Beijing aims to build a cumulative computing capacity of 200,000 petaflops by 2027 to support the artificial intelligence industry, an official with the city's bureau of economy and information technology said on Thursday. A petaflop, equal to one quadrillion floating-point operations per second, is a standard unit for measuring computing capacity.

The deployment of large-scale computing clusters is expanding beyond Beijing to surrounding regions, including Hebei Province and Tianjin Municipality, spokesperson Zhang Jinrui said at the conference.

Furthermore, the government encourages internet companies and telecom operators to build computing infrastructure in western regions like the Inner Mongolia Autonomous Region, the Ningxia Hui Autonomous Region and Qinghai Province, where renewable energy resources are abundant.  

The government provides computing subsidies to companies in need of computing resources. From 2023 to date, Beijing has distributed more than 7 billion yuan (around $1 billion) in subsidies, helping nearly 100 companies cut computing costs by more than 15 percent, said Zhang.

As of June 2025, China has released 1,509 large language models, accounting for about 40.2 percent of all large models launched globally, a report by the China Academy of Information and Communications Technology said.

Energy supply is a key constraint on future computing capacity growth. Beijing will push ahead in 2026 with low-carbon upgrades at 40 existing data centers, imposing higher electricity prices on facilities that exceed energy consumption standards.