Cambricon, often dubbed “China’s little Nvidia,” posted a staggering 4,348% year-on-year revenue surge—reaching RMB 2.88 billion (about US $403.8 million) in H1 2025—its highest since going public in 2020, reports South China Morning Post. It flipped from a net loss of RMB 533 million in H1 2024 to a profit of RMB 1.04 billion in the same period this year. The stock has rallied dramatically, doubling in the past month and trading near RMB 1,400 per share, amid market enthusiasm over AI and home-grown chip momentum.
What this really means is that Cambricon isn’t just selling chips, it’s building the practical infrastructure that makes AI models work. The company credits its growth to “continued market expansion and active support for the implementation of AI applications.” That means its chips and software are being embedded in real, working AI systems, supporting inference tasks such as running chatbots, analytics, and LLMs. It’s not theoretical; it’s real-world execution.
Cambricon’s improved inference-software platform is key. By making it easier for customers to run AI models on its chips, it lowers barriers to adoption and speeds deployment. Partnerships are ramping up too; its chips now support major local AI models such as DeepSeek (renowned for V3.1), Qwen, and Tencent’s Hunyuan.
What this really shows is that the company has become more than a hopeful chip designer. It is a provider of both hardware and software that plays a tangible role in actual AI rollouts across China. That’s what “support for the implementation of AI applications” looks like on the ground.