NVIDIA Reportedly Reallocates TSMC Capacity to Accelerate Vera Rubin Production

NVIDIA had planned to use TSMC to produce H200 AI chips for the Chinese market, but U.S. export controls and Chinese regulatory reviews stalled sales. Redirecting capacity to the next-generation Vera Rubin platform signals that NVIDIA is reducing policy risk while prioritizing global data-center demand for advanced AI computing.

Market reports in July 2026 said NVIDIA had reassigned TSMC capacity originally allocated to the H200 to Vera Rubin, accelerating mass production of its next-generation AI hardware. CEO Jensen Huang said Vera Rubin had entered mass production and expected TSMC and Taiwan’s server, cooling and component supply chains to reach peak deliveries in the second half of 2026. No related order value has been disclosed.