Nvidia Unveils AI Advancement with New TSMC-Made Chips at CES 2026

Nvidia has launched a groundbreaking AI computing platform, Vera Rubin, featuring six new TSMC-made chips. Announced by CEO Jensen Huang at CES 2026, this development features a multi-chip approach to meet the rapid growth in AI demands. Key tech giants are set to adopt this innovative platform.

Nvidia has made a remarkable stride in AI technology with the debut of its new computing platform, Vera Rubin, incorporating cutting-edge chips crafted by Taiwan Semiconductor Manufacturing Co. (TSMC). Unveiled by Nvidia CEO Jensen Huang during a keynote at the 2026 International Consumer Electronics Show (CES) in Las Vegas, this announcement signifies a shift in the company's chip production strategy, introducing multiple new chips simultaneously rather than the usual one or two.

The transition to this multi-chip platform was hinted at during Huang's visit to Taiwan in August 2025, as rapid industry growth demanded a shift to more aggressive, simultaneous innovation. Addressing the audience at CES, Huang emphasized the necessity of extreme co-design to cope with AI models' 10-fold growth and five-fold increase in generated tokens annually.

At the core of the Vera Rubin platform are six distinct components: Vera CPU, Rubin GPU, NVLink 6 Switch, ConnectX-9 SuperNIC, BlueField-4 DPU, and Spectrum-X Ethernet Switch, all leveraging TSMC's advanced 3-nanometer process. The flagship NVL72 supercomputer, honoring astronomer Vera Rubin, offers significant efficiency enhancements, cutting inference costs and reducing necessary GPUs by 75% compared to previous architectures. Major tech companies including AWS, Meta, Google, and Microsoft are poised to embrace this revolutionary AI solution, with Foxconn leading the manufacturing of AI servers using the Rubin platform.

(With inputs from agencies.)