Nvidia At CES 2026: Rubin Platform, Physical AI, Gaming Innovations, And More
Nvidia introduced its Rubin AI platform, Physical AI models, and major gaming and autonomous tech updates at CES 2026.
Published : January 6, 2026 at 7:19 PM IST
Hyderabad: Nvidia founder and CEO Jensen Huang took the stage at the Fontainebleau Las Vegas today to open CES 2026 and introduce the company's new products, technologies, and future plans. Declaring that AI is scaling into every domain and every device, Huang talked about "Physical AI", which he described as AI models that are trained in a virtual environment using computer-generated “synthetic” data and then deployed as physical machines once they’ve mastered their purpose.
At the stage, Huag also unveiled Rubin, Nvidia's first extreme-codesigned, six-chip AI platform, hoping to "push AI to the next frontier". He also introduced Alpamayo, an open reasoning model family for autonomous vehicle development. Let's take a look at every major announcement Nvidia made at CES 2026.
The Rubin Platform
Already in production, the Rubin Platform is the successor to the Nvidia Blackwell architecture and is named after the pioneering astronomer Vera Rubin. It aims to significantly reduce the cost of generating AI tokens, making large-scale AI deployment more economical. Key features of the platform are as follows:
- Rubin GPUs: 50 petaflops of NVFP4 inference.
- Vera CPUs: Engineered for data movement and agentic processing.
- NVLink 6: Scale-up networking.
- Spectrum-X Ethernet Photonics: Scale-out networking.
- ConnectX-9 SuperNICs and BlueField-4 DPUs.
Huang explained that extreme codesign is crucial for scaling AI to gigascale, as it requires integrated innovation across various components to eliminate bottlenecks and reduce costs.
Nvidia's open models and their applications
Huang stressed NVIDIA's commitment to open models, which are trained on Nvidia supercomputers and are being utilised in various sectors, including healthcare, climate science, robotics, and autonomous driving. The portfolio spans six domains, which are as follows:
- Clara: Healthcare applications.
- Earth-2: Climate science.
- Nemotron: Reasoning and multimodal AI.
- Cosmos: Robotics and simulation.
- GR00T: Embodied intelligence.
- Alpamayo: Autonomous driving.
“These models are open to the world,” Huang said. “You can create the model, evaluate it, guardrail it and deploy it.”
RTX, DGX Spark, and Personal Agents
At the CES 2026, Huang said that the future of AI is not only about supercomputers but is personal. He demonstrated a personalised AI agent running on the Nvidia DGX Spark desktop supercomputer, embodied through a Reachy Mini robot using Hugging Face models. He showed how open models, model routing, and local execution turn agents into responsive, physical collaborators.
Adding further, Huang said that the DGX Spark delivers up to 2.6x performance for large models, with new support for Lightricks LTX‑2 and FLUX image models, and upcoming NVIDIA AI Enterprise availability.
Advancements in Physical AI
Talking about Physical AI, Huang emphasised that Nvidia's technologies are now grounded in the physical world, with systems trained on synthetic data before real-world interactions. He introduced Cosmos, which generates realistic scenarios for AI training, and Alpamayo, an open portfolio for autonomous driving.
Cosmos can generate realistic videos from a single image, synthesise multi‑camera driving scenarios, model edge‑case environments from scenario prompts, perform physical reasoning and trajectory prediction, and drive interactive, closed‑loop simulation.
Alpamayo's open portfolio of reasoning vision action models, simulation blueprints, and datasets enables level 4-capable autonomy. This includes Alpamayo R1, the first open reasoning vision language action model for autonomous driving, and AlpaSim, a simulation blueprint for high-fidelity autonomous vehicle testing. The first passenger car featuring Alpamayo will be the Mercedes-Benz CLA, set to hit the roads soon.
Gaming and Creative Updates
At CES, NVIDIA unveiled a broad slate of gaming updates led by DLSS 4.5, which introduces Dynamic Multi Frame Generation, a new 6X Multi Frame Generation Mode, and a second-generation transformer model for improved image quality, enabling up to 240Hz 4K path-traced gaming on RTX 50 Series GPUs.
DLSS 4 now supports 250+ games and apps, with major upcoming titles adding support at launch.
Nvidia also announced G-SYNC Pulsar monitors with over 1,000Hz effective motion clarity and ambient adaptive display tuning; RTX Remix Logic, allowing modders to trigger dynamic graphics effects in classic games without engine access; and expanded NVIDIA ACE AI, powering intelligent in-game advisors and AI teammates with long-term memory in titles like Total War: PHARAOH and PUBG.
Additionally, GeForce NOW expanded RTX cloud gaming to more devices, including Linux systems and Amazon Fire TV, with day-one streaming support for major new releases.
At CES 2026, Nvidia announced new AI upgrades for GeForce RTX GPUs and laptops that significantly boost on-device generative AI performance, cutting VRAM usage and speeding up image, video, audio-video and language model inference through CUDA, precision optimisations, and expanded RTX acceleration across popular tools and frameworks.
