NVIDIA CEO Jensen Huang and Industry Visionaries to Unveil What’s Next in AI at GTC 2025

NVIDIA announced GTC 2025, the world’s premier AI conference, will return March 17-21 to San Jose, Calif. — bringing together the brightest minds in AI to showcase breakthroughs happening now in physical AI, agentic AI and scientific discovery. GTC will bring together 25,000 attendees in person — and 300,000 attendees virtually — for an in-depth look at the technologies shaping the future.

NVIDIA founder and CEO Jensen Huang will deliver the keynote from SAP Center on Tuesday, March 18, at 10 a.m. PT focused on AI and accelerated computing technologies changing the world. It will be livestreamed and available on demand at nvidia.com. Registration is not required to view the keynote online.

Onsite attendees can arrive at SAP Center early to enjoy a live pregame show hosted by the “Acquired” podcast and other surprise festivities. Virtual attendees can catch the pregame show live online.

“AI is pushing the limits of what’s possible — turning yesterday’s dreams into today’s reality,” Huang said. “GTC brings together the brightest scientists, engineers, developers and creators to imagine and build a better future. Come and be first to see the new advances in NVIDIA computing and breakthroughs in AI, robotics, science and the arts that will transform industries and society.”

AI is here, and it’s mainstream — powering the everyday brands that shape people’s lives. At GTC, some of the world’s largest companies, groundbreaking startups and leading academic minds will convene to explore the transformative impact of AI across industries.

With over 1,000 sessions, 2,000 speakers and nearly 400 exhibitors, GTC will showcase how NVIDIA’s AI and accelerated computing platforms tackle the world’s biggest and toughest challenges — spanning climate research to healthcare, cybersecurity, humanoid robotics, autonomous vehicles and more. From large language models and physical AI to cloud computing and scientific discovery, NVIDIA’s full-stack platform is driving the next industrial revolution.

At the conference, attendees can also look forward to curated experiences, including dozens of demos spanning every industry, hands-on training, autonomous vehicle exhibits and rides, and a new GTC Night Market featuring street food and wares from 20 local vendors and artisans.

More than 900 organizations will participate, including Accenture, Adobe, Arm, Airbnb, Amazon Web Services (AWS), BMW Group, The Coca-Cola Company, CoreWeave, Dell Technologies, Disney Research, Field AI, Ford, Foxconn, Google Cloud, Kroger, Lowe’s, Mercedes-Benz, Meta, Microsoft, MLB, NFL, OpenAI, Oracle Cloud Infrastructure, Pfizer, Rockwell Automation, Salesforce, Samsung, ServiceNow, SoftBank, TSMC, Uber, Volvo, Volkswagen, Wayve and Zoox.

Quantum Day Arrives

NVIDIA will host its first Quantum Day at GTC on March 20. The event will bring together the global quantum computing community and key industry figures.

Leaders from the quantum computing industry will join a panel with Huang from 10 a.m. to 12 p.m. PT, shedding light on the current state and future of quantum computing. The panel will be livestreamed and available on demand, and feature pioneers in quantum computing.

AI Training and Certification for Developers

NVIDIA is training the workforce of the future to equip them with critical skills for navigating and leading in an AI-driven future.

GTC attendees can participate in more than 80 hands-on instructor-led workshops and training labs provided by NVIDIA Training.

For the first time, onsite attendees can take certification exams for free — gaining a tremendous opportunity to validate their AI and accelerated computing skills and advance their careers.

In addition, new professional certifications will be available in accelerated data science and AI networking, as well as workshops in generative AI, agentic AI and accelerated computing with CUDA® C++.

 

Explore How RTX AI PCs and Workstations Supercharge AI Development at NVIDIA GTC 2025

 

Generative AI is redefining computing, unlocking new ways to build, train and optimize AI models on PCs and workstations. From content creation and large and small language models to software development, AI-powered PCs and workstations are transforming workflows and enhancing productivity.

At GTC 2025, running March 17–21 in the San Jose Convention Center, experts from across the AI ecosystem will share insights on deploying AI locally, optimizing models and harnessing cutting-edge hardware and software to enhance AI workloads — highlighting key advancements in RTX AI PCs and workstations.

Develop and Deploy on RTX
RTX GPUs are built with specialized AI hardware called Tensor Cores that provide the compute performance needed to run the latest and most demanding AI models. These high-performance GPUs can help build digital humans, chatbots, AI-generated podcasts and more.

With more than 100 million GeForce RTX and NVIDIA RTX™ GPUs users, developers have a large audience to target when new AI apps and features are deployed. In the session “Build Digital Humans, Chatbots, and AI-Generated Podcasts for RTX PCs and Workstations,” Annamalai Chockalingam, senior product manager at NVIDIA, will showcase the end-to-end suite of tools developers can use to streamline development and deploy incredibly fast AI-enabled applications.

