Peng Cheng Laboratory has begun trial operation of 10,000P of domestically built AI computing on Peng Cheng Cloud Brain III, its large-scale science facility in Shenzhen. The milestone was announced at the opening ceremony of the 2026 World Computing Network Conference (WCNC 2026), held in Shenzhen's Nanshan district on October 15 and 16, according to the laboratory's official website.
The lab said the move takes Cloud Brain III into a phase of 'building while operating.' The 10,000P now online, measured in petaflops, is already supporting domestic large-model training and a dedicated research network for AI for science, while the remaining capacity is still being installed. Chinese tech outlet IT Home reported that the completed system is planned to reach 16,000P, or 16 EFLOPS, and described it as the largest single AI computing facility planned in China.
On architecture, the laboratory's announcement says only that the system uses a pioneering domestic intelligent-computing supernode architecture designed to ease data-transfer bottlenecks. IT Home, citing public reports, said Cloud Brain III is built on Huawei's Ascend 384 SuperPoD architecture; the laboratory does not name a chip supplier in its own release. In late December 2025, the lab said, first-phase compute boards totaling 4,500P and supporting equipment were powered on for testing.
Lu Yutong, director of the National Supercomputing Center in Shenzhen, told local media that Cloud Brain III gives the city a solid base for cross-district computing interconnection and for making computing services widely accessible, adding that high-speed links should extend to Hong Kong and Macao. The facility is designed to be a core node of the China Computing NET (C2NET), a national pilot program that laboratory director Gao Wen said has now run for four years.
In his keynote, Gao set out the next target, which he called ZPG: zettaflop-scale computing, petabit-scale communication rates and gigawatt-scale green power supply. He described a three-step path for the computing network: pooling computing resources, scheduling tasks across regions, and eventually making computing available on demand like electricity.
The conference also released two white papers. One, on the development of a global computing network, was written by Peng Cheng Laboratory with the WCNC alliance. The other, on coordinating computing and power systems, was produced with Huairou Laboratory and lays out a roadmap through 2035, with 2026 to 2030 focused on planning and building the framework and 2031 to 2035 on coordinated operation and system-wide optimization.
SEE ALSO: Ascend, MetaX, Biren and Moore Threads Announce Day-0 Support for StepFun's Step 5 Preview · Huawei Unveils Ascend 960 SuperPoD With Near-Package Optics at Connect 2026 · Huawei to Publicly Demonstrate Atlas 950 SuperPoD AI Computing Hardware at WAIC 2026