Key Highlights

  • Nvidia (NASDAQ:NVDA) said its Vera Rubin platform generated nearly USD 20 billion of Revenue in its first quarter of production shipments.
  • Management called it the fastest product ramp in the company's history, with hyperscalers moving systems into production in about one month.
  • Data center revenue reached USD 107.6 billion in the quarter ended October 25, up 110% from a year earlier and 21% from the prior quarter.
  • Vera Rubin Ultra is scheduled for next year, followed by the Feynman architecture in 2028.

A Record Debut for a New Platform

Nvidia (NASDAQ:NVDA) began production shipments of Vera Rubin in August, and by the end of its fiscal third quarter on October 25, 2026, the platform had generated nearly USD 20 billion of revenue. Interim chief executive Colette Kress described it on the November 17 Earnings Call as the fastest product ramp in the company's history. The speed of adoption stood out as much as the dollar figure. Hyperscale customers brought Vera Rubin systems into production in about one month, roughly half the time it took them to deploy Grace Blackwell. Within about three months of first shipments, the platform was ramping at every major hyperscaler and at the leading specialised AI cloud providers. That pace matters because each new architecture carries engineering risk, from rack-level integration to cooling and power delivery. A slow or troubled ramp can compress margins and delay revenue for several quarters. Vera Rubin appears to have avoided those problems, which helped total revenue reach USD 115.8 billion, up 103% from a year earlier and 20% from the prior quarter. Inventory rose to USD 39.7 billion from USD 31.6 billion as the company built Supply for the continued ramp, a sign that management expects Demand to stay strong through the coming quarters.

Why Customers Are Moving Quickly

The Economics of AI inference explain much of the urgency. Industry benchmark results cited by the company show Vera Rubin delivering about ten times more tokens per watt than the prior GB300 systems. For Data Center operators whose growth is limited by available electricity, performance per watt translates directly into revenue per megawatt. Management also pointed to firm pricing for older hardware. Token prices have fallen nearly 60% since May, which has driven a surge in usage, and that usage has pushed up rental prices for installed accelerators. Rental rates for B300 systems have risen more than 60% since launch, and some Hopper capacity has been recontracted under multiyear deals at up to twice its original price. Agentic workloads, in which AI systems carry out multistep tasks, now account for about 80% of tokens processed on one large model routing platform. These trends suggest demand is broadening from model Training toward everyday inference, which tends to be more recurring. Management said compute is fully utilised across every hyperscaler and specialised cloud it serves, and it estimated the cloud industry Backlog at USD 2.5 trillion, up from USD 2 trillion three months earlier.

The Roadmap After Rubin

Nvidia's product cadence remains annual. Vera Rubin Ultra is scheduled for next year, and the Feynman architecture is planned for 2028. The company also highlighted its Vera CPU, which is sold as a standalone product as well as inside Rubin systems, with demand described as far outstripping supply. For investors, the Rubin ramp underpins the revenue outlook of USD 130.0 billion, plus or minus 3%, for the fourth quarter, which includes an extra week, and the raised growth expectation of about 75% for fiscal 2028. The main constraints are supply related rather than demand related. Management said it has secured memory and advanced packaging capacity, and it signed new substrate agreements during the quarter. Execution on these supply contracts will decide how quickly Rubin volumes can grow from here. Any delay in memory deliveries or packaging capacity could cap shipments even if orders remain strong.

Where Things Stand

Vera Rubin delivered nearly USD 20 billion in its debut quarter and is ramping faster than any prior Nvidia platform. Supply of memory and packaging, rather than demand, is the main constraint, and the Ultra and Feynman roadmap keeps the annual product cycle in place.