TAIPEI (TVBS News) — Nvidia announced Monday (Jan. 5) that its next-generation AI supercomputer, Vera Rubin, has entered full-scale production, highlighting its partnership with Taiwan's TSMC (台積電). The platform comprises six new chipsets and marks a significant technological advancement unveiled at the Consumer Electronics Show (CES) in the U.S. TSMC, the world's largest contract chipmaker, manufactures Nvidia's most advanced processors.
Nvidia CEO Jensen Huang (黃仁勳), speaking at an international press conference, emphasized the extensive collaboration with TSMC, noting that both parties update progress weekly, sometimes even daily. The Vera Rubin platform integrates the Vera CPU, Rubin GPU, NVLink 6 switch, ConnectX-9 SuperNIC, BlueField-4 data processor, and Spectrum-6 Ethernet switch, showcasing cutting-edge AI technology development.
Huang also revealed plans for mass shipment of the Grace Blackwell superchip and reiterated Nvidia's deep appreciation for its over 25-year partnership with TSMC, which manufactures the company's most advanced processors. On AI chip shipments to China, Huang highlighted high demand for Nvidia's H200 chips, stating that the supply chain is fully operational and ready to meet market needs.
Last month, U.S. President Donald Trump announced that Washington agreed to open H200 sales to China with a 25% fee, expanding market access for the chipmaker. Huang confirmed H200 production is underway, with Nvidia finalizing export permit details with the U.S. government. Despite the H200's current competitiveness in China, Huang acknowledged that more powerful Blackwell and Rubin architecture chips will be introduced in the future to maintain market leadership.