Nvidia Faces Charges as US Tightens Chip Sales to China, May Penalize DeepSeek

In a significant development, U.S. chip manufacturer Nvidia was caught off guard with new export license requirements for its H20 chips to China. On April 9, the U.S. government notified Nvidia that the export of H20 chips to China would now require a license, marking a further tightening of restrictions on China's access to advanced semiconductor technology. Major Chinese cloud service companies, including Alibaba, ByteDance, and Tencent, had been expecting delivery of these H20 chips by the end of the year and were unaware of the upcoming restrictions. Remarkably, Nvidia's sales team in China was also not informed beforehand. In response to the new requirements, Nvidia stated on April 11 that it would strictly adhere to U.S. export regulations. "The U.S. government directs when and where U.S. companies can sell their products, and we fully comply with the government's guidance," the company said in a statement. However, the export control poses a substantial threat to Nvidia's business in China, one of its largest markets. As of January 26, 2023, the end of Nvidia's fiscal year, China contributed $17 billion to the company's revenue, accounting for 13% of its total sales. Since the beginning of the year, Nvidia has secured $18 billion worth of H20 chip orders, with a significant portion coming from Chinese customers. The unexpected change in export policy has significant financial implications for Nvidia. The company announced that it expects to incur up to $5.5 billion in costs for the quarter ending April 27, 2023. These costs include inventory, procurement commitments, and related reserves. Following the announcement, Nvidia's stock price fell by 6% in aftermarket trading on Tuesday. Alibaba and Tencent's stocks also took a hit, dropping 4.1% and 1.8%, respectively, in Hong Kong trading on Wednesday. The H20 chip, introduced in October 2023, was designed to comply with the U.S. government's export restrictions on advanced chips to China. Since 2022, the U.S. has banned the export of Nvidia's most advanced chips to China, citing concerns that these technologies could be used to enhance China's military capabilities. The H20, while slower in training AI models compared to Nvidia's top-tier products in other markets, excels in the inference phase, which is crucial for AI models to respond to user requests. This phase is becoming increasingly important in the AI chip market. The new export restrictions have raised concerns among industry experts and analysts. Some suggest that these restrictions could inadvertently benefit Chinese AI chip manufacturers, such as Huawei. According to Nori Chiou, the investment director at Singapore's White Oak Capital Partners, "By limiting the H20 system, U.S. regulators are actually pushing Nvidia's Chinese clients to Huawei's AI chips." He further noted that with more clients and accumulated development experience, Huawei's chip design and software capabilities could rapidly improve. Adding to the complexity, the Trump administration is reportedly considering penalties for DeepSeek, a Chinese AI startup, to prevent it from acquiring U.S. technology and to restrict the use of its services by American citizens. DeepSeek recently unveiled a low-cost AI model, which has had a notable impact on the U.S. AI ecosystem. The U.S. House Select Committee on the Strategic Competition between the United States and the Chinese Communist Party has issued a formal letter to Nvidia, requesting an explanation of whether and how the company's chips are used in DeepSeek's AI models. On April 14, the U.S. government clarified that the new license requirement for H20 chip exports to China would continue indefinitely. The likelihood of Nvidia obtaining any export licenses remains unclear. This regulation is expected to affect Nvidia's inventory, procurement commitments, and related reserves, potentially leading to a $5.5 billion loss. The "Progressive Policy Institute," a bipartisan think tank based in Washington D.C., has reported that at least one Chinese buyer, Tencent, has already installed H20 chips for training large AI models, potentially violating the U.S. restrictions on supercomputer chip usage. However, Tencent denyed these claims, stating that it has not violated any laws and has not built a supercomputer. Tencent argued that all assertions to the contrary are entirely incorrect. One of the report's authors, Tim Fiist, responded that the conclusion is based on industry-wide practices. He acknowledged that Tencent might comply with U.S. rules, but doing so would impose performance limits and additional installation costs. The tightening of export controls reflects the U.S. government's stringent approach to regulating technology exports to China, especially in the realms of AI and supercomputing. Analysts and industry insiders believe that this move may intensify the technological competition between the U.S. and China and could have broader implications for the global chip supply chain. Nvidia, as a leading producer of graphics processors and AI computing chips, has been at the forefront of designing products that comply with U.S. export restrictions. However, the H20 ban not only imposes substantial financial losses on Nvidia but also highlights the profound impact of such restrictions on business strategies and operations. Despite these challenges, Nvidia's CEO, Jensen Huang, has expressed confidence in the company's ability to maintain its market leadership. He emphasized that Nvidia is well-positioned to adapt to market changes and could potentially explore alternative channels or methods to mitigate the financial impact. Industry experts agree that Nvidia's strong technological and market credentials will likely help it navigate these turbulent waters. Nvidia's situation underscores the ongoing and intensifying technological competition between the U.S. and China. The event highlights the critical role of semiconductor technology in this contest and the complex dynamics of international tech regulation. As the global tech landscape continues to evolve, it is clear that the interplay between regulatory policies and market forces will shape the future of AI and supercomputing. Nvidia's response to this challenge will be closely watched by both the tech industry and financial markets.