National compute platform expands as domestic inference systems debut
China’s AI compute updates centered on nationwide compute coordination, domestic chip-based inference, and company clarifications on NVIDIA and Huawei links.
• China’s national computing power platform has achieved integrated monitoring and coordination across all 31 provincial-level regions, according to the Ministry of Industry and Information Technology.
• The China Computing Power Platform has more than 10,000 registered enterprise users, over 2,000 listed computing power products, more than 300 integrated large models, and billions of computing power monitoring data records.
• China unveiled its first heterogeneous mixed-inference system combining domestic GPUs and neuromorphic chips for large models at the 2026 China Computing Power Conference in Langfang, Hebei.
• Tests using DeepSeek V4 showed the domestic GPU and neuromorphic chip system delivered more than twice the cost efficiency of comparable domestic GPU clusters while cutting operating costs by more than 40%.
• Goldenmax International said it has had no contact with NVIDIA or Huawei to date, while CCTC said it currently supplies no products to NVIDIA.
• Wulanchabu’s AI computing power boom was linked to institutional advantages arising from historical burdens in the West Inner Mongolia power grid.
• TuringQ launched its Gen3 large-scale photonic quantum computer in a standard IDC rack form factor, with a chip-based modular architecture and proprietary low-latency XLink interconnect.
• DeepSeek has begun gray-scale testing of AI voice conversations, according to an IT Morning Brief item.