NVIDIA's Vera Rubin Enters Mass Production Across Over 350 Factories

en.Wedoany.com Reported - On July 15, during a developer event in Tokyo, Japan, NVIDIA CEO Jensen Huang stated that the next-generation Vera Rubin AI computing platform has entered production, and related products will be delivered to customers as planned, adding that "massive volumes are coming." This statement primarily addressed market rumors that Vera Rubin server cabinets might face delivery delays due to manufacturing issues. Huang denied the reported delays but did not disclose specific production volumes or delivery timelines.

Vera Rubin is not a single AI chip but a rack-scale computing platform designed for large data centers and AI smart factories. Its production scope covers the Vera central processing unit, Rubin graphics processing unit, BlueField data processing unit, NVLink interconnect system, Spectrum-X networking equipment, and liquid-cooled server cabinets. The platform's entry into the production phase means that related AI chips, networking equipment, storage systems, and server cabinets have transitioned from product validation to coordinated mass production across the supply chain.

NVIDIA previously disclosed that the Vera Rubin mass production system covers over 350 supply chain factories across 30 countries, with more than 150 participating companies and partners in Taiwan, China. Server manufacturers including Dell Technologies, Hewlett Packard Enterprise, Super Micro Computer, and Lenovo have initiated volume production of Vera Rubin systems. Companies such as Hon Hai, Pegatron, Quanta Cloud Technology, Wistron, and Wiwynn in Taiwan, China, are also involved in the manufacturing and assembly of server cabinets and related computing infrastructure.

This production system indicates that Vera Rubin encompasses not only AI chip capacity but also multi-layer supply capabilities including complete systems, cabinets, optical networking, data storage, and liquid cooling systems. NVIDIA stated that Vera Rubin adopts an open MGX system design, enabling different server manufacturers to organize production according to unified specifications and integrate AI chips, compute nodes, and networking equipment into directly deployable data center systems.

The Spectrum-X Ethernet photonic switching equipment, which complements Vera Rubin, has also entered production. This equipment is used to connect large-scale GPU clusters and is a network component for building million-GPU AI factories. With related server cabinets entering mass production, Vera Rubin's production capabilities have extended from chip manufacturing to the delivery of complete computing infrastructure.

According to previously announced plans, production versions of the Vera Rubin system are scheduled to begin shipping in the fall of 2026. On July 15, Huang reaffirmed the production progress, indicating that the supply chain is currently expanding capacity for AI chips, server cabinets, advanced packaging, memory, and high-speed networking equipment, preparing hardware supply for subsequent data center and AI factory deployments.