The head of NVIDIA, Jensen Huang, captivated the audience during his keynote at the CES 2025, presenting the company’s vision for the future of artificial intelligence across various sectors such as video gaming, autonomous vehicles, and robotics. Huang stated that AI is progressing at an unprecedented pace, now heading towards an era defined by what he calls “physical AI,” capable of perceiving, reasoning, planning, and acting.
NVIDIA’s projects and platforms continue to generate revolutionary advancements across various industries. New technologies such as the Cosmos platform, the GeForce RTX 50 series GPUs, as well as the compact AI supercomputer named Project DIGITS, were unveiled, confirming the company’s prominent role in the tech field.
GeForce RTX 50: the power of GPUs
The highlight presentation at CES 2025 was the GeForce RTX 50 series, built on the Blackwell architecture. The RTX 5090 model was showcased, featuring an impressive total of 92 billion transistors and achieving 3.352 trillion AI operations per second (TOPS). Huang referred to it as a “beast,” highlighting its advanced features such as dual cooling, while noting the use of AI for groundbreaking real-time graphics.
Scheduled for staggered release in early 2025, the series also includes the RTX 5080 model, expected on January 30, followed by the RTX 5070 Ti and RTX 5070 models in February. Dedicated GPUs for laptops will complement this offering in March.
NVIDIA also unveiled DLSS 4 and its image generation technology called “Multi-Frame Generation,” boosting gaming performance up to eight times by generating three additional frames for each rendered shot, ensuring unprecedented smoothness.
Cosmos: physical AI for robotics
Huang described the Cosmos platform as “revolutionary” for the robotics and autonomous vehicle industries. Much like the impact of language models in generative AI, Cosmos asserts itself as a new frontier for the application of AI in autonomous systems.
This platform integrates generative models, video processing technologies, and frameworks allowing robots and vehicles to simulate various scenarios, predict the best choices based on real-world conditions, lighting, and environmental context. Leading companies such as XPENG, Hyundai Motor Group, and Uber have already committed to adopting Cosmos, accessible via an open license on GitHub.
Empowering developers with AI models
NVIDIA introduced new foundational AI models for RTX computers, aiming to boost content creation and productivity within businesses. These models, designated as NVIDIA NIM (Neural Interaction Model), seamlessly integrate with the hardware of the RTX 50 series.
These tools are accessible to all through NVIDIA GPUs, present in all cloud services. Developers also benefit from “Blueprints AI,” pre-configured tools for designing agents for specific enterprise needs such as fraud detection and video management. Huang stated that these resources are entirely open source.
Towards safer and smarter autonomous vehicles
NVIDIA also highlighted its position in the automotive industry, emphasizing the DRIVE Hyperion platform focused on the safety and intelligence of autonomous vehicles. Based on the NVIDIA AGX Thor system on chip, this technology paves the way for much more capable autonomous features, integrating generative AI models to ensure enhanced safety.
Huang affirmed that the revolution of autonomous vehicles is “underway.” The development of these vehicles requires three fundamental elements: NVIDIA DGX for training AI models, Omniverse for generating synthetic data, and DRIVE AGX, a supercomputer embedded in the vehicle.
Project DIGITS: compact AI supercomputer
The closing remarks of Huang’s keynote were reserved for the announcement of Project DIGITS, NVIDIA’s most powerful and compact AI supercomputer, powered by the new GB10 Grace Blackwell Superchip microprocessor. Huang stated that this device could fit in a pocket, breaking traditional barriers in high-performance computing.
Project DIGITS enables developers and engineers to create and deploy AI models directly from their work environment, thus making the full power of the NVIDIA ecosystem accessible. Set for launch in May, this device represents a significant step towards democratizing AI computing.
Vision for the future
Huang reflected on the journey since the invention of the programmable GPU in 1999. The past twelve years have witnessed profound transformations driven by AI. Huang envisions a future where all technologies discussed during his presentation promise unexpected advancements in robotics and other AI domains in the coming years, marking a new era of innovation.
(Image credit: NVIDIA)
Major players in AI must adapt to an ever-evolving market. To learn more about the topic, check out the article on the fierce competition in the AI chip sphere.
Investors are continually reevaluating their strategy in light of AI advancements. Read this article to better understand the challenges of the first quarter of the current century.
Demand for computing power continues to grow with the needs of AI. Discover the details of the initiative from Elon Musk regarding funding for new developments in AI.
The ambitions in AI are both exciting and uncertain. The article on AI anticipating intentions places this issue in a necessarily urgent context.
Competition intensifies around AI supercomputers. The Trainium chip from Amazon offers a significant alternative to NVIDIA.
FAQ on NVIDIA innovations at CES 2025
What are the major innovations presented by NVIDIA at CES 2025 regarding artificial intelligence?
NVIDIA announced several major innovations, including the GeForce RTX 50 GPU series, the Cosmos platform for physical AI, and the compact Project DIGITS supercomputer, all aimed at optimizing performance in gaming, robotics, and autonomous vehicles.
How does the GeForce RTX 50 series improve gaming performance?
The GeForce RTX 50 series, particularly the RTX 5090, leverages Blackwell architecture and is capable of executing 3.352 trillion AI operations per second, providing a significant enhancement in real-time graphics thanks to features like DLSS 4 that generates additional images to smooth gameplay.
What does the Cosmos platform entail and how does it affect robotics?
The Cosmos platform enables robots and vehicles to utilize generative models, integrating video processing and data management to simulate potential outcomes and plan optimal actions in real-world environments, which will dramatically transform automation and robotics.
What impact will Project DIGITS have on AI developers?
Project DIGITS, the compact supercomputer, aims to make the power of AI accessible directly from developers’ desks, thereby facilitating the training and deployment of models by offering a portable and efficient solution.
How does NVIDIA envision the future of autonomous vehicles with its new technologies?
NVIDIA highlights its DRIVE Hyperion platform which, integrated into vehicles, aims to enhance safety and autonomy through generative AI models, allowing cars to become smarter and capable of navigating complex situations.
What are the advantages of the new AI models proposed by NVIDIA for content creation?
The new AI models, presented as NIM microservices, are designed to optimize content creation and productivity in businesses while being integrated into RTX PCs, thus facilitating their adoption in various Cloud environments.
What challenges does the Cosmos platform help overcome in robot learning?
The Cosmos platform addresses the challenges of data scarcity and variability necessary for robot learning by generating photorealistic scenarios, reducing reliance on costly real data captures.
What features make RTX 50 GPUs “powerhouses” in gaming?
The RTX 50 GPUs, notably the RTX 5090, stand out for their incredible power thanks to numerous transistors and advanced features like dual cooling, enabling revolutionary real-time graphics generation.
What is the ultimate objective behind NVIDIA’s announcements at CES 2025?
NVIDIA’s goal is to radically transform several sectors by integrating advanced AI technologies into real-world applications, from gaming to autonomous vehicles, thus redefining how artificial intelligence interacts with our daily lives.
