinical-trial administration. Crucially, he also acknowledges that the evidence remains early and that experiments and clinical work still take time. That distinction separates an ambitious scientific thesis from an earnings forecast. Amodei’s argument is that intelligence itself has been a constraint on discovery: too much biological complexity, too many possible explanations and too few researchers able to investigate them. Lilly’s collaboration with NVIDIA addresses a related practical question: How can a pharmaceutical company turn greater computational capability into better experiments, rather than merely more computational activity? The answer being pursued is a continuous learning system. Computational models help inform experiments, laboratory results generate new data, and those data help improve the models. Scientists remain in the loop. The objective is not to persuade biology to obey software. It is to interrogate biology more effectively.
Why Lilly’s AI Strategy Has Substance
Lilly’s AI effort rests on an established research organization rather than a newly decorated corporate presentation. Its 2025 Form 10-K reports approximately 12,000 employees engaged in research and development at year-end. Research and development expenses increased 21% during that year, reflecting continued investment across early- and late-stage programs. That gives the company a broad scientific base on which to deploy new tools. The spending continued into 2026. Lilly’s third-quarter Form 10-Q reports research and development expenses of approximately $11.6 billion for the first nine months, compared with approximately $9.5 billion in the same period a year earlier. Separately, the company generated approximately $27.4 billion in operating cash flow during those nine months. Those figures matter because AI-enabled discovery requires more than purchasing processors. It requires sustained spending on experiments, data systems, scientific talent and the difficult work of integrating them. Lilly has both the incentive and the financial capacity to attempt that integration. Its annual filing explicitly identifies the continual discovery or acquisition, development and commercialization of innovative medicines as essential to its long-term success. AI, in that context, is not an extracurricular activity. It is a potential improvement to the company’s core production process.
NVIDIA’s Opportunity Extends Beyond Selling Chips
The partnership also illustrates why NVIDIA’s healthcare opportunity deserves attention. In October 2025, Lilly announced an NVIDIA-based supercomputer powered by more than 1,000 B300 graphics processing units. Lilly described the system as the foundation for an “AI factory” supporting the lifecycle from data ingestion and model training to fine-tuning and large-scale inference. The subsequent co-innovation laboratory broadens that relationship. The companies said its infrastructure would use NVIDIA’s BioNeMo platform and incorporate the Vera Rubin architecture, bringing Lilly’s scientists together with NVIDIA’s AI researchers and engineers. For NVIDIA, the strategic attraction is a deeper position in scientific workflows. The announced collaboration encompasses not only drug discovery but potential applications in clinical development, manufacturing, medical imaging, robotics and digital twins, which are virtual representations used to study physical systems. That creates several possible avenues for demand. It does not establish how much revenue NVIDIA will earn from them, nor does the announced combined investment represent a disclosed chip order. Still, the direction is commercially appealing: the more closely advanced computing becomes embedded in pharmaceutical research and production, the less it resembles an occasional equipment purchase.
The Data Advantage May Be Hiding in Failed Experiments
One of Lilly’s most interesting AI assets may be information that once looked like a scientific disappointment. In its explanation of the supercomputer initiative, Lilly says it can build models using internal data that include lessons from millions of molecules that did not succeed. The company argues that this proprietary information can help deepen biological understanding and support the design of better research programs. The implication is important. A failed molecule may still reveal something useful about a target, a chemical structure or an experimental approach. A system capable of learning across those results could make historical research more valuable. In pharmaceuticals, the archive of what did not work may deserve a better office. Lilly has also moved to extend access to AI-enabled discovery tools through TuneLab. Its third-quarter 2025 earnings release highlighted the platform’s launch and described its models as built through more than $1 billion in research investment. That historical research investment is distinct from the later commitment of up to $1 billion for the NVIDIA co-innovation laboratory. Lilly’s 2026 proxy filing likewise identifies access to AI-enabled discovery models and the planned AI supercomputer among its innovation initiatives. Taken together, the disclosures suggest an effort to build a connected research capability rather than a collection of isolated demonstrations.
A Growth Business Can Fund Its Next Growth Business
Lilly’s AI ambitions are developing alongside substantial commercial momentum. In the first quarter of 2026, the company reported revenue of $19.8 billion, up 56% from a year earlier. The earnings release attributed the increase primarily to higher volume, partly offset by lower realized prices. By the third quarter, revenue reached $24.4 billion, up 39% year over year. Lilly raised its full-year 2026 revenue guidance to $91 billion to $92 billion. Third-quarter Mounjaro revenue was approximately $10.7 billion, while worldwide Zepbound revenue was approximately $5.4 billion. The distinction is worth preserving: those sales are evidence of Lilly’s existing commercial strength, not proof that its new AI infrastructure has already improved drug-discovery returns. They nevertheless provide a powerful foundation. A business generating substantial revenue and operating cash flow can finance long-duration scientific projects while continuing to invest in its current portfolio. The market opportunity is therefore not simply “Lilly plus AI.” It is a successful pharmaceutical enterprise attempting to make its future research spending more productive. The numbers also require context. Lilly said third-quarter revenue included approximately $900 million in benefits, primarily from milestones and favorable rebate and discount adjustments. Worldwide realized prices declined 15%, partially offsetting a 53% increase in volume. Growth is strong, but it is not free of commercial friction.
