NextFin News - In a landmark development for the global semiconductor industry, the Chinese government has officially granted regulatory approval for domestic technology giants to resume large-scale purchases of high-end artificial intelligence hardware from Nvidia. According to reports from the Wall Street Journal and Reuters on January 28, 2026, Beijing has authorized firms including ByteDance, Alibaba, and Tencent to acquire the H200 Tensor Core GPU, a move that effectively ends a months-long period of informal pressure on Chinese enterprises to prioritize domestic alternatives.
The approval follows a high-stakes diplomatic and commercial negotiation period. U.S. President Donald Trump, following his inauguration on January 20, 2025, reached an agreement with Chinese leadership to transition from a "presumption of denial" to a transactional licensing framework. Under this new regime, the U.S. Department of Commerce confirmed on January 13 that Nvidia could export the H200—which falls under specific performance ceilings—provided the U.S. government receives a 25% revenue cut from the sales. While Washington cleared the path from the export side, the final hurdle remained Beijing’s willingness to allow its champions to deepen their reliance on American silicon. This week’s green light confirms that China’s Ministry of Industry and Information Technology (MIIT) has prioritized immediate compute capacity over immediate total self-reliance.
The H200 represents a critical upgrade for China’s AI ecosystem. While the previous H20 model was a significantly throttled version designed to bypass earlier export controls, the H200 offers nearly six times the performance in specific AI workloads. According to industry data, Chinese firms have already signaled intent to place orders totaling upwards of $14 billion for 2026. However, the approval is not an unconditional surrender to foreign technology. Sources familiar with the matter indicate that Beijing is implementing a "pairing" policy, where for every Nvidia chip purchased, companies are encouraged or required to procure a specific ratio of domestic chips, such as Huawei’s Ascend series, to ensure the local industry continues to receive investment and testing data.
From a financial perspective, the news triggered an immediate rally in the semiconductor sector. Nvidia’s stock rose 1.6% in premarket trading following the announcement, while the VanEck Semiconductor ETF jumped more than 3%. For Nvidia CEO Jensen Huang, who recently concluded a high-profile visit to major Chinese tech hubs including Shanghai and Shenzhen, the approval is a vital victory. Huang noted in Taipei that the company is "looking forward to returning to China to compete vigorously," acknowledging that the market remains one of Nvidia’s most lucrative, despite the $8 billion in lost sales previously projected due to trade restrictions.
The decision by Beijing to allow these imports highlights a sobering reality for the domestic chip industry. Despite massive state subsidies and the rapid development of firms like Moore Threads and Biren Technology, a significant performance gap remains. The H200’s HBM3e memory architecture provides a bandwidth of approximately 4,800 GB/s, a threshold that domestic foundries like SMIC still struggle to hit at scale. By allowing the H200, Beijing is ensuring that its Large Language Model (LLM) development—led by Alibaba’s Qwen and ByteDance’s Doubao—does not fall two to three generations behind U.S. counterparts like OpenAI or Anthropic, who are already moving toward Nvidia’s Blackwell and forthcoming Rubin architectures.
Looking ahead, this "transactional" phase of the chip war suggests a new era of managed interdependence. The U.S. maintains a "technological leash" by keeping the most advanced Blackwell (B200) chips prohibited, while China uses the H200 as a temporary bridge to sustain its AI ambitions. However, the long-term trend toward bifurcation remains. The "AI Overwatch Act" currently moving through the U.S. House of Representatives could grant Congress veto power over these very licenses, creating a permanent state of volatility. For investors, the H200 approval provides a short-term revenue catalyst for Nvidia, but the underlying race for "silicon sovereignty" in Beijing suggests that the window for American dominance in the Chinese market is being measured in months, not decades.
Explore more exclusive insights at nextfin.ai.
