Nvidia has cleared a long-standing dual regulatory bottleneck, receiving approvals from both the US and Chinese governments to resume sales of its H200 AI chips in China. The announcement came Tuesday during Nvidia's GTC 2026 conference, where CEO Jensen Huang confirmed the company had received purchase orders from "many" Chinese customers.

"Our supply chain is getting fired up," Huang said at a press conference.

Breaking the Regulatory Logjam

The H200 — Nvidia's second-most powerful AI chip — had been stuck in limbo for months. Despite strong demand and existing US export licenses granted in February, Beijing's hesitation to approve imports was the final barrier. Sources told Reuters that China has now also granted licenses for many customers, including ByteDance, Tencent, Alibaba, and AI startup DeepSeek, which received preliminary Chinese approval in January.

China once accounted for roughly 13% of Nvidia's total revenue, making the reopened market a significant business development.

Groq Inference Chip Adapted for China

In a separate development, Nvidia is preparing a version of its Groq chip — licensed from the inference chip startup for $17 billion last December — specifically adapted for Chinese market compatibility. The chips are not downgraded versions; they are designed to interoperate with existing systems in China and are expected to be available in May.

The move targets the AI inference market, where Nvidia faces stiffer competition from domestic Chinese players including Baidu, which produces its own inference silicon. Nvidia's forthcoming Vera Rubin chips, its most powerful next-generation hardware, remain off-limits for Chinese sales under current export regulations.