NVIDIA kicked off SC24 in Atlanta with a wave of AI and supercomputing instruments set to revolutionize industries like biopharma and local weather science.
The bulletins, delivered by NVIDIA founder and CEO Jensen Huang and Vice President of Accelerated Computing Ian Buck, are rooted within the firm’s deep historical past in reworking computing.
“Supercomputers are amongst humanity’s most significant devices, driving scientific breakthroughs and increasing the frontiers of data,” Huang stated. “Twenty-five years after creating the primary GPU, we now have reinvented computing and sparked a brand new industrial revolution.”
NVIDIA’s journey in accelerated computing started with CUDA in 2006 and the primary GPU for scientific computing, Huang stated.
Milestones like Tokyo Tech’s Tsubame supercomputer in 2008, the Oak Ridge Nationwide Laboratory’s Titan supercomputer in 2012 and the AI-focused NVIDIA DGX-1 delivered to OpenAI in 2016 spotlight NVIDIA’s transformative function within the subject.
“Since CUDA’s inception, we’ve pushed down the price of computing by a millionfold,” Huang stated. “For some, NVIDIA is a computational microscope, permitting them to see the impossibly small. For others, it’s a telescope exploring the unimaginably distant. And for a lot of, it’s a time machine, letting them do their life’s work inside their lifetime.”
At SC24, NVIDIA’s bulletins spanned instruments for next-generation drug discovery, real-time local weather forecasting and quantum simulations.
Central to the corporate’s developments are CUDA-X libraries, described by Huang as “the engines of accelerated computing,” which energy the whole lot from AI-driven healthcare breakthroughs to quantum circuit simulations.
Huang and Buck highlighted examples of real-world affect, together with Nobel Prize-winning breakthroughs in neural networks and protein prediction, powered by NVIDIA expertise.
“AI will speed up scientific discovery, reworking industries and revolutionizing each one of many world’s $100 trillion markets,” Huang stated.
CUDA-X Libraries Energy New Frontiers
At SC24, NVIDIA introduced the brand new cuPyNumeric library, a GPU-accelerated implementation of NumPy, designed to supercharge purposes in knowledge science, machine studying and numerical computing.
With over 400 CUDA-X libraries, together with cuDNN for deep studying and cuQuantum for quantum circuit simulations, NVIDIA continues to steer in enhancing computing capabilities throughout numerous industries.
Actual-Time Digital Twins With Omniverse Blueprint
NVIDIA unveiled the NVIDIA Omniverse Blueprint for real-time computer-aided engineering digital twins, a reference workflow designed to assist builders create interactive digital twins for industries like aerospace, automotive, vitality and manufacturing.
Constructed on NVIDIA acceleration libraries, physics-AI frameworks and interactive, bodily based mostly rendering, the blueprint accelerates simulations by as much as 1,200x, setting a brand new customary for real-time interactivity.
Early adopters, together with Siemens, Altair, Ansys and Cadence, are already utilizing the blueprint to optimize workflows, reduce prices and convey merchandise to market quicker.
Quantum Leap With CUDA-Q
NVIDIA’s deal with real-time, interactive applied sciences extends throughout fields, from engineering to quantum simulations.
In partnership with Google, NVIDIA’s CUDA-Q now powers detailed dynamical simulations of quantum processors, lowering weeks-long calculations to minutes.
Buck defined that with CUDA-Q, builders of all quantum processors can carry out bigger simulations and discover extra scalable qubit designs.
AI Breakthroughs in Drug Discovery and Chemistry
With the open-source launch of BioNeMo Framework, NVIDIA is advancing AI-driven drug discovery as researchers acquire highly effective instruments tailor-made particularly for pharmaceutical purposes.
BioNeMo accelerates coaching by 2x in comparison with different AI software program, enabling quicker improvement of lifesaving therapies.
NVIDIA additionally unveiled DiffDock 2.0, a breakthrough software for predicting how medicine bind to focus on proteins — essential for drug discovery.
Powered by the brand new cuEquivariance library, DiffDock 2.0 is 6x quicker than earlier than, enabling researchers to display screen thousands and thousands of molecules with unprecedented pace and accuracy.
And the NVIDIA ALCHEMI NIM microservice, NVIDIA introduces generative AI to chemistry, permitting researchers to design and consider novel supplies with unbelievable pace.
Scientists begin by defining the properties they need — like energy, conductivity, low toxicity and even coloration, Buck defined.
A generative mannequin suggests 1000’s of potential candidates with the specified properties. Then the ALCHEMI NIM types candidate compounds for stability by fixing for his or her lowest vitality states utilizing NVIDIA Warp.
This microservice is a game-changer for supplies discovery, serving to builders sort out challenges in renewable vitality and past.
These improvements display how NVIDIA is harnessing AI to drive breakthroughs in science, reworking industries and enabling quicker options to international challenges.
Earth-2 NIM Microservices: Redefining Local weather Forecasts in Actual Time
Buck additionally introduced two new microservices — CorrDiff NIM and FourCastNet NIM — to speed up local weather change modeling and simulation outcomes by as much as 500x within the NVIDIA Earth-2 platform.
Earth-2, a digital twin for simulating and visualizing climate and local weather situations, is designed to empower climate expertise firms with superior generative AI-driven capabilities.
These instruments ship higher-resolution and extra correct predictions, enabling the forecasting of maximum climate occasions with unprecedented pace and vitality effectivity.
With pure disasters inflicting $62 billion in insured losses within the first half of this 12 months — 70% greater than the 10-year common — NVIDIA’s improvements handle a rising want for exact, real-time local weather forecasting. These instruments spotlight NVIDIA’s dedication to leveraging AI for societal resilience and local weather preparedness.
Increasing Manufacturing With Foxconn Collaboration
As demand for AI programs just like the Blackwell supercomputer grows, NVIDIA is scaling manufacturing by means of new Foxconn amenities within the U.S., Mexico and Taiwan.|
Foxconn is constructing the manufacturing and testing amenities utilizing NVIDIA Omniverse to convey up the factories as quick as doable.
Scaling New Heights With Hopper
NVIDIA additionally introduced the normal availability of the NVIDIA H200 NVL, a PCIe GPU based mostly on the NVIDIA Hopper structure optimized for low-power, air-cooled knowledge facilities.
The H200 NVL affords as much as 1.7x quicker massive language mannequin inference and 1.3x extra efficiency on HPC purposes, making it ultimate for versatile knowledge middle configurations.
It helps quite a lot of AI and HPC workloads, enhancing efficiency whereas optimizing current infrastructure.
And the GB200 Grace Blackwell NVL4 Superchip integrates 4 NVIDIA NVLink-connected Blackwell GPUs unified with two Grace CPUs over NVLink-C2C, Buck stated. It supplies as much as 2x efficiency for scientific computing, coaching and inference purposes over the prior era. |
The GB200 NVL4 superchip might be out there within the second half of 2025.
The speak wrapped up with an invite to attendees to go to NVIDIA’s sales space at SC24 to work together with numerous demos, together with James, NVIDIA’s digital human, the world’s first real-time interactive wind tunnel and the Earth-2 NIM microservices for local weather modeling.
Be taught extra about how NVIDIA’s improvements are shaping the way forward for science at SC24.
