10 NVIDIA GTC 2025 Announements that You Must Know

I still remember last year’s NVIDIA GTC, Jensen Huang with its visionary approach along with a touch of humour introduced the developers with promises to redefine technology. From Blackwell architecture, Generative AI with NIM, GB200 AI Chip to Project Groot and other things, we got a glimpse into the future of technology. Now the future is here! Today at NVIDIA GTC 2025, we got a lot of updates. This year’s event highlighted not only the pace of the AI revolution but also how NVIDIA is transforming the very essence of computing.

Jensen Huang Said:

“This is the computer of the age of AI. This is what computers should look like. And this is what computers will run in the future.”

The future, without a doubt, is here.

Key Takeaways from the NVIDIA GTC 2025

• AI Real-Time CGI Rendering: Demonstrated the capabilities of real-time AI-driven CGI rendering.
• GM & Nvidia Autonomous Vehicle Partnership: Announced a collaboration between GM and Nvidia focused on autonomous vehicle technology.
• Blackwell GPU Data Centers: Introduced Blackwell GPU data centers, highlighting advancements in data processing and storage.
• Nvidia Blackwell System: Unveiled the Nvidia Blackwell System, showcasing its innovative features and applications.
• Nvidia Dynamo: Presented Nvidia Dynamo, a new technology aimed at enhancing AI workflows and performance.
• DGX Station: Showcased the DGX Station, a powerful AI computing platform designed for enterprise use.
• Nvidia Robotics: Discussed advancements in Nvidia’s robotics division, emphasizing AI integration and innovation.
• Nvidia Omniverse: Highlighted the Nvidia Omniverse platform, focusing on its capabilities for virtual collaboration and simulation.
• Partnerships with Disney and Google DeepMind: Announced strategic partnerships with Disney and Google DeepMind, aimed at advancing AI and technology integration.

AI Real-Time CGI Rendering

GeForce brought Cuda to the world, it enabled AI and AI is now coming back to revolutionize computer graphics.

GeForce and the AI Revolution in Graphics

NVIDIA’s journey began with GeForce, and 25 years later, it continues to redefine computer graphics. The introduction of the GeForce 5090 highlights remarkable improvements:

• 30% smaller in volume than the previous generation.
• 30% better energy dissipation, enhancing efficiency.
• Blackwell Generation
• AI-powered rendering: For every pixel that is mathematically rendered, AI infers 15 additional pixels, ensuring real-time, 100% path-traced graphics with temporal stability.

NVIDIA showcased significant advancements in real-time AI-driven computer-generated imagery (CGI) rendering. By leveraging artificial intelligence, the company has enhanced the efficiency and quality of computer graphics, enabling more realistic and temporally stable images. This technology allows for the prediction and rendering of pixels with remarkable precision, revolutionizing the way visual content is created and experienced.

Jensen Huang’s Favourite Slide of GTC 2025

At GTC 2025, NVIDIA CEO Jensen Huang highlighted a pivotal slide, calling it his favourite. He showcased the CUDA-X libraries, a collection of acceleration frameworks driving advancements in science and engineering, all powered by GH200 and GB200 Superchips.

Huang emphasized that CUDA-X libraries are the foundation of GTC itself, reflecting two decades of relentless innovation in accelerating computing across various domains. It represents the evolution from a single library to a vast ecosystem supporting AI, physics, quantum computing, biology, lithography, and beyond.

Key Highlights from the Slide

• cuPyNumeric: A zero-change, drop-in acceleration for NumPy, the most downloaded Python library, enhancing numerical computing.
• cuLitho: Revolutionizing computational lithography, a critical technology in semiconductor manufacturing, already embraced by TSMC, Samsung, ASML, and major EDA leaders.
• Aerial Sionna: Transforming GPUs into 5G radios, leveraging AI-driven signal processing to optimize wireless networks.
• cuQuantum & CUDA-Q: Accelerating quantum computing and simulation, paving the way for the future of computing.
• Parabricks & MONAI: Powering gene sequencing and medical imaging, revolutionizing healthcare and life sciences.
• Physics and AI Frameworks: From multi-physics simulations to deep learning frameworks, enabling breakthroughs across industries.

The Future: Computational Factories

The concept is clear: in the AI-driven future, every industry will need two factories:

• One is for building physical products (cars, chips, smart devices).
• One for computing the intelligence behind them (AI models, simulations, and optimizations).

