【Voice_over】 In China, domestically developed microchips and other tech hardware are fueling the country's explosive AI surge, with their rapid iteration closing performance gaps to meet domestic demand. At an intelligent computing center in Beijing, made-in-China GPU modules form the backbone of the operation, powering everything from large AI model training to critical financial security applications. According to its operator, the facility has expanded its footprint from one to three floors this year and is gearing up for another major upgrade in 2026. 【Sound_bite】 Liu Tao, deputy general manager, Huitian Network Technology Co., Ltd.: "The types and number of customers coming for testing are both growing rapidly. Our next step is to expand the cluster from 3,000 to 10,000 cards in 2026." 【Voice_over】 For AI hardware manufacturers, the ever-increasing demand from computing centers like this one has meant massively ramping up production to meet booming orders. One accelerator card maker in south China's Guangzhou City, for instance, has seen orders almost triple over the past year. The firm's founder and chairperson says homemade AI hardware is gaining strong domestic traction thanks to their swift iteration and frequent lineup updates. 【Sound_bite】 Liu Miao, founder and chairperson, Turing: "The biggest advantage of domestic cards is their proximity to the market—they can iterate rapidly based on our diverse applications and quickly roll out a steady stream of new products." 【Voice_over】 According to a report from market research firm CCID Consulting, the size of China's AI industry is projected to reach some 250 billion U.S. dollars by 2035, accounting for more than 30 percent of the world's total. [Restrictions: No access Chinese mainland]