NVIDIA founder Jensen Huang: AI will completely transform drug discovery, collaborating with Eli Lilly to build the world's largest biological supercomputer.

Written by: Techub News Compilation

Introduction

At the recent JP Morgan Healthcare Conference, a highly anticipated dialogue unfolded between giants in the AI and biopharmaceutical fields. NVIDIA founder and CEO Jensen Huang and Eli Lilly Chairman and CEO David Ricks appeared on stage to announce a deep strategic partnership aimed at fundamentally transforming the drug discovery paradigm. This dialogue is significant as it marks the shift of AI computing capabilities from empowering general tasks to tackling one of humanity's most complex challenges—understanding and designing therapies for human diseases. Huang, as the vanguard of the accelerated computing and AI revolution, is embedding his technological blueprint deep within the life sciences; Ricks, leading the 150-year-old pharmaceutical giant Eli Lilly, represents the industry’s urgent desire to improve R&D efficiency and success rates. Their collaboration signifies that the era of “computer-aided drug design” may truly be upon us.

Summary

NVIDIA and Eli Lilly announced the establishment of a deep partnership to jointly build the world’s largest dedicated bio supercomputer and set up a joint AI laboratory.
The core goal is to leverage AI and accelerated computing to transform drug discovery from a reliance on chance, an "art," into a predictable, engineerable "computer-aided design" science.
The collaboration will combine NVIDIA’s BioNeMo platform, computational infrastructure, with Eli Lilly’s domain knowledge and data, creating an “AI R&D flywheel” from target discovery to molecular design.
Eli Lilly shared its successful paradigm represented by GLP-1 drugs and envisioned the tremendous potential of AI in tackling age-related diseases such as dementia and optimizing new therapy designs.

From Accelerated Computing to Conquering Biology: Huang's Vision

Jensen Huang opened by reflecting on NVIDIA's 33-year journey in deepening accelerated computing. He noted that through software-hardware co-design, NVIDIA has increased AI computing performance by approximately one million times over the past decade, far surpassing Moore’s Law’s hundredfold growth. This exponential leap in capability makes it possible to tackle highly complex fields such as human biology.

Huang used ray tracing and autonomous driving technologies as examples to illustrate how different computing fields have reached tipping points over time. He emphasized that breakthroughs in generative AI, multimodal understanding, and reasoning capabilities are creating new revolutions. For life sciences, multimodal AI can comprehend various biological data forms, including protein structures, molecular geometry, and genomic sequences, and can reason like humans, greatly expanding the "operating domain" of drug development. He explained that NVIDIA has constructed dedicated software platforms, including Parabricks (gene sequencing), MONAI (medical imaging), and BioNeMo (biomolecules), among which BioNeMo has become the foundation for most structural prediction models globally.

"I believe the moment for your industry has arrived," Huang told the life science professionals in the audience. He drew a parallel to the revolution that computer-aided design (CAD) brought to the chip industry forty years ago: transforming from the "art of creation" reliant on experience and luck to an engineering discipline capable of precise simulation, prediction, and iteration in the silicon realm. He believes that with today's computing capabilities increasing by millions or even tens of millions of times, humanity finally has the possibility to carry out similar digital representation and simulation of the most complex system—human biology—thus upgrading "drug discovery" to "computer-aided drug design."

Eli Lilly's Philosophy of Survival and Efficiency Engine: Ricks's Insights

David Ricks shared Eli Lilly's survival and development logic as a century-old pharmaceutical company. He pointed out that the pharmaceutical industry has long been trapped in the paradox of "innovation cycles exceeding return cycles," leading to cyclical fluctuations in company performance and R&D pipelines. Eli Lilly is one of the few major pharmaceuticals that have navigated downturns without large-scale mergers, which has forged its unique DNA of "persevering through difficult times."

To break this cycle, Eli Lilly, under Ricks's leadership, adopted a dual-pronged strategy: pursuing more inventions and maximizing the speed of invention. The company systematically deconstructed hundreds of R&D steps from selecting candidate compounds to receiving first approvals, striving to compress time. Currently, Eli Lilly's average R&D speed is about 40% faster than its peers, allowing it to complete innovations within patent periods and achieve sustainable growth.

The tremendous success of GLP-1 drugs (such as tirzepatide) is a product of this efficiency engine combined with scientific foresight. Ricks recalled that Eli Lilly began cultivating its GLP-1 field as early as 2006, and after more than a decade of iterations, it eventually launched a more effective therapy by merging the two hormones, GIP and GLP-1, becoming a market leader in obesity treatment. However, he is also acutely aware of the challenges: how to identify the next breakthrough before the current successful cycle ends. He believes that the collaboration with NVIDIA aims to systematically create the next "success conditions," rather than simply waiting for the next chance discovery.

Co-creating the Blueprint: Supercomputers, Joint Laboratories, and R&D Flywheels

The core content of this collaboration encompasses multiple aspects, aiming to construct a complete new paradigm of "AI-empowered drug discovery."

1. The World’s Largest Bio Supercomputer: Eli Lilly is building a massive dedicated supercomputing facility at its headquarters in Indianapolis, built on NVIDIA’s chips and technology, which will be completed soon. This will become the largest locally deployed supercomputer focused on biological discovery globally, providing an unprecedented power foundation for subsequent research.

2. Silicon Valley Joint AI Laboratory: The two parties will jointly establish a collaborative AI research lab in the San Francisco Bay Area (most likely in South San Francisco or the heart of Silicon Valley). This laboratory will attract the world’s top computer scientists and biologists, forming a cross-disciplinary innovation center. Huang emphasized that creating top-notch research conditions (such as supercomputers) is a prerequisite for attracting top talent and fostering great inventions, aligned with his earlier investments to build supercomputers for NVIDIA to attract AI researchers.