This applies to every sector:

• Automotive → Car manufacturing + AI for self-driving.
• Semiconductors → Chip fabrication + computational lithography.
• Consumer Electronics → Smart speakers + AI models for voice recognition.

Read more about it here: Nvidia Cuda x Libraries

GM & NVIDIA Autonomous Vehicle Partnership

NVIDIA GTC 2025 announced a strategic partnership with General Motors (GM) to integrate AI across all aspects of vehicle manufacturing and operation:

• AI for Manufacturing: Optimizing production processes.
• AI for Enterprise: Enhancing vehicle design, simulation, and workplace productivity.
• AI in Vehicles: Advanced infrastructure to power the next generation of autonomous cars.

The Big Shift

With this collaboration, NVIDIA is not just supplying technology but actively shaping the AI-driven transformation of the automotive industry.

A strategic collaboration between General Motors (GM) and NVIDIA focuses on the development of autonomous vehicle technology. This partnership aims to integrate NVIDIA’s AI capabilities into GM’s vehicles, enhancing manufacturing processes, enterprise operations, and in-car AI systems. The collaboration signifies a major step toward the widespread adoption of self-driving cars, with NVIDIA’s technology playing a crucial role in advancing vehicle autonomy and safety.

AI Safety: The HALOS Initiative

Jensen said, “One of the lesser-discussed yet critical areas of innovation at NVIDIA is automotive safety“.

The company’s HALOS program ensures:

• Safety assessment of all software: Every 7 million lines of code are reviewed.
• Third-party validation: Ensuring transparency, diversity, and explainability in AI systems.
• End-to-end safety: From chip design to software algorithms, NVIDIA takes a holistic approach to making AI safer for automotive applications.

This level of commitment to AI safety ensures reliability, making AI-powered self-driving systems more transparent, predictable, and trustworthy.

Enterprise AI Adoption: Full-stack AI solutions, AI-powered workforce.

Huang forecasted a transformative shift in computing, with AI revolutionizing everything from hardware architecture to software applications. AI-driven agents will seamlessly integrate into business operations, and Nvidia is constructing the foundation to support this evolution.

• A new workforce of 10 billion digital AI agents is on the horizon.
• By the end of the year, AI will assist 100% of Nvidia’s operations.
• Traditional programming will give way to AI-powered coding.
• This isn’t merely about substituting human workers—it’s about empowering businesses to scale intelligence in unprecedented ways.

NVIDIA Blackwell System: The “Beauty”

NVIDIA’s Blackwell architecture marks a fundamental shift in computing, designed to power the next generation of AI supercomputers. With unprecedented scale, efficiency, and performance, Blackwell introduces a disaggregated computing architecture that moves beyond traditional designs.

At its core, Blackwell is engineered to handle exascale AI workloads, providing trillions of operations per second, with massive memory bandwidth and liquid cooling technology—a game changer for AI data centers.

1. The Evolution from HGX to Blackwell

HGX: The Foundation of AI Computing

Before Blackwell, NVIDIA HGX systems led the AI revolution by integrating multiple GPUs into a single unit for AI and HPC (High-Performance Computing) workloads.

• HGX System:
  • Consisted of eight GPUs interconnected via NVLink.
  • Connected to CPU shelves via PCI Express.
  • Scaled up using Infiniband, forming an AI supercomputer.

Blackwell’s Major Leap Forward

Blackwell doubles down on performance and efficiency by:

• Integrating two GPUs into one package: Each Blackwell package houses two GPUs for increased processing power.
• Disaggregating the NVLink system: NVLink is no longer embedded within the GPU enclosure but is now a centralized switch.
• Scaling up computing capacity: Allowing every GPU to communicate at full bandwidth simultaneously.

NVIDIA rearchitected how AI supercomputers scale, moving from a single-unit model to a disaggregated network of GPUs—significantly improving performance and flexibility.

2. NVLink Switch: The Heart of Blackwell’s Disaggregated Design

One of Blackwell’s defining features is the introduction of the NVLink switch, the highest-performance GPU switch ever built.

Why NVLink Switch Matters

• Previously, GPUs communicated within a single enclosure.
• Now, NVLink is extracted and centralized, creating a switch-based interconnect.
• This allows every GPU to talk to every other GPU at full bandwidth, simultaneously.

