Nvidia begins Vera Rubin superchip production: powers AI, AV

Nvidia announced on January 5, 2026 that its Vera Rubin family is in full production. Production includes the NVL72 rack-scale Rubin system and the Alpamayo autonomous-vehicle (AV) platform. The company also introduced six new Rubin chips. Coverage: NVIDIA press release; TechCrunch; The Verge; WIRED.

Rubin is a rack-scale architecture (NVL72) with dedicated Rubin chips and the Alpamayo edge platform. Nvidia says Rubin boosts training throughput and supports on-vehicle reasoning. The company published specific performance and power numbers. Independent outlets have described the launch and system design but have not yet verified third-party benchmarks (NVIDIA; TechCrunch; The Verge; WIRED).

For enterprises and auto OEMs, Rubin creates a direct path from data-center training to in-vehicle inference. Expect early system availability and partner trials this year. Procurement teams should assess rack power, cooling, and floor space for NVL72 deployments. Automotive teams should evaluate Alpamayo integration timelines against vehicle validation schedules.

Why It Matters

• Operations teams must assess rack power, cooling, and floor space for NVL72 deployments before procurement to avoid capacity shortfalls.
• AI training teams can plan to port large models to Rubin-enabled clusters to reduce training time and improve throughput.
• Automotive engineers should request Alpamayo specs and trial availability from Nvidia to align integration with vehicle validation timelines.
• Security and MLOps teams should evaluate firmware, telemetry, and runtime isolation for on-vehicle Rubin-based systems before field trials.