- In February 2025, China’s NDRC and semiconductor firms unveiled landmark regulation reforms to support open-source, domain-specific DNN models. This initiative aims to democratize advanced AI development by enabling training on affordable GPU setups, promoting local innovation and reducing reliance on foreign infrastructure
- In 2024, Huawei completely overhauled its open‑source deep learning framework MindSpore (v2.3), optimized for ARM‑based NPUs on HarmonyOS and Ascend chips. This update bolsters on‑device DNN performance in smartphones, IoT devices, and edge computing platforms across Asia‑Pacific
- In February 2025, the Nature journal reported accelerating competition between Chinese and Western AI models, with Chinese small‑scale DNNs narrowing the performance gap. This reflects APAC’s maturing ecosystem of high‑quality, locally developed neural network models
- In early 2025, Origin Quantum partnered with Phoenix to leverage its “Wukong” superconducting quantum chips in DNN training. This cutting‑edge collaboration in China signifies emerging interest in integrating quantum computing with neural network workflows
- In June 2025, the MLANN 2025 conference was hosted in Xiamen, China, bringing together leading researchers and industry practitioners in machine learning and neural networks. The event showcased new architectures, optimization techniques, and real-world DNN applications across healthcare, robotics, and smart manufacturing
