GTC 2025 Spring Healthcare Special Address

Category: Healthcare Special Address · Year: 2025 · ▶ Watch

Speakers: Kimberly Powell, Vice President of Healthcare and Life Sciences, NVIDIA

Switch language → 中文

Segments (13)

  • 00:00 · Introduction
    • A video montage showcasing NVIDIA’s impact on healthcare over the years.
  • 02:08 · Welcome and GTC Overview
    • Kimberly Powell welcomes attendees and highlights the scale of healthcare presence at GTC.
  • 03:56 · Accelerated Computing in Healthcare
    • A review of how accelerated computing has transformed medical imaging, genomics, and drug discovery.
  • 05:56 · Decoding Biology and Evo 2
    • Discussion on understanding biology with AI and the announcement of the Evo 2 genomic foundation model.
  • 10:53 · Announcing DGX Spark
    • Introduction of the DGX Spark, a benchside AI supercomputer for researchers.
  • 12:28 · Every Wet Lab Needs a Dry Lab
    • Exploring the integration of digital AI labs with physical wet labs using the BioNeMo platform.
  • 16:05 · AI Drug Discovery Ecosystem
    • Highlighting partners using AI to accelerate drug discovery and improve R&D productivity.
  • 17:39 · AI Agents in Healthcare
    • Introduction of reasoning models and the NVIDIA AI Agent Platform for healthcare applications.
  • 19:55 · Synchron BCI Patient Story
    • A video demonstrating how a patient uses an AI-enabled Brain-Computer Interface to interact with his environment.
  • 22:50 · Physical AI and Robotics
    • Announcing Isaac for Healthcare to build physical AI for medical devices and robotics.
  • 24:12 · MONAI Multimodal
    • Updates to the MONAI platform, introducing multimodal and agent frameworks for medical imaging.
  • 26:48 · Holoscan 3.0 and Ecosystem
    • Launch of Holoscan 3.0 for real-time sensor processing and highlights of the growing ecosystem.
  • 28:44 · Conclusion
    • Summary of the AI revolution in healthcare across drug discovery, digital health, and medical devices.

Product Announcements (8)

  • [07:48] Evo 2 Genomic Foundation Model
    • The world’s largest biology foundation model.
    • specs: Trained on 9 trillion nucleotides, 1 million context length.
    • availability: Available now
  • [10:53] NVIDIA DGX Spark
    • A benchside AI supercomputer.
    • specs: 1 petaflop of FP4 performance, fits on a desk.
    • availability: Shipping in July
  • [14:10] NVIDIA BioNeMo Platform Updates
    • Platform for drug discovery.
    • specs: New NIMs and Blueprints for generative chemistry and virtual screening.
    • availability: Available now
  • [19:50] NVIDIA AgentIQ
    • Toolkit to connect agents, data, and tools.
    • specs: Uses reasoning models to orchestrate complex workflows.
    • availability: Available now
  • [20:25] Epic with NVIDIA NIM on Microsoft Azure
    • Integration of NVIDIA AI into Epic’s EHR.
    • specs: Accelerates AI deployment for healthcare providers.
    • availability: Available now
  • [23:30] Isaac for Healthcare
    • Platform to build physical AI for medical devices.
    • specs: Includes training, simulation (Omniverse), and deployment (Holoscan).
    • availability: Early access
  • [24:45] MONAI Multimodal and Agent Framework
    • Updates to the medical imaging AI platform.
    • specs: Supports DICOM, text, EHR, video, and omics data with reasoning agents.
    • availability: Available now
  • [26:50] NVIDIA Holoscan 3.0
    • Multimodal real-time AI sensor processing platform.
    • specs: Dynamic flow control for runtime decision making.
    • availability: Available now

Specific Numbers (9)

Timestamp Metric Value Context
02:58 Healthcare companies at GTC 700+ Number of healthcare companies participating in the conference.
07:58 Evo 2 training data 9 trillion Number of nucleotides the Evo 2 model was trained on.
08:54 Evo 2 context length 1 million The context window size for the Evo 2 model.
11:28 DGX Spark performance 1 petaflop FP4 compute performance of the DGX Spark.
15:51 R&D industry spend $300 billion Annual spending on R&D and drug discovery.
16:26 Drug discovery timeline 13 months Time taken by Insilico Medicine to go from target to candidate.
19:23 Synthetically accessible compounds 1.2 trillion Number of compounds explored by Cadence’s Orion platform.
24:15 MONAI downloads 4.5 million Total downloads of the MONAI platform.
30:40 Epic patient records 325 million Number of patient records managed by Epic worldwide.