3. Building Data and Model Flywheels: Both parties recognize that high-quality, large-scale data is the cornerstone of AI models. The collaboration will include utilizing robotic laboratories (wet labs) to generate high-quality experimental data continuously 24/7 for training and validating AI models. Simultaneously, based on existing foundational models for protein design, molecular synthesis, and toxicity prediction on platforms like NVIDIA BioNeMo, researchers will perform fine-tuning and innovation. The goal is to form a reinforcement learning closed loop of "AI designs molecule -> robotic lab validates -> data feedback optimizes models," enabling the R&D flywheel to operate at high speed.

4. Tune Labs and Federated Learning: Ricks specifically mentioned the Tune Labs platform constructed by Eli Lilly. This platform, based on NVIDIA’s federated learning technology (such as NVFlare), allows Eli Lilly to collaborate with external biotech firms and startups to train AI models without sharing raw data, advancing research in specific areas together. This protects the intellectual property of all parties while gathering collective intelligence to accelerate innovation.

The Focus of AI in Pharmaceuticals: From Design Optimization to New Target Discovery

Regarding the specific research directions of the joint laboratory, Ricks proposed two core and complementary paths:

Path One: Drug Design and Optimization. This is a field where progress can currently be made relatively easily. For known and clearly defined biological targets (such as certain enzymes or receptors), AI is utilized to design or optimize corresponding molecules (“keys”), including small molecules, peptides, antibodies, and RNA therapies. AI can quickly explore vast chemical spaces and screen candidate compounds with ideal characteristics (such as potency, selectivity, pharmacokinetics), significantly improving the efficiency and success rate of the "key fitting the lock."

Path Two: New Target Discovery and Understanding. This is a more challenging but potentially groundbreaking "holy grail." Many diseases lack clear drug targets, or the biological functions of existing targets are not fully understood (which is one of the main reasons for drug development failure). The collaboration will leverage robotic laboratories to conduct large-scale, systematic experimental analyses on some promising but structurally complex targets, generating unprecedented deep datasets. AI models will learn from these data, helping scientists to more comprehensively understand target functions and even discover entirely new biological pathways that can be used for drug intervention.

Huang added that the ultimate goal is to establish a "synthetic data flywheel," where AI models can not only analyze real experimental data but also generate high-quality synthetic data, interacting in a closed loop with a "world model" simulating the biological system. This would enable countless "virtual experiments" to be conducted at extremely low costs in the digital world, greatly accelerating the process from hypothesis to validation.

Beyond GLP-1: Prospects for Future Disease Treatments

The conversation naturally extended to Eli Lilly's current most successful GLP-1 field. Ricks envisioned the future evolution of these drugs: better efficacy and tolerability (such as dual-targeted and triple-targeted drugs), longer dosing intervals (once a month or even longer), and the introduction of oral formulations (set to launch this spring), which will greatly enhance drug accessibility, especially in high-demand markets like China and India.

More importantly, the potential of GLP-1 drugs is expanding from weight loss to broader chronic disease management. Ricks pointed out that obesity is a core driver of many chronic diseases. GLP-1 drugs not only reduce weight but also significantly improve metabolic health (such as reducing the risk of transitioning from pre-diabetes to diabetes by 93%) and alleviates chronic inflammation, potentially having positive effects on arthritis and even some brain diseases (such as addiction and dementia). Eli Lilly has launched multiple clinical studies on these new indications.

When discussing the cutting-edge "longevity" field, Ricks expressed a pragmatic attitude. Eli Lilly’s focus is not on pursuing "immortality," but rather on extending human "healthspan" by conquering diseases. He pointed out the next frontier is age-related brain diseases like dementia. As other organ diseases are better controlled, brain health becomes crucial in affecting the quality of life for the elderly. He hopes that by collaborating with NVIDIA, they can simulate brain tissue to explore complex interactions of inflammation and protein misfolding, leading to new breakthrough targets in this area.

Reshaping the Industry Ecosystem: Collaboration Between Large Companies and Startups

In the context of the JP Morgan Conference, the two CEOs also discussed how AI is reshaping the ecosystem of the life sciences industry. Huang listed various innovative companies that emerged at the conference: "AI scientist" companies focused on robotic wet labs, companies developing healthcare dialogue agents, and biotech companies exploring new therapies such as bispecific antibodies. He believes that the clash between large companies and small firms is an important source of innovation.

Ricks fully agreed with this and pointed out that Eli Lilly has a tradition of embracing computing technology—decades ago, Eli Lilly purchased the first supercomputer in the pharmaceutical industry (Cray "Big Red") and designed the first drug entirely developed through computer assistance (insulin lispro). The current collaboration with NVIDIA is a natural extension of this tradition. He also mentioned that AI has enormous potential in improving the efficiency of healthcare services (such as automating administrative tasks and diagnostic assistance), and Eli Lilly is willing to leverage its market influence to collaborate with startups to jointly promote these changes, allowing the healthcare system to allocate more resources to genuinely innovative therapies.

Finally, Huang concluded that dedicating his life to computer science and collaborating with companies like Eli Lilly to tackle humanity's most important and complex challenge of health is an incredibly exciting journey. He believes that the synergy between the two companies is likely to "bend the arc of history," bringing about fundamental changes in drug discovery. Ricks responded that even setting moderate goals, this collaboration can significantly accelerate biomedical advancements, finding new pathways for an industry that desperately needs to enhance R&D productivity.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。