NVIDIA GTC Berlin: Vera Rubin racks revealed to power the whole agent, from model to tools
AjdDavison · x · 2026-10-21
At NVIDIA GTC in Berlin, Jensen Huang showcased the Vera Rubin rack lineup under the theme "NVIDIA Vera Rubin Powers the Whole Agent", with four dedicated racks: Vera Rubin NVL72 for model processing, Vera MGX for secure agent processing, Vera CMX storage node for knowledge serving, and Vera RTX Compute Node for CUDA-X tools processing. He also highlighted European partner companies on stage, signaling NVIDIA's push to split agent workloads across specialized racks.
More from Companies & People
- EPFL expands CS faculty hiring with openings in security, math of learning and AI alignment — caglarml · 2026-10-21
- Jensen Huang spent 10 keynote minutes telling microducks to sit before Nvidia's roadmap — Thom_Wolf · 2026-10-21
- OpenAI DevDay community hackathon hits London Nov 1 with ChatGPT Pro prizes — paw_lean · 2026-10-21
- OpenAI DevDay heads to Paris with hackathon, finale, and three tracks — paw_lean · 2026-10-21
- First OpenAI Codex Community Meetup in Brighton draws lively demos — paw_lean · 2026-10-21
- NHS to Roll Out Copilot to 500,000 Staff, Starting with Yorkshire Ambulance — satyanadella · 2026-10-21