Indicates the type of a location. Location Type Defined by CCTVPLUS Indicates the role that a specific instance of a description takes among all descriptions. Description Role Defined by CCTVPLUS The role in the overall content stream. Part Role Indicates a kind of content rendition. Content Rendition Defined by CCTVPLUS CCTV Newscontent Co.,Ltd. CCTV Newscontent Co.,Ltd. All Rights Reserved. No access Chinese mainland CCTV Newscontent www.cctvplus.com +86-10-83949886 +86-10-63978447 contact@cctvplus.com 2026-04-03T21:39:18Z 2026-04-03T21:39:18Z 2026-04-03 2014-11-06T09:22:45Z China Central Television China Economy/Other Recent/File China-AI/Power/Computing China's AI token exports surge with strong computing power, robust electricity infrastructure China's AI industry is entering a new phase, marked by a steady increase in the volume of tokens generated by its large language models, thanks to the support of the country's strong computing power and robust electricity infrastructure. Token, the AI usage unit, now has an official Chinese name "ciyuan." From food delivery and short videos creation, to healthcare and education, China has the world's densest digital landscape -- and every scenario constantly consumes tokens. In March, China's average daily token calls surpassed 140 trillion, a 1,000-plus-fold surge since early 2024. "The growing number of token calls reflects the rapid development of China's AI industry. China's consumer-facing token calls are already among the global leaders. With AI widely integrated into e-commerce, travel, and social platforms, the deployment of large AI models in China has clearly reached a mature stage," said Thomas Zhou, vice president of enterprise system and software research at International Data Corporation (IDC) China. According to OpenRouter, a platform that aggregates over 300 mainstream models and serves as a barometer for developer and startup preferences, Chinese models swept the top four spots in token usage over the past month, far surpassing the United States. Users clearly prefer Chinese large language models and are willing to pay for the tokens they consume. They are opting for Chinese models not only for performance but also for cost-efficiency, according to IDC research. "Currently, token prices in China are very competitive. Some overseas companies are also using the computing power provided by Chinese large language model firms. Compared with the United States, the cost of tokens offered by domestic model providers is roughly 1/20 to 1/8 of that in America, a remarkably low price. This cost factor is extremely important for enterprises aiming to scale their AI deployments. According to our estimates, in an AI project, token costs can account for about 60 to 70 percent of total expenditure," said Zhou. This AI boom is also pushing up China's demand for computing and electricity. Yet, the country's stable energy supply keeps its computing power dependable. By the end of 2025, China's installed power capacity reached 3.89 billion kilowatts, with renewables exceeding 60 percent. "China has a huge advantage in electricity generation. Our power output is about 2.5 times that of the United States, far ahead of any other country in the world. China has in recent years begun deploying major projects, such as the Yajiang Group's hydropower initiatives. Looking ahead, China is expected to remain a global leader in power supply and will be well-positioned to support growing computing power needs," said Weng Xi, a professor at Peking University's Guanghua School of Management. This year, the "coordinated development of computing capacity and electricity supply" was emphasized in the Government Work Report for the first time. "The coordinated development of computing capacity and electricity supply is an important strategy. It requires differentiating between computing needs, specifically, providing power support based on task urgency. Urgent tasks can be handled locally using pricier thermal power, while non-urgent ones can tap into green electricity from more remote regions, which would also lead to lower token prices. The country is going to use a more market-driven and flexible system to manage the diverse electricity demands of the new AI era," said Weng. With deep strengths in electricity supply, data centers, and computing infrastructure, China is well-positioned to power both its AI innovation and the export of digital services. Chinese/Part Mute FILE: China - Date Unknown (CCTV - No access Chinese mainland) 1. Various of programmer working Beijing, China - Recent (CGTN - No access Chinese mainland) 2. SOUNDBITE (Chinese) Thomas Zhou, vice president of enterprise system and software research, IDC China: "The growing number of token calls reflects the rapid development of China's AI industry. China's consumer-facing token calls are already among the global leaders. With AI widely integrated into e-commerce, travel, and social platforms, the deployment of large AI models in China has clearly reached a mature stage." FILE: China - Date Unknown (CCTV - No access Chinese mainland) 3. Various of programmer working Beijing, China - Recent (CGTN - No access Chinese mainland) 4. SOUNDBITE (Chinese) Thomas Zhou, vice president of enterprise system and software research, IDC China (partially overlaid with shots 5-7): "Currently, token prices in China are very competitive. Some overseas companies are also using the computing power provided by Chinese large language model firms. Compared with the United States, the cost of tokens offered by domestic model providers is roughly 1/20 to 1/8 of that in America, a remarkably low price. This cost factor is extremely important for enterprises aiming to scale their AI deployments. According to our estimates, in an AI project, token costs can account for about 60 to 70 percent of total expenditure." ++SHOTS OVERLAYING SOUNDBITE++ FILE: China - Date Unknown (CCTV - No access Chinese mainland) 5. Programmer working 6. Various of people using cell phone 7. Various of screens showing AI models, procedure code ++SHOTS OVERLAYING SOUNDBITE++ FILE: Qinghai Province, northwest China - Date Unknown (CCTV - No access Chinese mainland) 8. Various of power facilities FILE: China - Date Unknown (CCTV - No access Chinese mainland) 9. Various of computing facilities Beijing, China - Recent (CGTN - No access Chinese mainland) 10. SOUNDBITE (Chinese) Weng Xi, professor Guanghua School of Management, Peking University (partially overlaid with shot 11): "China has a huge advantage in electricity generation. Our power output is about 2.5 times that of the United States, far ahead of any other country in the world. China has in recent years begun deploying major projects, such as the Yajiang Group's hydropower initiatives. Looking ahead, China is expected to remain a global leader in power supply and will be well-positioned to support growing computing power needs." ++SHOT OVERLAYING SOUNDBITE++ FILE: Yunnan Province, southwest China - Date Unknown (CCTV - No access Chinese mainland) 11. Various of mountains, river, hydropower project ++SHOT OVERLAYING SOUNDBITE++ FILE: China - Date Unknown (CCTV - No access Chinese mainland) 12. Various of wind turbines Beijing, China - Recent (CGTN - No access Chinese mainland) 13. SOUNDBITE (Chinese) Weng Xi, professor Guanghua School of Management, Peking University (partially overlaid with shots 14-15): "The coordinated development of computing capacity and electricity supply is an important strategy. It requires differentiating between computing needs, specifically, providing power support based on task urgency. Urgent tasks can be handled locally using pricier thermal power, while non-urgent ones can tap into green electricity from more remote regions, which would also lead to lower token prices. The country is going to use a more market-driven and flexible system to manage the diverse electricity demands of the new AI era." ++SHOTS OVERLAYING SOUNDBITE++ FILE: China - Date Unknown (CCTV - No access Chinese mainland) 14. Various of power plant FILE: Guizhou Province, southwest China - Date Unknown (CCTV - No access Chinese mainland) 15. Various of computing center, AI training base ++SHOTS OVERLAYING SOUNDBITE++ FILE: Inner Mongolia, north China - Date Unknown (CCTV - No access Chinese mainland) 16. Various of computing facilities