Huang arrived in Taipei on Thursday following a visit to China, where he said he met with customers, business partners, and government officials. Speaking to reporters at Taipei’s Songshan Airport, the Nvidia boss noted that while U.S. approval for exporting the H200 chip is already in place, the final decision now rests with Chinese regulators.
According to Huang, the H200 licence is in its final stages, and Nvidia is awaiting a favourable outcome from Beijing. He described the chip as beneficial not only to U.S. technological leadership but also to China’s fast-growing artificial intelligence market, adding that Chinese customers have shown strong interest in the product. However, he acknowledged that the company would have to wait patiently for the decision.
The H200, Nvidia’s second most powerful AI chip, has emerged as a sensitive issue in U.S.–China relations. Despite strong demand from Chinese technology firms and clearance from U.S. authorities, Beijing’s reluctance to quickly approve imports has become the primary obstacle to shipments. Chinese officials have not publicly explained the delay, but analysts say Beijing is attempting to balance the needs of its domestic AI industry with its broader goal of nurturing a homegrown semiconductor sector.
Earlier in the week, Reuters reported, citing sources, that China had granted conditional approval to ByteDance, Alibaba, and Tencent to purchase more than 400,000 H200 chips in total. One source said the conditions attached to the approvals were restrictive, and that customers had yet to convert them into firm purchase orders. Huang, however, said Nvidia had not received any such information and that his understanding was that the Chinese government was still in the process of making a decision.
Huang also acknowledged the competitive landscape within China, noting that the country has several strong domestic chipmakers. He said Nvidia would need to compete aggressively in the market, emphasising that securing customer orders remains the company’s top priority.
Addressing concerns about manufacturing capacity, Huang said Nvidia currently has sufficient supply to support its existing customers. If the H200 receives approval for sale in China, he said the company would work closely with Taiwan Semiconductor Manufacturing Co. (TSMC) to plan production schedules and deliver the chips as quickly as possible. Packaging capacity, particularly for advanced AI chips, remains constrained across the industry.
Beyond China, Huang also signalled interest in participating in a new funding round for OpenAI, the developer of ChatGPT. While he did not confirm the size of any potential investment, he said he would “love to invest” in the company. Media reports indicate that Nvidia, alongside Amazon and Microsoft, is in talks to invest up to $60bn in OpenAI, with Nvidia potentially contributing as much as $30bn.
Nvidia’s operations remain deeply intertwined with Taiwan’s semiconductor ecosystem. The company relies heavily on TSMC, the world’s largest contract chipmaker, as well as major manufacturers such as Foxconn and Wistron. TSMC is also expanding its global footprint, with plans to invest $165bn in Arizona to build new chipmaking facilities aimed at meeting growing demand from customers.
Huang said he expects demand for TSMC’s wafers and manufacturing capacity to eventually exceed the energy available in Taiwan, a trend he described as positive for the chipmaker’s global expansion. The Taiwan-born executive added that he plans to meet with TSMC Chief Executive Officer C.C. Wei and other key supply chain partners during his visit, underscoring the strategic importance of Taiwan to Nvidia’s long-term growth.