Nvidia Inihahanda ang Groq Inference Chips para sa Paglulunsad sa Pamilihang Tsino
Ni Wendy Frey
Nvidia will ship Groq AI chips to China starting in May 2026. The company unveiled the plan on March 17 after securing regulatory approval for H200 sales. Jensen Huang is betting that inference processors will bypass the training restrictions that killed H20 and B30A shipments. The $20 billion Groq licensing deal from December 2025 suddenly makes sense. China gets language processing units instead of GPUs. Beijing approved the imports despite banning training chips.
The timing follows a painful pattern. The Biden administration blocked H20 sales in January 2026. The Trump administration maintained the restrictions. Alibaba and Tencent scrambled for alternatives while Nvidia lost $15 billion in quarterly revenue. Groq chips slip through because they accelerate inference rather than train large language models. Huang told investors to exclude China from forecasts. Then he found the loophole.
Huang Inanunsyo ang Groq 3 sa GTC 2026 para sa Maglulunsad ng Mayo
The GTC 2026 conference in San Jose revealed the strategy. Huang introduced Groq 3 chips built for inference speed rather than model training. Chinese customers will receive the first export versions. Reuters reported the timeline, citing two sources familiar with the matter. The chips handle token generation for deployed AI models, not training. That distinction keeps them within legal boundaries while satisfying demand from Alibaba, ByteDance, and Tencent.
The technical design focuses on language processing units. Groq technology accelerates AI inference through a compiler optimized for token generation. Nvidia redesigned the architecture specifically to comply with Chinese export regulations. No tensor cores for training. Pure inference throughput. Sources say the chips will slot into existing H200 servers as companion processors. The result is a hybrid system that technically complies with U.S. restrictions while delivering the performance Chinese technology companies need.
$20 Bilyong Pagsusugal sa Inference na Iwasan ang mga Restriksyon ni Trump
Nvidia paid $20 billion for Groq technology in December. That acquisition now serves as the company's backdoor into China. The deal included licensing agreements that transferred compiler technology and chip designs to Nvidia's control. Huang immediately redirected engineering resources toward export-compliant products. Two million H200 orders already sit in the queue. Groq creates a new revenue stream without triggering national security restrictions.
The political calculation remains risky. Trump administration officials could close the inference loophole overnight. But Huang is betting that inference chips fall outside Washington's focus on training capabilities. China needs inference to run existing AI models. It needs training hardware to build new ones. The United States banned the latter. Nvidia now sells the former. If the strategy holds, Groq could generate billions through Alibaba Cloud services and ByteDance's AI ecosystem. If regulators intervene, Huang loses Nvidia's last major bridge into China.
