TSMC Loses Nvidia H200 Orders, Gains Vera Rubin Production Capacity
Nvidia halted production of its H200 AI chips for the Chinese market at TSMC and reallocated wafer capacity to its next-generation Vera Rubin platform. This reallocation will reduce TSMC’s H200 chip revenue in the near term while boosting potential Vera Rubin wafer volumes once the new platform ramps up.
1. Production Shift at TSMC
TSMC reallocated wafer fabrication lines originally designated for Nvidia’s H200 AI chips bound for China to the manufacturing of Nvidia’s upcoming Vera Rubin architecture. The change reduces the volume of H200 wafers processed in TSMC’s advanced nodes and reallocates those schedules to a platform with anticipated higher margins.
2. Financial Implications
The halt in H200 production will lower TSMC revenue generated from these chips in the current quarter, potentially impacting foundry revenue growth by an estimated mid-single-digit percentage. Offsetting this, Vera Rubin orders could contribute incremental wafer sales once volume production begins, cushioning near-term revenue headwinds.
3. Vera Rubin Ramp-Up Prospects
Nvidia plans to ramp Vera Rubin in the coming months, shifting priority at TSMC to its next-gen hardware. TSMC stands to benefit from long-term volume and pricing stability if Vera Rubin adoption matches expectations, though initial yields and process optimizations will be key.