Nvidia Redirects H200 Production to Vera Rubin Hardware After Export Delays
Nvidia halts production of H200 AI chips bound for China and redirects TSMC capacity to next-generation Vera Rubin accelerators due to constrained US export approvals. This shift signals minimal near-term H200 China sales and underscores Nvidia’s emphasis on higher-performance Vera Rubin hardware.
1. Production Strategy Change
Nvidia has stopped manufacturing its H200 AI accelerators for the Chinese market, reallocating contracted wafer output at TSMC toward its next-generation Vera Rubin GPUs. This change follows limited authorization for H200 exports, prompting the company to prioritize hardware expected to drive future AI deployments.
2. Regulatory Constraints
Recent US export rules require case-by-case approval for shipments to China and Macau, slowing large-scale deliveries of advanced AI chips. Although a small number of H200 units received licenses, the ongoing uncertainty led Nvidia to scale back China-bound H200 volumes.
3. Market Impact
By shifting capacity to Vera Rubin accelerators, Nvidia aims to capture higher-performance workload demand globally and mitigate revenue risks from stalled H200 sales in China. The move could delay near-term China revenue growth but positions the company for faster launch of its flagship AI hardware.