NVIDIA Healthcare

Today we announce results from our deployment of NVIDIA Blackwell B300 GPUs in partnership with @nvidia and @nebiusai, a leading AI cloud provider for training and inference, achieving over 4x training speedups and more than 60% faster inference for our generative models powering drug discovery.

In our new blog post, we team up with @NVIDIAHealth to make it possible to map how drug effects cascades through gene regulatory networks.

Introducing PerturbAI. Today we announced our emergence from stealth with the release of the world’s largest in vivo CRISPR data engine, interrogating the effects of thousands of genetic perturbations across 8 million cells throughout the whole brain. This dataset represents a new category of biological data: organism-level, circuit-resolved causal genomics leading to novel targets and therapeutics. By combining scalable in vivo CRISPR perturbation with AI, we model biological systems at unprecedented resolution and simulate therapeutic interventions before committing to expensive downstream development. We’re grateful to our collaborators at @NVIDIAHealth and @10xGenomics for helping make this landmark dataset possible. Read More: perturb.ai/news #CRISPR #AI #DrugDiscovery #FunctionalGenomics #Biotech

Earlier this week at GTC, we announced our partnership with Nvidia. We will work with Nvidia to build strong, American open-source models that are at the frontier of scientific reasoning. These models will be essential for the US to compete with China on science in the coming decades. Jensen is committing to spend tens of billions of dollars developing open-source models, and we are excited to be a partner with them in figuring out how to benchmark, train and use those agents to accelerate scientific research. We have already open-sourced some of the work we have done with them, and are looking forward to open-sourcing more. There are few things today that are more important. See our blog post below, and watch the video to learn more, narrated by the man himself.

It has been great to collaborate with the @NVIDIAHealth team on adding Context Parallelism to the Boltz models! Announced in the keynote at #GTC, @NVIDIA showed you can now scale up Boltz-2 to predict structures up to 30k residues (up from ~2k)! 🤯 This opens the door to a number of very cool new applications, although performance starts to drop around 3-4k residues because the current models have only been trained on smaller crops. Try it out at: github.com/jwohlwend/bolt… 🧬

@ManifoldBio and @NVIDIAHealth announce a joint study validating Proteina-Complexa, NVIDIA's latest BioNeMo model for protein binder design.

LIVE IN 1 HOUR 🚨 Our #NVIDIAGTC Developer Community Livestream is about to start. Remember to use #ShowMeGTC to let us know what you want to see in real-time on stream. Start dropping your suggestions below or in chat. Join 👉 nvda.ws/4sYobzX

Scientific discovery shouldn’t drown in data. ⚡ Kosmos, built by @EdisonSci on NVIDIA, turns months of papers, code, and analysis into new hypotheses in a single day. This is what it looks like when AI becomes a real collaborator in the lab. Watch here ▶️ nvda.ws/4saUA6a

Read the full @huggingface blog nvda.ws/40EIt5j

Imagine a robot that can suture tissue as skillfully as a surgeon - turning AI from perception into precise, life‑saving action. Explore the three breakthroughs powering this leap: 1️⃣ Open‑H‑Embodiment - the first open, community‑driven dataset delivering standardized robot‑body, vision, force, kinematics, and sim‑to‑real data so AI can do rather than just perceive. 2️⃣ GR00T‑H - a Vision‑Language‑Action model trained on ~600 hours of Open‑H‑Embodiment that autonomously executes a full end‑to‑end suture in the SutureBot benchmark. 3️⃣ Cosmos‑H‑Surgical‑Simulator - a World Foundation Model fine‑tuned on Open‑H‑Embodiment that generates physically plausible surgical video directly from kinematic actions, closing the sim‑to‑real gap 30× faster than benchtop experiments. 🔗 👇

🏥 Discover how NVIDIA is revolutionizing healthcare with Project Rheo

📢📢 Proteina-Complexa 📢📢 Atomistic Binder Design with Generative Pretraining and Test-Time Compute + Experimental Validation at Scale ⭐️ Project page (research.nvidia.com/labs/genair/pr…) for: 📜 Method paper (ICLR 2026 Oral) 🧬 Wet lab paper 🛠️ Code & models 📁 Data 🧵 Thread (1/n)

Drug discovery has a scale problem. The complexes that govern signaling, regulation, and disease aren't single proteins, or even tidy pairs. They're sprawling assemblies of 10, 20, even 50+ subunits. The field is forced to sidestep that level of complexity, not for lack of interest, but for lack of tools. And yet, simulating cellular biology requires modeling the complexes and interactions these proteins form, not just individual proteins. NeoLink, @proximabio's proteomics platform, has experimentally mapped millions of protein-protein interactions across these higher-order complexes. Predicting their structures at this scale demanded something new. We built it with @NVIDIAHealth. By integrating NVIDIA's Fold-CP into our pipeline, we distribute computation for long biological sequences across multiple GPUs, removing the memory bottleneck that kept these complexes out of reach. New preprint out now with the NVIDIA BioNeMo team: bit.ly/4rDQnXA

At @GoogleDeepMind, we believe AI is the ultimate catalyst for science. 🧬 The best example of this has been the AlphaFold database (AFDB) of protein structure predictions which has been used free of cost by more than 3.3 millions researchers across the world! Today, in collaboration with @emblebi, @Nvidia and @SeoulNatlUni, we are expanding the database by adding millions of AI-predicted protein complex structures to the AlphaFold Database. To maximise global health impact, we’ve prioritised proteins that are important for understanding human health and disease, including homodimers from 20 of the most studied organisms, including humans, as well as the @WHO’S bacterial priority pathogens list. Read more here: embl.org/news/science-t…

Read tech blog nvda.ws/4sJbQzb

🏥 Don't teach robots inside hospitals - train the hospital in simulation first. Project Rheo is a new blueprint for Hospital Automation & Physical AI. See how you can help solve the data gap in healthcare robotics, helping make deployment scalable. ✅ GR00T VLA Models: Train Physical AI using Vision-Language-Action (VLA) architectures for intuitive, multi-modal robot control. ✅ Isaac Sim & Lab-Arena: Develop in high-fidelity digital twins to master complex loco-manipulation and precision bimanual tasks. ✅ Cosmos-H Synthetic Data: Scale training datasets exponentially with generative AI to bridge the domain gap between simulation and real-world hospitals. ✅ RL Post-Training: Use Reinforcement Learning (PPO) to refine high-precision stages, ensuring robust performance in chaotic clinical environments. #NVIDIAGTC

@IQVIA_global is integrating Agent Toolkit software into IQVIA.ai, its unified agentic AI platform, to help organizations scale AI responsibly across clinical, commercial, and real-world operations.

Lights, cameras, almost go time. 🎥 Tomorrow morning we kick off #NVIDIAGTC 2026 with the keynote pregame, GTC Live, starting at 8 a.m. PT. Join @saranormous (Conviction), @GavinSBaker (Atreides), @Alfred_Lin (Sequoia Capital), and Tiffany Janzen (TiffinTech) as they meet with 20 industry leaders to set the stage before Jensen Huang’s GTC 2026 keynote. Big conversations. Big announcements. One big week. 🔗 nvda.ws/4ryLqiL

📖 Read story nvda.ws/4sJvtr2 🗓️ Join her GTC Session nvda.ws/4tloqWH

It’s easy to get caught up in AI parameters and compute power, but @kpowgerade reminds us of the technology's true purpose: to tackle the world’s hardest problems. In a powerful new interview with @ALFANewsCo, Kimberly grounds the AI revolution in what matters most: human health. Don't miss her special address at #NVIDIAGTC next week! Details in 🔗👇