NVIDIA's next-generation AI platform is now in production. Speaking at GTC Berlin on 20-22 October 2026, CEO Jensen Huang confirmed that the Vera Rubin architecture has entered volume production and is already taking orders from major hyperscalers and AI cloud providers. The platform pairs Rubin GPUs with Vera central processors, high-speed NVLink and networking into complete 'AI factory' racks. Huang said revenue per gigawatt of compute rises from about $18 billion with Hopper to $25 billion with Blackwell and now roughly $40 billion with Vera Rubin - a 60% jump over the current generation. The economics matter because power has become the binding constraint for AI data centers. NVIDIA reported $96.2 billion in revenue for its fiscal second quarter (ended 26 July), up 106% year-on-year, with data-center revenue of $89 billion. A 1-gigawatt AI factory costs about $50 billion to build and could generate some $40 billion in annual revenue, Huang noted, so a three-month delay can cost $10 billion in lost sales.
Sources: AOL / The Motley Fool; Longbridge; CoinPost.
