Noting that 80% of startups are building on open models, Huang also emphasized the role of NVIDIA open models<\/a> across every domain, trained on NVIDIA supercomputers, forming a global ecosystem of intelligence that developers and enterprises can build on.<\/p>\n \u201cEvery single six months, a new model is emerging, and these models are getting smarter and smarter,\u201d Huang said. \u201cBecause of that, you could see the number of downloads has exploded.\u201d<\/p>\n Find all NVIDIA news from CES in this online press kit<\/a>.<\/p>\n A New Engine for Intelligence: The Rubin Platform<\/p>\n Introducing the audience to pioneering American astronomer Vera Rubin, after whom NVIDIA named its next-generation computing platform, Huang announced that the NVIDIA Rubin platform<\/a>, the successor to the record\u2011breaking NVIDIA Blackwell architecture and the company\u2019s first extreme-codesigned, six\u2011chip AI platform, is now in full production.<\/p>\n Built from the data center outward, Rubin platform components span:<\/p>\n Rubin GPUs with 50 petaflops of NVFP4 inference Extreme codesign \u2014 designing all these components together \u2014 is essential because scaling AI to gigascale requires tightly integrated innovation across chips, trays, racks, networking, storage and software to eliminate bottlenecks and dramatically reduce the costs of training and inference, Huang explained.<\/p>\n He also introduced AI-native storage with NVIDIA Inference Context Memory Storage Platform<\/a> \u2014 an AI\u2011native KV\u2011cache tier that boosts long\u2011context inference with 5x higher tokens per second, 5x better performance per TCO dollar and 5x better power efficiency.<\/p>\n Put it all together and the Rubin platform promises to dramatically accelerate AI innovation, delivering AI tokens at one-tenth the cost. \u201cThe faster you train AI models, the faster you can get the next frontier out to the world,\u201d Huang said. \u201cThis is your time to market. This is technology leadership.\u201d<\/p>\n Open Models for All<\/p>\n NVIDIA\u2019s open models \u2014 traained on NVIDIA\u2019s own supercomputers \u2014 are powering breakthroughs across healthcare, climate science, robotics, embodied intelligence and autonomous driving.<\/p>\n \u201cNow on top of this platform, NVIDIA is a frontier AI model builder, and we build it in a very special way. We build it completely in the open so that we can enable every company, every industry, every country, to be part of this AI revolution.\u201d<\/p>\n The portfolio spans six domains \u2014 Clara for healthcare, Earth-2 for climate science, Nemotron for reasoning and multimodal AI, Cosmos for robotics and simulation, GR00T for embodied intelligence<\/a> and Alpamayo for autonomous driving \u2014 creating a foundation for innovation across industries.<\/p>\n \u201cThese models are open to the world,\u201d Huang said, underscoring NVIDIA\u2019s role as a frontier AI builder with world-class models topping leaderboards. \u201cYou can create the model, evaluate it, guardrail it and deploy it.\u201d<\/p>\n AI on Every Desk: RTX, DGX Spark and Personal Agents<\/p>\n Huang emphasized that AI\u2019s future is not only about supercomputers \u2014 it\u2019s personal.<\/p>\n Huang showed a demo featuring a personalized AI agent running locally on the NVIDIA DGX Spark desktop supercomputer and embodied through a Reachy Mini robot using Hugging Face models \u2014 showing how open models, model routing and local execution turn agents into responsive, physical collaborators.<\/p>\n \u201cThe amazing thing is that is utterly trivial now, but yet, just a couple of years ago, that would have been impossible, absolutely unimaginable,\u201d Huang said.<\/p>\n The world\u2019s leading enterprises are integrating NVIDIA AI to power their products, Huang said, citing companies including Palantir, ServiceNow, Snowflake, CodeRabbit, CrowdStrike, NetApp and Semantec.<\/p>\n \u201cWhether it\u2019s Palantir or ServiceNow or Snowflake \u2014 and many other companies that we\u2019re working with \u2014 the agentic system is the interface.\u201d<\/p>\n At CES, NVIDIA also announced that DGX Spark delivers up to 2.6x performance<\/a> for large models, with new support for Lightricks LTX\u20112 and FLUX image models, and upcoming NVIDIA AI Enterprise availability.<\/p>\n Physical AI<\/p>\n AI is now grounded in the physical world, through NVIDIA\u2019s technologies for training, inference and edge computing.<\/p>\n These systems can be trained on synthetic data in virtual worlds long before interacting with the real world.<\/p>\n Huang showcased NVIDIA Cosmos<\/a> open world foundation models trained on videos, robotics data and simulation. Cosmos:<\/p>\n Generates realistic videos from a single image Advancing this story, Huang announced Alpamayo<\/a>, an open portfolio of reasoning vision language action models, simulation blueprints and datasets enabling level 4\u2011capable autonomy. This includes:<\/p>\n Alpamayo R1 \u2014 the first open, reasoning VLA model for autonomous driving \u00a0<\/p>\n \u201cNot only does it take sensor input and activates steering wheel, brakes and acceleration, it also reasons about what action it is about to take,\u201d Huang said, teeing up a video showing a vehicle smoothly navigating busy San Francisco traffic.<\/p>\n Huang announced the first passenger car featuring Alpamayo built on NVIDIA DRIVE full-stack autonomous vehicle platform will be on the roads soon in the all\u2011new Mercedes\u2011Benz CLA<\/a> \u2014 with AI\u2011defined driving coming to the U.S. this year, and follows the CLA\u2019s recent EuroNCAP five\u2011star safety rating.<\/p>\n
\nVera CPUs engineered for data movement and agentic processing
\nNVLink 6 scale\u2011up networking
\nSpectrum\u2011X Ethernet Photonics scale\u2011out networking
\nConnectX\u20119 SuperNICs
\nBlueField\u20114 DPUs<\/p>\n![\"\"](\"https:\/\/www.newsbeep.com\/il\/wp-content\/uploads\/2026\/01\/ces26-nvidialive-open-model-ecosystem-1-1680x1200.jpg\")
<\/p>\n![\"\"](\"https:\/\/www.newsbeep.com\/il\/wp-content\/uploads\/2026\/01\/26ces-blog-1920x1080-physicalai.jpg\")
<\/p>\n
\nSynthesizes multi\u2011camera driving scenarios
\nModels edge\u2011case environments from scenario prompts
\nPerforms physical reasoning and trajectory prediction
\nDrives interactive, closed\u2011loop simulation<\/p>\n
\nAlpaSim \u2014 a fully open simulation blueprint for high\u2011fidelity AV testing<\/p>\n![\"\"](\"https:\/\/www.newsbeep.com\/il\/wp-content\/uploads\/2026\/01\/26ces-blog-1280x680-alpamayo.jpg\")
<\/p>\n