NVIDIA CEO Jensen Huang delivered a landmark keynote at CES 2026 in Las Vegas, declaring the era of “AI Everywhere, for Everyone.” The focus was overwhelmingly on AI infrastructure, physical AI, and future platforms, with no new consumer GeForce RTX GPUs announced (the first time in five years). Instead, NVIDIA unveiled massive leaps in data centre AI, autonomous driving, gaming enhancements, and developer tools. Here’s everything Nvidia announced at CES 2026.
Vera Rubin Platform: Next-Gen AI Supercomputing
NVIDIA launched the Rubin platform ahead of schedule, now in full production. Named after astronomer Vera Rubin, it’s the successor to Blackwell and the company’s first extreme co-designed, six-chip AI supercomputer system.
Key components include:
- Vera CPU
- Rubin GPU
- NVLink 6th-gen switch
- ConnectX-9 SuperNIC
- BlueField-4 DPU
- Spectrum-X 102.4T CPO Ethernet switch
Rubin delivers massive gains: up to 5x faster inference, 10x lower cost per token vs. Blackwell, and triple the speed in some workloads. It’s optimised for mixture-of-experts (MoE) models and agentic AI.
The Rubin NVL72 rack-scale system is highlighted as a “supercomputer in a rack.” First cloud deployments (AWS, Google Cloud, Microsoft, OCI, CoreWeave, etc.) start in the second half of 2026.
Self-Driving AI: Alpamayo Family & Autonomous Tech
NVIDIA introduced Alpamayo, an open-source family of vision-language-action (VLA) models that bring chain-of-thought reasoning to autonomous vehicles. This lets cars “think” through rare scenarios, explain decisions, and drive safely in complex situations.
Key highlights:
- Alpamayo models, simulation tools, and datasets (1,700+ hours of driving data) on Hugging Face.
- First production vehicle: Mercedes-Benz CLA with full NVIDIA DRIVE stack (including Alpamayo) starts shipping in Q1 2026 in the US, expanding to Europe/Asia.
- Partnerships: Mercedes-Benz for AI-defined driving; Lucid, Nuro, and Uber for robotaxi work.
Huang called it the “ChatGPT moment for physical AI,” pushing toward full autonomy.
DLSS 4.5: Major Gaming Upgrade
NVIDIA announced DLSS 4.5, the next evolution of its AI upscaling tech:
- 2nd-generation transformer model for Super Resolution (better detail, less ghosting, sharper edges).
- 6X Dynamic Multi Frame Generation (adjusts frame multiplier dynamically for smooth 240+ FPS at 4K path-traced gaming).
- Available now for Super Resolution (all RTX GPUs); Dynamic 6X MFG arrives spring 2026 (RTX 50-series exclusive).
Over 250 games already support DLSS 4 Multi Frame Generation, with more coming (007 First Light, Phantom Blade Zero, PRAGMATA, Resident Evil Requiem, etc.). Path tracing and DLSS integrations expand rapidly.
Other Notable Announcements
G-SYNC Pulsar monitors: New motion clarity tech; first models available soon.
GeForce NOW expansions: Native Linux and Fire TV apps; RTX 5080-class performance on more devices.
- RTX Remix Logic: Empowers modders with dynamic effects in remastered classics.
- ROCm 7.2 updates: Better support for Ryzen AI and developer workflows.
- Partnerships & Ecosystem: Expanded work with Siemens, OpenAI, HPE, and more for industrial AI, robotics, and cloud AI.
AMD CEO Lisa Su countered with MI400 series chips and MI500 preview, but NVIDIA dominated headlines with Rubin and physical AI vision.
CES 2026 showed NVIDIA’s shift: AI infrastructure and real-world applications over consumer GPUs. The future looks massive for data centres, autonomy, and AI agents. Stay tuned for more updates as the show continues!