Model Behavior
Large language models (LLMs) can be used for an abundance of use cases — and scale to tackle complex tasks like writing code or translating Japanese into Greek. But since they’re typically trained with a wide spectrum of knowledge for broad applications, they may not be the right fit for specific tasks, like nonplayer character dialog generation in a video game. In contrast, small language models balance need with reduced size, maintaining accuracy while running locally on more devices.

In the session “Watch Your Language: Create Small Language Models That Run On-Device,” Oluwatobi Olabiyi, senior engineering manager at NVIDIA, will present tools and techniques that developers and enthusiasts can use to generate, curate and distill a dataset — then train a small language model that can perform tasks designed for it.

Maximizing AI Performance on Windows Workstations
Optimizing AI inference and model execution on Windows-based workstations requires strategic software and hardware tuning due to diverse hardware configurations and software environments. The session “Optimizing AI Workloads on Windows Workstations: Strategies and Best Practices,” will explore best practices for AI optimization, including model quantization, inference pipeline enhancements and hardware-aware tuning.

A team of NVIDIA software engineers will also cover hardware-aware optimizations for ONNX Runtime, NVIDIA TensorRT and llama.cpp, helping developers maximize AI efficiency across GPUs, CPUs and NPUs.

Advancing Local AI Development
Building, testing and deploying AI models on local infrastructure ensures security and performance even without a connection to cloud-based services. Accelerated with NVIDIA RTX GPUs, Z by HP’s AI solutions provide the tools needed to develop AI on premises while maintaining control over data and IP

Dell Pro Max and NVIDIA: Unleashing the Future of AI Development: This session introduces Dell Pro Max PCs, performance laptops and desktops for professionals, powered by NVIDIA RTX GPUs. Discover how this powerful duo can help jumpstart AI initiatives and transform the way AI developers, data scientists, creators and power users innovate.
Develop and Observe Gen AI On-Prem With Z by HP GenAI Lab and AI Studio: This session demonstrates how Z by HP solutions simplify local model training and deployment, harnessing models in the NVIDIA NGC catalog and Galileo evaluation technology to refine generative AI projects securely and efficiently.
Supercharge Gen AI Development With Z by HP GenAI Lab and AI Studio: This session explores how Z by HP’s GenAI Lab and AI Studio enable on-premises LLM development while maintaining complete data security and control. Learn how these tools streamline the entire AI lifecycle, from experimentation to deployment, while integrating models available in the NVIDIA NGC catalog for collaboration and workflow efficiency.
Developers and enthusiasts can get started with AI development on RTX AI PCs and workstations using NVIDIA NIM microservices. Rolling out today, the initial public beta release includes the Llama 3.1 LLM, NVIDIA Riva Parakeet for automatic speech recognition (ASR), and YOLOX for computer vision.

NIM microservices are optimized, prepackaged models for generative AI. They span modalities important for PC development, and are easy to download and connect to via industry-standard application programming interfaces.

NVIDIA Launches NIM Microservices for Generative AI in Japan, Taiwan

NVIDIA Launches NIM Microservices for Generative AI in Japan, Taiwan

Nations around the world are pursuing sovereign AI to produce artificial intelligence using their own computing infrastructure, data, workforce and business networks to ensure AI systems align with local values, laws and interests.

In support of these efforts, NVIDIA announced the availability of four new NVIDIA NIM microservices that enable developers to more easily build and deploy high-performing generative AI applications.

The microservices support popular community models tailored to meet regional needs. They enhance user interactions through accurate understanding and improved responses based on local languages and cultural heritage.

In the Asia-Pacific region alone, generative AI software revenue is expected to reach $48 billion by 2030 — up from $5 billion this year, according to ABI Research.

Llama-3-Swallow-70B, trained on Japanese data, and Llama-3-Taiwan-70B, trained on Mandarin data, are regional language models that provide a deeper understanding of local laws, regulations and other customs.

Training a large language model (LLM) on regional languages enhances the effectiveness of its outputs by ensuring more accurate and nuanced communication, as it better understands and reflects cultural and linguistic subtleties.

Nations worldwide — from Singapore, the United Arab Emirates, South Korea and Sweden to France, Italy and India — are investing in sovereign AI infrastructure.

The new NIM microservices allow businesses, government agencies and universities to host native LLMs in their own environments, enabling developers to build advanced copilots, chatbots and AI assistants.

High-Tech Highways: India Uses NVIDIA Accelerated Computing to Ease Tollbooth Traffic

India is home to the globe’s second-largest road network, spanning nearly 4 million miles, and has over a thousand tollbooths, most of them run manually.

Traditional booths like these, wherever in the world they’re deployed, can contribute to massive traffic delays, long commute times and serious road congestion.

To help automate tollbooths across India, Calsoft, an Indian-American technology company, helped implement a broad range of NVIDIA technologies integrated with the country’s dominant payment system, known as the unified payments interface, or UPI, for a client.

