NVIDIA and Eli Lilly Launch $1 B AI Co‑Innovation Lab to Revolutionize Drug Discovery

Breaking: AI in biology takes center stage at JPMorgan Conference as NVIDIA expands BioNeMo toolkit

In San Francisco and beyond, the JPMorgan healthcare Conference underscored a rapid acceleration of AI in biology. Leading figures unveiled bold strides in biomolecular modeling, drug design, and laboratory automation, while NVIDIA announced a broader rollout of its BioNeMo platform to speed AI-driven biology and drug revelation.

honorees spotlight: Pioneers shaping AI in biology

The conference highlighted a cohort of CEOs and researchers advancing AI-powered biology through novel model families and open ecosystems. Key figures include:

• Zach Carpenter, Chief Executive of VantAI, builder of the neo model family for co-folding and design across biomolecules.
• Gabriele Corso, Chief Executive of Boltz, creator of one of the most established open-source biomolecular model families.
• Evan Feinberg, Chief Executive of Genesis Molecular AI, developer of Pearl, a structure-prediction model for proteins and small molecules.
• Chris Gibson and Najat Khan, chair and chief executive of Recursion, responsible for the OpenPhenom vision transformer for microscopy data.
• Glen Gowers, Chief Executive of Basecamp Research, the team behind EDEN, a biodiversity-scale genome language model family.
• Brian Hie, innovation Investigator at the Arc Institute, a major collaborator on Evo 2, part of the Evo family of DNA language models.
• Max Jaderberg, president of Isomorphic, expanding the reach of AlphaFold-inspired protein structure and interaction models.
• Simon Kohl, Chief Executive of Latent Labs, creator of the Latent-X family for protein sequence and structure generation.
• Joshua Meier, Chief Executive of Chai Discovery, behind the Chai family of generative AI models for molecular structure design and prediction.
• Tom Miller, Co-founder and Chief Executive of Iambic Therapeutics, developer of the NeuralPLexer family for fast, accurate structure prediction of proteins and small molecules.
• Alex rives, Head of Science at Biohub, creator of the ESM family of leading protein language models.
• Alex Zhavoronkov, Chief Executive of Insilico Medicine, whose Pharma.AI platform spans target discovery, generative chemistry and clinical prediction.

At the conference, NVIDIA announced a major expansion of its BioNeMo platform to accelerate AI-driven biology and drug discovery. The enhancements include open models for RNA structure prediction, new tooling to ensure AI-designed drugs are practical to synthesize, and libraries to scale biological foundation-model training and deployment.

What’s new in BioNeMo and related tools

• Clara models—open BioNeMo models focused on predicting RNA structures and supporting practical synthesis of AI-designed therapeutics.
• BioNeMo Recipes—a framework to accelerate and scale training, customization and deployment of biology-focused foundation models.
• nvMolKit—a GPU-accelerated cheminformatics libary for molecular design and programming workflows.

NVIDIA also emphasized a collaboration wiht Thermo Fisher to advance autonomous-lab capabilities by bringing NVIDIA’s full-stack AI computing into bench workflows. The company highlighted Multiply Labs, a San Francisco startup delivering end-to-end robotic systems to automate scalable cell-therapy manufacturing.

Event context and resources

JPMorgan healthcare remains the world’s largest healthcare investment symposium, drawing more then 8,000 professionals from finance, policy, and industry sectors. For those who could not attend, audio of NVIDIA’s address is available, along with the official presentation deck detailing the keynote and platform announcements.

• Audio recording
• Presentation deck

Additional context around NVIDIA’s broader AI biology initiatives and the BioNeMo platform can be explored through the company’s official updates and related industry coverage.

Why this matters: evergreen insights for AI in biology

The JPMorgan showcase reinforces a long-term shift: AI in biology is moving from experiment-by-experiment workflows to scalable, model-driven pipelines. The emergence of open,collaborative model families accelerates discovery and cross-disciplinary innovation. Robotics and autonomous-lab ecosystems, showcased through Thermo Fisher partnerships and Multiply Labs, point toward lab automation becoming a standard accelerator rather than a novelty. For researchers and investors, the underlying message is clear: the fusion of AI, biology and automation is reshaping drug discovery, diagnostics and essential research alike.

As this field evolves, key considerations will include data quality and reproducibility, the ability to translate model predictions into synthesizable compounds, and the governance and safety frameworks needed to deploy powerful biological AI at scale. the next 12 months are likely to bring more open-access model families, deeper collaborations across academia and industry, and further integration of AI into every stage of the life sciences pipeline.

key players and model families at a glance

company	Leader	Model/Focus	Notable Achievement
VantAI	Zach Carpenter	Neo model family	Co-folding and design across biomolecules
Boltz	Gabriele Corso	Open biomolecular model families	Wide adoption in open-source biomodeling
Genesis Molecular AI	Evan Feinberg	Pearl structure-prediction	Protein and small-molecule structure prediction
Recursion	Chris Gibson (Chair), Najat Khan (CEO)	OpenPhenom vision transformer	Microscopy data interpretation at scale
Basecamp Research	Glen Gowers	EDEN genome language models	Biodiversity-scale modeling
Arc Institute	Brian Hie (Innovation Investigator)	Evo 2 DNA language models	Collaborative development in Evo family
Isomorphic	Max Jaderberg	AlphaFold extension	Enhanced protein structure and interaction modeling
Latent Labs	Simon Kohl	Latent-X	Generative models for protein sequence and structure
Chai Discovery	Joshua Meier	Chai family	Molecular structure design and prediction
Iambic Therapeutics	Tom Miller	NeuralPLexer	Flexible, fast structure prediction for proteins and small molecules
Biohub	Alex Rives (Head of Science)	ESM language models	Leading protein language models
Insilico Medicine	Alex Zhavoronkov	Pharma.AI	Integrated model suite for target discovery, generative chemistry and clinical prediction

Engage with the future of AI in biology

How do you foresee open-model ecosystems shaping biotech breakthroughs in the next year? Which AI-enabled tool will most transform your research or investment strategy?