The Scale-Up Approach

• 18 NVLink switches are placed across nine switch trays, enabling unprecedented AI workloads.
• Computing is now in a single, unified rack, reducing complexity.
• Eliminates bottlenecks, allowing seamless AI model training at an industrial scale.

The disaggregation of NVLink enables higher bandwidth, more scalability, and faster GPU-to-GPU communication than ever before.

3. Liquid Cooling: The Key to High-Density AI Computing

To accommodate the massive power requirements of Blackwell GPUs, NVIDIA has transitioned from air cooling to liquid cooling, enabling unparalleled compute density.

Why Liquid Cooling?

1. Traditional air-cooled systems → Struggled with increasing heat output as AI models became more powerful.
2. Blackwell’s liquid-cooled architecture:
   • Compresses computing power into a single rack.
   • Reduces energy consumption while maximizing efficiency.
   • Enables one exaFLOP of AI computation per rack—a milestone in computing.

The Numbers That Matter

• From 60,000 components per AI system: To 600,000 components per rack.
• From air-cooled racks: To fully liquid-cooled racks consuming 120kW.
• Memory bandwidth of 570 terabytes per second: Accelerating AI model training.

By integrating liquid cooling, NVIDIA has pushed the boundaries of AI computing, allowing denser, more power-efficient supercomputers that can train trillion-parameter AI models with ease.

4. Exascale AI Computing: A New Performance Benchmark

Blackwell is built to power the AI factories of the future, with performance benchmarks that are off the charts:

• One exaFLOP (ExaFLOPS) per rack: Equivalent to one million trillion floating-point operations per second.
• Trillions of calculations per second, enabling breakthroughs in AI, deep learning, and scientific computing.
• Memory bandwidth of 570 terabytes per second, ensuring data moves at unprecedented speeds.

Real-World Applications

• AI Model Training: Training large language models (LLMs) in record time.
• Autonomous Systems: Enabling real-time decision-making for self-driving cars and robotics.
• Scientific Research: Powering breakthroughs in climate modeling, medical research, and quantum simulations.

Blackwell represents the most extreme scale-up ever in AI computing, setting a new standard for high-performance AI training and inference.

The Future of AI Supercomputing with Blackwell

With Blackwell, NVIDIA is redefining how AI supercomputers are built. It introduces:

• A fully disaggregated computing model: Allowing flexibility and massive scalability.
• NVLink switch architecture: Enabling seamless, full-bandwidth communication between GPUs.
• Liquid cooling: Allowing AI supercomputers to be denser and more power-efficient.
• Exascale AI performance: Enabling the next generation of AI breakthroughs.

With Blackwell, NVIDIA isn’t just scaling AI computing—it’s revolutionizing how AI supercomputers are designed, built, and deployed. As AI models grow exponentially, Blackwell is the infrastructure that will power the future of artificial intelligence.

NVIDIA Dynamo

“Dynamo was the first instrument that sparked the last Industrial Revolution—the revolution of energy. Water comes in, electricity comes out. It’s pretty fantastic. You take water, light it on fire, turn it to steam, and what emerges is this invisible force of immense value. It took another 80 years to transition to alternating current, but Dynamo is where it all began. That’s why NVIDIA named this incredibly complex piece of software—its AI factory operating system—NVIDIA Dynamo.”

NVIDIA introduced Dynamo, an open-source inference software system designed to serve as the operating system for AI factories. Dynamo facilitates the scaling and acceleration of AI models, optimizing the deployment and management of AI workflows across various platforms. By streamlining AI operations, Dynamo enables organizations to harness the full potential of AI technologies, improving efficiency and reducing operational complexities.

AI is changing the nature of computing, requiring a new operating system designed specifically for AI workloads. Enter NVIDIA Dynamo—an AI-centric OS for AI Factories.

How does it compare to Traditional OS?

• Traditional Data Centers: Run enterprise software using VMware.
• Future AI Factories: Run AI agents using Dynamo instead of VMware.

Dynamo represents a paradigm shift: Data centres of the past processed traditional enterprise workloads, while future AI Factories will be optimized for AI model training and inference.

DGX Station

The DGX Station is a powerful AI computing platform tailored for enterprise use. Equipped with 20 petaFLOPS of performance and 72 CPU cores, it provides a robust infrastructure for data scientists and researchers, supporting complex AI computations and workflows. The DGX Station enables organizations to conduct advanced AI research and development, facilitating innovation and accelerating time-to-market for AI-driven solutions.

