In China, numerous companies are dedicated to the design of GPUs for artificial intelligence (AI) applications. Notable names include Stepfun from Tencent Holdings, Infinigence AI, Siliconflow from Huawei, Metax, Biren Technology, Focus Me, Iluvatar Corex, and Moore Threads. However, one company is currently outshining the rest. As highlighted in this article, this company represents China’s leading asset in the quest to dispense with NVIDIA chips.
While not as famous as tech giants like Huawei or Moore Threads, Cambricon Technologies is one of the firms specializing in GPU design for AI with immense growth potential . Recently, the company was approved by the Shanghai Stock Exchange to raise $560 million. This funding will be channeled into the design of four chips specifically for training and inference of AI models, along with the development of an alternative to NVIDIA’s CUDA platform.
Cambricon appears to be thriving. In the last twelve months, the value of its stock has tripled. Given the strategic importance of AI in the technological and commercial battle with the United States, Chinese firms focusing on AI hardware are receiving substantial support. Additionally, the Chinese government has mandated that all state-owned data centers must utilize at least 50% Chinese integrated circuits on their servers, thereby boosting business prospects not just for Cambricon but also for other domestic GPU design firms.
Cambricon Technologies is Not Just Another Emerging Company
In its quest to compete with the United States, China recognizes the importance of talent . The government has been promoting elite educational institutions to attract the country’s best students. Among these are the Chen brothers, the founders and top executives of Cambricon Technologies. Chen Tianshi serves as the company’s president and general director, while Chen Yunji is an advisor and technology lead, focusing on processor development.
<img alt="China is already at the forefront in particle physics. Its new neutrino detector is the most advanced that exists" width="375" height="142" src="https://i.blogs.es/3b2e34/junofinal-ap/375_142.jpeg"/>Both brothers graduated from an elite talent program at the Chinese Academy of Sciences and currently serve as researchers there. Their complementary skills have fueled Cambricon’s success: Tianshi excels in chip design, while Yunji has a forte in AI.
Chen Tianshi and Chen Yunji obtained their doctorates in computer science at age 24
The duo initiated a project at the Chinese Academy of Sciences that sought to create a processor tailored for deep learning . Their project materialized and ultimately facilitated the establishment of Cambricon. Their impressive academic backgrounds, having attained their doctorates in computer science by age 24, lend credibility to their vision. However, what sets Cambricon apart from traditional emerging firms is its growth trajectory, backed significantly by the Chinese government, which views this company as pivotal for achieving technological self-sufficiency .
While Huawei has recently emerged as a key player in Chinese GPU design for AI, Cambricon has something that Huawei currently lacks: a highly ambitious hardware-software integration . While Huawei’s Ascend chips are known for competitiveness, along with their Cann software (Compute Architecture for Neural Networks) aimed at rivaling CUDA, Cambricon has proven its agility by quickly adapting its Neuware software to meet customer demands. In a market dominated by CUDA, this adaptability is invaluable.
Presently, Cambricon’s flagship products competing against NVIDIA and Huawei in the local market are the MLU series (Machine Learning Unit). Notably, insiders believe that Cambricon’s Siyuan 690 GPU could achieve performance levels comparable to NVIDIA’s H100 chip.
Moreover, Cambricon ensures that its products are compatible with widely-used models in China, such as Deepseek , Alibaba’s Qwen, and Tencent’s Hunyuan, establishing trust within the Chinese industry . Furthermore, reports suggest that, according to the Financial Times, developers find Neuware more user-friendly than Cann, suggesting that Cambricon may capture significant attention within the tech industry in the upcoming months.
Image | Cambricon Technologies
In conclusion, as the landscape of AI GPU design evolves, companies like Cambricon Technologies demonstrate remarkable promise. By leveraging local talent, nurturing innovation, and securing government support, they are well-positioned to challenge established players in the GPU market. Amid geopolitical hurdles and technological competition, Cambricon emerges not just as a contender, but as a transformative force in the global AI landscape.