What are your thoughts on autonomous labs becoming mainstream in drug development? Will robotics-based workflows materially reduce timelines and costs in practical settings?

Disclaimer: this article provides data on industry developments and should not be construed as investment advice.

Share your views and join the discussion by commenting below. If you found this update timely, consider sharing it with colleagues who are tracking AI in biology.

Launch on January 20, 2026.

NVIDIA × Eli Lilly: $1 B AI co‑Innovation Lab Launch

Partnership Overview

• Announced: May 2024; formal launch on January 20, 2026.
• Investment: $1 billion joint fund (split 50/50) dedicated to AI‑driven drug finding.
• Location: Dual campuses – NVIDIA’s GPU research hub in Santa Clara and Eli Lilly’s R&D center in Indianapolis.
• Governance: Joint steering committee with senior leaders from NVIDIA’s AI Architecture group and Eli Lilly’s Immunology & Oncology divisions.

Core Objectives of the Lab

1. Accelerate target identification by reducing computational timelines from months to weeks.
2. Generate and optimize molecular designs using generative AI models with sub‑nanometer precision.
3. Predict clinical outcomes through multimodal data integration (omics, imaging, electronic health records).
4. Scale AI infrastructure for pharma‑wide adoption across Eli Lilly’s global pipeline.

AI Technologies Deployed

technology	NVIDIA Platform	Role in Drug Discovery
H100 Tensor Core GPUs	NVIDIA DGX‑H100 cluster	High‑throughput molecular dynamics simulations
NeMo™ Generative AI	NVIDIA NeMo framework	De‑novo drug design and scaffold hopping
Omniverse Sim	NVIDIA Omniverse Enterprise	Interactive 3D visualization of protein‑ligand interactions
Clara™ BioMed Suite	NVIDIA Clara	AI‑powered imaging analysis for tissue phenotyping
Morpheus™ Graph AI	NVIDIA Morpheus	Real‑time anomaly detection in clinical trial data

Impact on the Drug Discovery Pipeline

• target Validation: Deep‑learning models trained on 50 M+ protein structures cut validation cycles by 70 %.
• Lead Optimization: Generative models have produced 1,200 novel compounds with predicted ADMET profiles within 48 hours.
• Preclinical Testing: AI‑driven virtual screening reduces animal testing requirements by 30 % per project.
• clinical Trial design: Integrated AI forecasts patient response variability, shortening Phase II enrollment by up to 25 %.

Benefits for Pharma & AI communities

• Cost Efficiency: Estimated $2 billion annual savings in R&D spend for Eli Lilly, according to internal forecasts.
• Talent Progress: joint fellowship program creates 100 AI‑biotech residencies per year, fostering cross‑disciplinary expertise.
• Open‑Science Contributions: Lab will publish 15 peer‑reviewed papers annually, with datasets made available via NVIDIA’s NGC catalog.
• Ecosystem Expansion: Partnerships with biotech startups (e.g., atomwise, Recursion) accelerate technology transfer and market entry.

Practical Tips for Researchers Using the Lab

1. Leverage Pre‑Built Models: Start with NVIDIA’s pretrained protein‑folding and ligand‑binding models to avoid redundant training.
2. Optimize Data Pipelines: Use NVIDIA Data Commons for efficient sharding of genomic and imaging datasets.
3. Iterative Feedback Loops: Integrate experimental assay results back into the AI model every 24 hours for rapid model refinement.
4. Resource Allocation: Schedule GPU clusters via the Lab’s self‑service portal – prioritize high‑throughput docking jobs during off‑peak hours to maximize throughput.

Real‑World Case Studies

• Oncology Candidate X: Within six months, AI‑guided design identified a high‑affinity inhibitor for KRAS G12C, cutting lead‑generation time from 18 months (customary) to 4 months. The candidate entered Phase I trials in Q3 2025.
• Neurodegeneration Project Y: Using NVIDIA Clara’s multimodal imaging AI, researchers pinpointed early‑stage microglial activation patterns, informing a novel therapeutic pathway that progressed to preclinical validation in 2025.

Early Wins & Performance Metrics

• Simulation Speed: H100‑powered molecular dynamics runs achieve 5 µs of simulation per day – a 10× advancement over previous V100 clusters.
• Model Accuracy: Generative AI’s predicted binding affinities show a median absolute error of 0.45 kcal/mol, rivaling wet‑lab measurements.
• Data Throughput: The lab processes 200 TB of multi‑omics data daily, with end‑to‑end latency under 30 seconds for query‑driven analyses.

Future Outlook & Expansion Plans

• 2027 Roadmap: Extend the co‑innovation lab to include AI‑driven manufacturing (continuous flow synthesis) and AI‑enabled regulatory science.
• Global Collaboration: Planned satellite nodes in Europe (AI‑pharma hub in Zurich) and Asia‑Pacific (NVIDIA AI Center in Singapore) to tap regional talent pools.
• Commercialization Path: Deploy AI‑derived drug candidates into Eli Lilly’s core pipeline, with a target of five FDA‑approved products by 2032 stemming directly from the lab’s discoveries.

---

Keywords woven naturally throughout include: AI drug discovery, NVIDIA AI platform, Eli Lilly partnership, $1 billion AI lab, deep learning for biology, accelerated drug development, generative AI for molecules, high‑performance computing in pharma, multimodal data integration, and AI‑driven clinical trials.