Media Probe into NVIDIA Rubin: Dozens of Customers Have Received Test Racks, Production Smoother Than Blackwell, Volume Ramp Expected Next Year

NVIDIA's next-generation Vera Rubin server system is accelerating its rollout, with scaled mass production expected next year, eagerly anticipated by AI developers, cloud service providers, and Wall Street investors.

On July 22, The Information reported that several NVIDIA executives recently hosted media visits, providing detailed disclosures on the mass production progress and technical specifications of Vera Rubin.

Andrew Bell, Senior Vice President at NVIDIA, stated that the manufacturing process for Vera Rubin is significantly superior to the previous-generation Blackwell products, expressing optimism about the smooth progression of the launch.

The mass production pace of Vera Rubin racks will have a decisive impact on NVIDIA's future revenue. Based on maximum capacity estimates, the potential revenue scale far exceeds NVIDIA's current quarterly performance, driving growing market attention to its commercial prospects.

Dozens of Customers Have Received Test Racks

According to The Information, dozens of customers, including CoreWeave, Microsoft, OpenAI, Anthropic, and SpaceX AI, have been among the first to receive small batches of Vera Rubin test racks, with each rack housing 72 GPUs.

Regarding pricing, one buyer revealed that each Vera Rubin rack is priced between $7 million and $8 million. In comparison, the current flagship Grace Blackwell 300 rack sells for approximately $5 million, marking a significant price increase.

During their visit to NVIDIA's headquarters in Santa Clara, media reporters also toured a semi-confidential testing site, where they observed around 30 server racks in operation, many of which were Vera Rubin units. NVIDIA executives indicated that OpenAI is currently using some of these racks.

In terms of physical form, a Vera Rubin rack is roughly the size of a tall filing cabinet but weighs up to 4,000 pounds, equivalent to a pickup truck.

Production Process Smoother Than Blackwell, Cable-Free Design Boosts Assembly Efficiency

Compared to the numerous setbacks encountered during the Blackwell launch, the manufacturing progress for Vera Rubin has been markedly smoother.

In an interview, NVIDIA Senior Vice President Andrew Bell admitted that Blackwell initially "broke almost everything"—with issues arising in hardware, software, diagnostic systems, and manufacturing, ultimately necessitating a complete redesign.

Bell noted that Vera Rubin features significant design improvements: the racks are nearly cable-free, and robotic systems handle most of the assembly processes. This substantially reduces assembly complexity, thereby enhancing production efficiency and yield rates.

He expressed cautious optimism that Vera Rubin would avoid the technical bottlenecks encountered by Blackwell.

Maximum Daily Capacity Reaches 1,000 Racks, Potential Quarterly Revenue Exceeds $630 Billion

Regarding capacity planning, Bell revealed that more than ten manufacturing partners collaborating with NVIDIA will eventually possess the capability to produce up to 1,000 Vera Rubin racks per day.

Based on this capacity, a rough estimate suggests that potential quarterly revenue would reach at least $630 billion, whereas NVIDIA's actual revenue for the quarter ending in April was $82 billion.

However, possessing maximum capacity does not imply that deliveries will actually occur at that rate; actual shipment volumes will still depend on customer demand and installation schedules.

Ian Buck, Vice President at NVIDIA, stated that the company's decisions on GPU allocation are directly linked to whether customers are physically prepared to install the servers. He jokingly hinted that the final authority on allocation decisions rests with CEO Jensen Huang.