Benchmark Claims (1)

  • [08:55] Context Length: 1 million
    • vs: Evo 1 (100k)
    • gain: 10x increase in context length.

Customer Stories (7)

  • [16:15] Insilico Medicine
    • Used AI for drug discovery.
    • outcome: Went from target to candidate in 13 months and brought 10 assets to the clinic.
  • [17:39] BIOVIA
    • Built a generative therapeutics design platform with BioNeMo.
    • outcome: Integrated generative chemistry NIMs into their workflow for high-quality synthesizable compounds.
  • [18:29] Sapio Sciences
    • Built a no-code informatics platform with BioNeMo.
    • outcome: Enabled scientists to use AI microservices without writing code.
  • [19:02] Cadence
    • Built the Orion molecular design platform with BioNeMo.
    • outcome: Allowed researchers to efficiently explore 1.2 trillion synthetically accessible compounds.
  • [19:55] Synchron
    • Developed an implantable Brain-Computer Interface.
    • outcome: Enabled a patient with ALS to control smart home devices and communicate using Apple Vision Pro and NVIDIA technology.
  • [25:03] Hippocratic AI
    • Created a marketplace of healthcare AI agents.
    • outcome: Deployed agents for over 400 use cases across 25 specialties.
  • [27:00] GE HealthCare
    • Partnered with NVIDIA to build physical AI capabilities.
    • outcome: Developing autonomous ultrasound and X-ray systems to improve access and early detection.

Key Technologies (7)

  • Accelerated Computing: Speeds up complex calculations for imaging, genomics, and simulations.
  • Generative AI / Foundation Models: Understands and generates biological sequences (DNA, RNA, proteins).
  • BioNeMo: Platform for generative AI in drug discovery.
  • AI Agents: Autonomous systems that can reason, plan, and execute tasks.
  • Physical AI (Isaac): AI that understands and interacts with the physical world via robotics.
  • MONAI: Medical Open Network for AI, used for medical imaging.
  • Holoscan: Real-time AI sensor processing platform for medical devices.

Demos Shown (2)

  • [19:55] Video of a patient using the Synchron BCI to control an Apple Vision Pro and smart home devices.
    • Yes
  • [27:16] UI walkthrough of a Biomedical Research Agent Blueprint performing literature search, hypothesis generation, and virtual screening.
    • Yes

Predictions / Commitments (3)

  • [12:28, Current/Near future] Every wet lab needs a dry lab.
  • [21:48, Current/Near future] Reasoning is an incredibly new frontier of large language models.
  • [28:46, Current] This is the most exciting time in healthcare in at least my 20 years.

Companies Mentioned (5)

Arc Institute · ASUS, Lenovo, Dell, HPE · Epic, Microsoft Azure · Apple · GE HealthCare

Notable Quotes (2)

Every wet lab needs a dry lab. — Kimberly Powell @ 12:28

This is the most exciting time in healthcare in at least my 20 years. — Kimberly Powell @ 28:46

Key Topics

Accelerated Computing · Generative AI · Genomics · Drug Discovery · BioNeMo · AI Agents · Reasoning Models · Brain-Computer Interfaces · Physical AI · Medical Robotics · Medical Imaging · MONAI · Holoscan · Digital Health

Takeaways

  • NVIDIA is driving a massive transformation in healthcare through accelerated computing and AI.
  • The Evo 2 model represents a major leap in understanding biology, with a 1 million context length.
  • DGX Spark brings petaflop-scale AI supercomputing directly to the researcher’s desk.
  • The integration of ‘dry labs’ (AI) with ‘wet labs’ (physical) is essential for modern drug discovery.
  • AI agents with reasoning capabilities are automating complex workflows in research and clinical settings.
  • Physical AI, powered by platforms like Isaac and Holoscan, is enabling autonomous medical devices and robotics.
  • NVIDIA’s ecosystem of partners, from pharma to medical device manufacturers, is rapidly adopting these technologies.