Peng Cheng Cloud Brain III starts trial run of 10,000P domestic AI compute

Peng Cheng Laboratory in Shenzhen has begun trial operation of 10,000P of domestically built AI compute at its Cloud Brain III facility. IT Home reports the completed system is planned to reach 16,000P (16 EFLOPS), which would make it China's largest single AI computing facility; public reports link it to Huawei's Ascend 384 SuperPoD architecture.

• 10,000P is online now, with full capacity planned at 16,000P (16 EFLOPS)
• First-phase 4,500P of compute boards were powered on for testing in late December 2025
• The facility is designed as a core node of the national C2NET network
• The lab's next target, ZPG, calls for zettaflop compute, petabit links and gigawatt green power