NVIDIA Redefines AI Landscape with CES 2025 Announcements
Published on 1/8/2025 | 5 min read
Published on 1/8/2025 | 5 min read
NVIDIA has once again pushed the boundaries of artificial intelligence with its groundbreaking announcements at CES 2025. The company's latest innovations span multiple industries, including gaming, robotics, autonomous vehicles, and more, while introducing next-generation hardware and software platforms designed to empower developers and enterprises globally.
NVIDIA CEO Jensen Huang set the tone for the event, unveiling transformative AI solutions and emphasizing the company's vision for the future of "physical AI."
Huang began by discussing the rapid evolution of AI, stating:
AI has grown from perceiving and understanding images, words, and sounds to generating content. Now, we’re advancing into the era of ‘physical AI,’ where systems can perceive, reason, plan, and act in the physical world.
NVIDIA continues to fuel advancements across industries with cutting-edge platforms and GPUs, showcased through innovations such as the Cosmos platform, RTX 50 Series GPUs, and a compact supercomputer called Project DIGITS.
The highlight of the keynote was the introduction of the RTX 50 Series GPUs, powered by NVIDIA’s Blackwell architecture. Leading the lineup is the RTX 5090 GPU, which boasts 92 billion transistors and achieves 3,352 trillion AI operations per second (TOPS).
Huang described the RTX 5090 as “a beast,” showcasing its advanced features, such as dual cooling fans and AI-powered real-time graphics capabilities. The RTX 50 Series will have a phased release, with the RTX 5090 and RTX 5080 launching in January, followed by the RTX 5070 Ti and RTX 5070 in February. Laptop variants will hit the market in March.
In addition, NVIDIA introduced DLSS 4, featuring Multi-Frame Generation technology, which enhances gaming performance by generating three extra frames for every one rendered. Other innovations, such as RTX Neural Shaders and RTX Mega Geometry, aim to elevate realism in gaming, enabling lifelike facial and hair rendering through generative AI.
Another major highlight was NVIDIA’s Cosmos platform, described by Huang as a game-changer for robotics, industrial AI, and autonomous vehicles (AVs). Drawing parallels with the impact of large language models on generative AI, Cosmos sets the stage for advancements in robotics and autonomous systems.
The platform leverages generative models, tokenizers, and video processing frameworks to simulate real-world environments and predict optimal actions. Robots and AVs using Cosmos can ingest text, images, and video prompts to create “virtual world states,” enabling better decision-making in complex scenarios.
Early adopters of Cosmos include industry leaders like XPENG, Hyundai Motor Group, and Uber. Pras Velagapudi, CTO of Agility, highlighted its impact on robotics, noting how Cosmos addresses data scarcity challenges by generating photorealistic scenarios for model training.
NVIDIA further extended its commitment to developers by unveiling new AI foundation models for RTX PCs. These models, known as NVIDIA NIM (Neural Interaction Model) microservices, are designed to enhance content creation, productivity, and enterprise applications.
The company also introduced AI Blueprints—pre-configured tools for building AI-driven solutions tailored to specific needs, such as content generation and fraud detection. Additionally, the release of Llama Nemotron offers developers an open-source platform to build and deploy robust AI agents.
According to Huang, these tools are designed to be widely accessible:
These AI models run in every single cloud because NVIDIA GPUs are now available in every cloud.
Industry leaders such as Meta and SAP have praised the potential of NVIDIA’s tools for enterprise productivity and AI-driven collaboration. Meta’s VP of GenAI, Ahmad Al-Dahle, described agentic AI as the next frontier, while SAP’s Chief AI Officer, Philipp Herzig, emphasized its transformative impact on business processes.
NVIDIA also showcased its advancements in the automotive sector through the DRIVE Hyperion AV platform, built on the NVIDIA AGX Thor system-on-a-chip (SoC). This platform enables next-generation autonomous vehicles to achieve unparalleled safety and functionality using generative AI models.
Huang emphasized the importance of synthetic data in AV development, explaining how NVIDIA’s AI data factories—powered by Omniverse and Cosmos—generate diverse driving scenarios to enhance training datasets.
Major automakers, including Toyota, JLR, Mercedes-Benz, and Volvo, are leveraging NVIDIA’s technology to develop safer, smarter autonomous vehicles.
Huang’s keynote concluded with the introduction of Project DIGITS, NVIDIA’s smallest and most powerful AI supercomputer. Powered by the cutting-edge GB10 Grace Blackwell Superchip, Project DIGITS offers developers and engineers the ability to train and deploy AI models directly from their desks.
This compact AI supercomputer brings the full power of NVIDIA’s AI stack into a portable form, Huang said, noting its potential to make AI supercomputing accessible to individuals and organizations alike.
Scheduled for launch in May, Project DIGITS represents NVIDIA’s vision of democratizing AI development and deployment.
Reflecting on NVIDIA’s journey since inventing the programmable GPU in 1999, Huang highlighted the transformative impact of AI over the past decade. He expressed excitement about the future of AI-driven breakthroughs in robotics, gaming, and beyond.
All the technologies unveiled today will lead to astonishing advancements in general robotics and AI in the years to come, Huang concluded.