| November 20, 2024
At SC24 in Atlanta, NVIDIA’s founder and CEO, Jensen Huang, unveiled a series of groundbreaking AI and supercomputing tools designed to transform industries such as biopharma, climate science, and materials discovery.
Huang emphasized how supercomputers are driving humanity’s most critical advancements, stating, “Twenty-five years after creating the first GPU, we have reinvented computing and sparked a new industrial revolution.”
The announcements at SC24 are rooted in NVIDIA’s long history of accelerating computing, from the launch of CUDA in 2006 to recent innovations powering breakthroughs in neural networks, protein prediction, and real-time simulations.
1. Advancing AI With CUDA-X Libraries
NVIDIA’s CUDA-X libraries, described as the “engines of accelerated computing,” are pivotal to the company’s technological strides. New additions include:
- cuPyNumeric: A GPU-accelerated implementation of NumPy, ideal for data science and machine learning tasks.
- cuQuantum: Tools for simulating quantum circuits efficiently.
These libraries power critical applications across healthcare, climate modeling, and quantum simulations.
2. Omniverse Blueprint for Digital Twins
NVIDIA introduced the Omniverse Blueprint, a workflow enabling developers to build real-time digital twins for industries like aerospace, automotive, and energy.
- Simulations are accelerated by up to 1,200x, leveraging NVIDIA’s physics-AI frameworks and interactive rendering.
- Companies like Siemens and Altair are already optimizing workflows and reducing costs using this technology.
3. Real-Time Climate Modeling With Earth-2 NIM Microservices
In response to increasing natural disasters, NVIDIA announced two new microservices — CorrDiff NIM and FourCastNet NIM — as part of its Earth-2 platform.
- These tools accelerate climate modeling and simulation results by up to 500x, providing precise predictions of extreme weather events.
- They aim to mitigate the financial and human impact of climate-related disasters, which caused $62 billion in insured losses in the first half of 2024 alone.
4. AI-Driven Drug Discovery
NVIDIA unveiled significant advancements in biopharma with tools like:
- BioNeMo Framework: Open-source software that speeds up AI model training for drug discovery by 2x.
- DiffDock 2.0: A powerful tool for predicting drug-protein interactions, now 6x faster thanks to the new cuEquivariance library.
The ALCHEMI NIM microservice uses generative AI to design new materials by predicting desired properties like strength and low toxicity.
5. Accelerating Quantum Computing With CUDA-Q
In partnership with Google, NVIDIA’s CUDA-Q enables faster simulations of quantum processors. Developers can now perform large-scale simulations in minutes, significantly enhancing quantum research and development.
6. Scaling AI Production With Foxconn
To meet growing demand, NVIDIA is collaborating with Foxconn to establish new production facilities in the U.S., Mexico, and Taiwan. These factories, designed with NVIDIA’s Omniverse, will expedite manufacturing and testing.
7. Next-Gen Supercomputing With Hopper Architecture
NVIDIA announced the availability of its H200 NVL GPU, based on the Hopper architecture, designed for low-power, air-cooled data centers.
- Offers up to 1.7x faster large language model inference.
- The GB200 Grace Blackwell NVL4 Superchip doubles performance for scientific computing and AI applications.
Driving the Future of Scientific Discovery
Huang wrapped up the keynote by emphasizing the transformative potential of NVIDIA’s innovations, stating, “AI will accelerate scientific discovery, transforming industries and revolutionizing every one of the world’s $100 trillion markets.”
Attendees were invited to NVIDIA’s booth at SC24 to interact with cutting-edge demos, including:
- James, NVIDIA’s real-time digital human.
- Earth-2 microservices, showcasing real-time climate forecasting.
These advancements highlight NVIDIA’s commitment to using AI to solve global challenges, from advancing healthcare to combating climate change.