NVIDIA introduced the DGX Station and DGX Spark, defining what an AI-powered PC should look like:

• 20 petaFLOPS of AI computation.
• 72 CPU cores.
• Chip-to-chip high-bandwidth memory (HBM).
• PCI Express slots for expansion.

Availability

These workstations will be available from leading OEMs like HP, Dell, Lenovo, and Asus, ensuring accessibility to AI researchers and data scientists worldwide. Just as NVIDIA redefined GPUs, it is now redefining personal computing for the AI era.

NVIDIA Robotics and Omniverse

NVIDIA’s advancements in robotics were highlighted by the introduction of “Blue,” a robot developed in collaboration with Disney Research and Google DeepMind. This partnership led to the creation of Newton, a new physics engine designed to enhance robotics simulation and training, enabling more sophisticated and responsive robotic systems. These developments signify a leap forward in the field of robotics, with potential applications across various industries, including manufacturing, healthcare, and entertainment.

NVIDIA Omniverse

The NVIDIA Omniverse platform was showcased for its capabilities in virtual collaboration and simulation. By integrating AI and simulation technologies, Omniverse facilitates the creation of digital twins, enabling industries to design, simulate, and optimize complex systems in a virtual environment. This platform empowers teams to collaborate seamlessly, reducing development times and improving the accuracy of simulations in fields such as architecture, engineering, and entertainment.

With labour shortages expected to hit 50 million workers globally by 2030, robots are set to play a crucial role in various industries. NVIDIA is positioning itself at the forefront of this revolution with:

• Omniverse: An operating system for physical AI that simulates and trains robots in virtual environments.
• Cosmos: A generative AI model that simulates infinite real-world environments to train robots safely and efficiently.

Breakthroughs in Robotic Training

• AI-driven reinforcement learning: Robots improve by interacting with physics-based virtual environments(Verifiable Physics reward).
• Newton Physics Engine: A collaboration between DeepMind, Disney Research, and NVIDIA to create the most realistic, GPU-accelerated physics engine for robotics.

NVIDIA is pioneering the AI-driven future of robotics, where machines learn faster, move more naturally, and interact intelligently with the real world.

Open-Sourcing Robotics: Groot N1

NVIDIA’s commitment to robotics extends to open-source development, announcing that Groot N1 is now available to the public.

Why Open-Source?

By open-sourcing Groot N1, NVIDIA is:

• Accelerating collaboration in AI-powered robotics.
• Providing developers worldwide access to cutting-edge tools.
• Encouraging faster innovation and widespread adoption of AI-driven robots.

Partnerships with Disney and Google DeepMind

NVIDIA announced strategic partnerships with Disney and Google DeepMind, focusing on advancing AI and technology integration. These collaborations aim to leverage NVIDIA’s computing platforms to drive innovation in entertainment, research, and robotics, pushing the boundaries of what’s possible with AI. By combining expertise and resources, these partnerships are set to accelerate the development of cutting-edge technologies and applications, benefiting various sectors globally.

Watch the video here:

Conclusion

NVIDIA GTC 2025 keynote was a clear declaration: AI is no longer just a tool—it is the foundation of the next era of computing. Whether in graphics, autonomous vehicles, data centres, or robotics, NVIDIA GTC 2025 is driving the AI revolution with:

• Next-gen AI hardware (GeForce 5090, Blackwell, DGX workstations).
• AI-driven data centres (Dynamo OS, MVLink switches, liquid-cooled AI supercomputers).
• AI-powered robotics and automation (Omniverse, Cosmos, Newton engine, Groot N1).
• Major industry collaborations (GM for autonomous vehicles, DeepMind & Disney Research for robotics).

NVIDIA is not just creating new products—it is redefining industries, laying the groundwork for an AI-powered future.

These NVIDIA GTC 2025 collectively underscore NVIDIA’s commitment to advancing computing through AI, high-performance hardware, and strategic collaborations, setting the stage for the next era of technological innovation.

Let me know your thoughts on the NVIDIA GTC 2025 in the comment section below!

Pankaj Singh

Hi, I am Pankaj Singh Negi - Senior Content Editor | Passionate about storytelling and crafting compelling narratives that transform ideas into impactful content. I love reading about technology revolutionizing our lifestyle.