Cambricon Technologies, the Beijing-based chipmaker seen as a potential alternative to Nvidia’s graphics processing units (GPUs), reported a record revenue surge in the first half amid a Chinese stock market frenzy driven by DeepSeek’s breakthrough AI models.
Cambricon’s revenue surged 4,348 per cent year on year to 2.88 billion yuan (US$403.8 million) in the first six months of 2025, a record high for the company since it went public in 2020.
Profit reached 1.04 billion yuan, reversing a net loss of 533 million yuan seen in the first half of 2024, according to its financial report filed to the Shanghai Stock Exchange on Tuesday.
Do you have questions about the biggest topics and trends from around the world? Get the answers with SCMP Knowledge, our new platform of curated content with explainers, FAQs, analyses and infographics brought to you by our award-winning team.
The company’s shares surged 6 per cent on Wednesday morning to trade at 1,408.9 yuan per share, continuing a rally that has seen a nearly 10-fold increase over the past two years.
The stock has more than doubled over the past month, leading an AI-driven stock market frenzy that got a further boost after Chinese start-up DeepSeek said its latest V3.1 model was trained using a new data format which was “suitable for home-grown chips soon to be released”.
Cambricon chairman Chen Tianshi speaks at the World Artificial Intelligence Conference in Shanghai in 2021. Photo: Handout alt=Cambricon chairman Chen Tianshi speaks at the World Artificial Intelligence Conference in Shanghai in 2021. Photo: Handout>
Cambricon attributed the revenue surge to its “continued market expansion and active support for the implementation of AI applications”, according to the financial report.
Cambricon was founded in 2016 by two brothers, Chen Yunji and Chen Tianshi, who are now 42 and 40 years old, respectively. They both were students in the elite “genius youth class” at the Chinese Academy of Sciences, a special education programme designed to train top local talent.
In the financial report, the company claimed progress in both its hardware and software development for chips used in AI training and inference.
For large models, its training software platform, seen as a potential rival to Nvidia’s CUDA toolkit, has further expanded support for major AI models in China, including DeepSeek, Alibaba Group Holdings’ Qwen, and Tencent Holding’s Hunyuan, the company said, adding that the platform’s overall performance was now “comparable to those of mainstream competitors” in the field of reinforcement learning.