Where the Economic Payoff Could Appear
The most credible bullish scenario does not require AI to discover a blockbuster every quarter. It requires useful improvements across a costly, uncertain process. Based on the applications outlined by Lilly and NVIDIA, three areas deserve particular attention:
- Better research prioritization. Models trained on experimental results could help scientists identify, optimize and validate more promising molecules before committing additional resources.
- Faster learning between experiments. The planned connection between computational “dry labs” and experimental “wet labs” aims to support continuous, AI-assisted experimentation with scientists directing the work.
- Improvements beyond discovery. Lilly describes potential uses in clinical-trial design, imaging-based research and personalized care, while the broader collaboration includes manufacturing applications. These are development opportunities, not established savings.
Lilly’s own overview emphasizes the combination of scientific expertise and advanced AI as a way to improve both speed and confidence in medicine design. That second objective may prove just as valuable as the first. Accelerating a poor decision is not productivity. It is merely punctual disappointment.
The Milestones That Will Separate Progress From Promotion
The next phase should be judged by measurable scientific and operating results. For many tracking Lilly stock and NVIDIA’s healthcare expansion, the useful questions are whether these systems improve prospective predictions, shorten specific research workflows and help advance stronger candidates. Laboratory deployment alone will not establish that the economics have improved. Clinical validation remains essential. Lilly’s second-quarter 2026 Form 10-Q describes a pipeline progressing through clinical studies and regulatory processes, underscoring the distance between a promising research idea and a marketable medicine. Amodei makes a similar qualification in the supplied video: computation can accelerate parts of the process, but experiments and clinical work remain necessary. Lilly’s annual filing also identifies risks associated with AI, including competitive, regulatory, litigation and cybersecurity concerns. Better tools do not eliminate the obligations attached to using them. Nor does a compelling operating story automatically establish an attractive share price. The collaboration strengthens a long-term business thesis; it does not, by itself, resolve valuation.
The Bullish Case: A More Productive Pharmaceutical Engine
Lilly and NVIDIA are pursuing an appealing division of labor: one company understands how to turn scientific discoveries into medicines, while the other supplies technology intended to make complex computation more useful at scale. Their laboratory collaboration seeks to bring those capabilities into a tighter experimental loop. The opportunity is larger than a single AI-designed molecule. It is the possibility that proprietary data, better models and continuous experimentation can improve the productivity of an entire research organization. Lilly already has the commercial scale and research spending to make incremental gains meaningful. NVIDIA gains an opportunity to deepen its position in a scientific workflow that can extend from discovery to production. Neither outcome is assured, but both have a clearer foundation than the proposition that every industry needs AI because the earnings call has reached that slide. For many looking beyond the familiar debate over how much companies will spend on AI, this is the more interesting question: What happens when that spending helps a highly valuable industry learn faster? In medicine, buying back time could become a formidable business.
The Sources
- Supplied video, “Anthropic CEO Dario Reveals How AI Could Transform Medicine and Accelerate Cures,” speech transcript.youtu
- Eli Lilly, “Advancing Medicine Discovery With AI.lilly
- Eli Lilly and NVIDIA, “NVIDIA and Lilly Announce Co-Innovation AI Lab to Reinvent Drug Discovery in the Age of AI,” January 12, 2026.investor.lilly
- Eli Lilly, “Lilly Partners With NVIDIA to Build the Industry’s Most Powerful AI Supercomputer,” October 28, 2025.investor.lilly
- Eli Lilly, “Our New Supercomputer Could Change the Future of Medicine,” October 28, 2025.lilly
- SEC.gov, Eli Lilly, Form 10-K for the year ended December 31, 2025.sec
- SEC.gov, Eli Lilly, Form 10-Q for the quarter ended September 30, 2026.sec
- SEC.gov, Eli Lilly, third-quarter 2026 earnings release, Exhibit 99.1, October 29, 2026.sec
- SEC.gov, Eli Lilly, third-quarter 2025 earnings release, Exhibit 99.1, October 30, 2025, including the TuneLab announcement.sec
- SEC.gov, Eli Lilly, 2026 proxy filing, March 17, 2026.sec
- SEC.gov, Eli Lilly, first-quarter 2026 earnings release, Exhibit 99.1, April 30, 2026.sec
- SEC.gov, Eli Lilly, Form 10-Q for the quarter ended June 30, 2026.sec
Stay Updated with Vista Partners
Subscribe to receive market insights, investing ideas, and the latest updates directly in your inbox.