Manual tollbooths demand more time and labor compared to automated ones. However, automating India’s toll systems faces an extra complication: the diverse range of license plates.

India’s non-standardized plates pose a significant challenge to the accuracy of automatic number plate recognition (ANPR) systems. Any implementation would need to address these plate variations, which include divergent color, sizing, font styles and placement upon vehicles, as well as many different languages.

The solution Calsoft helped build automatically reads passing vehicle plates and charges the associated driver’s UPI account. This approach reduces the need for manual toll collection and is a massive step toward addressing traffic in the region.

Automation in Action

As part of a pilot program, this solution has been deployed in several leading metropolitan cities. The solution provides about 95% accuracy in its ability to read plates through the use of an ANPR pipeline that detects and classifies the plates as they roll through tollbooths.

NVIDIA’s technology has been crucial in this effort, according to Vipin Shankar, senior vice president of technology at Calsoft. “Particularly challenging was night-time detection,” he said. “Another challenge was model accuracy improvement on pixel distortions due to environmental impacts like fog, heavy rains, reflections due to bright sunshine, dusty winds and more.”

The solution uses NVIDIA Metropolis to track and detect vehicles throughout the process. Metropolis is an application framework, a set of developer tools and a partner ecosystem that brings visual data and AI together to improve operational efficiency and safety across a range of industries.

Calsoft engineers used NVIDIA Triton Inference Server software to deploy and manage their AI models. The team also used the NVIDIA DeepStream software development kit to build a real-time streaming platform. This was key for processing and analyzing data streams efficiently, incorporating advanced capabilities such as real-time object detection and classification.

Calsoft uses NVIDIA hardware, including NVIDIA Jetson edge AI modules and NVIDIA A100 Tensor Core GPUs in its AI solutions. Calsoft’s tollbooth solution is also scalable, meaning it’s designed to accommodate future growth and expansion needs, and can better ensure sustained performance and adaptability as traffic conditions evolve.

AI Chases the Storm: New NVIDIA Research Boosts Weather Prediction, Climate Simulation

As hurricanes, tornadoes and other extreme weather events occur with increased frequency and severity, it’s more important than ever to improve and accelerate climate research and prediction using the latest technologies.

Amid peaks in the current Atlantic hurricane season, NVIDIA Research today announced a new generative AI model, dubbed StormCast, for emulating high-fidelity atmospheric dynamics. This means the model can enable reliable weather prediction at mesoscale — a scale larger than storms but smaller than cyclones — which is critical for disaster planning and mitigation.

Detailed in a paper written in collaboration with the Lawrence Berkeley National Laboratory and the University of Washington, StormCast arrives as extreme weather phenomena are taking lives, destroying homes and causing more than $150 billion in damage annually in the U.S. alone.

It’s just one example of how generative AI is supercharging thundering breakthroughs in climate research and actionable extreme weather prediction, helping scientists tackle challenges of the highest stakes: saving lives and the world.

NVIDIA Earth-2 — a digital twin cloud platform that combines the power of AI, physical simulations and computer graphics — enables simulation and visualization of weather and climate predictions at a global scale with unprecedented accuracy and speed.In Taiwan, for example, the National Science and Technology Center for Disaster Reduction predicts fine-scale details of typhoons using CorrDiff, an NVIDIA generative AI model offered as part of Earth-2.

CorrDiff can super-resolve 25-kilometer-scale atmospheric data by 12.5x down to 2 kilometers — 1,000x faster and using 3,000x less energy for a single inference than traditional methods.

That means the center’s potentially lifesaving work, which previously cost nearly $3 million on CPUs, can be accomplished using about $60,000 on a single system with an NVIDIA H100 Tensor Core GPU. It’s a massive reduction that shows how generative AI and accelerated computing increase energy efficiency and lower costs.

The center also plans to use CorrDiff to predict downwash — when strong winds funnel down to street level, damaging buildings and affecting pedestrians — in urban areas.

Now, StormCast adds hourly autoregressive prediction capabilities to CorrDiff, meaning it can predict future outcomes based on past ones.

A Global Impact From a Regional Focus

Global climate research begins at a regional level.

Physical hazards of weather and climate change can vary dramatically on regional scales. But reliable numerical weather prediction at this level comes with substantial computational costs. This is due to the high spatial resolution needed to represent the underlying fluid-dynamic motions at mesoscale.

Regional weather prediction models — often referred to as convection-allowing models, or CAMs — have traditionally forced researchers to face varying tradeoffs in resolution, ensemble size and affordability.

CAMs are useful to meteorologists for tracking the evolution and structure of storms, as well as for monitoring its convective mode, or how a storm is organized when it forms. For example, the likelihood of a tornado is based on a storm’s structure and convective mode.

NVIDIA researchers trained StormCast on approximately three-and-a-half years of NOAA climate data from the central U.S., using NVIDIA accelerated computing to speed calculations.