Cambricon Technologies, co-founded by the two brothers Chen Yunyi and Chen Tianshi from the Institute of Computing Technology of the Chinese Academy of Sciences (ICT-CAS), has recently secured a $100 million Series A funding round. This round was led by SDIC Ventures, with participation from Alibaba Venture Capital, Lenovo Ventures, National Science Investment, Zhongke Turing, Yuanyu Origin (an Angel Wheel affiliate), and Yongxun Investment. China is making significant strides in developing artificial intelligence chips to close the gap with global tech leaders.
Named after the "Cambrian Explosion," a term referring to a period around 600 million years ago when a vast array of invertebrates emerged rapidly, the company's first AI processor was christened "Cambrian." This name symbolizes the rapid advancement and potential explosion of artificial intelligence in the near future.
This latest round of funding has valued Cambrian Technologies at $1 billion. The company's flagship 1A chip, launched last year, is recognized as the world's first commercially available "deep learning" neural network processor for smartphones, security surveillance systems, drones, wearables, and intelligent vehicles. Compared to traditional processors, the 1A chip demonstrates superior performance-to-power ratios when executing mainstream AI algorithms.
Chen Tianshi, CEO of Cambrian Technologies, explained that while deep learning applications are abundant, they often rely on outdated general-purpose processors like CPUs or GPUs. For instance, Google famously used tens of thousands of CPUs to train a cat face recognition model a few years ago. Expanding such neural networks to human brain-scale complexity would overwhelm even the most advanced CPUs or GPUs. Therefore, Chen emphasized the need for specialized AI processors in this era of smart computing.
He further stated that Cambrian is designed specifically for deep learning tasks, offering at least two orders of magnitude better performance in image and speech recognition compared to traditional processors. Its high integration makes it feasible to embed AI capabilities into mobile devices.
To illustrate the distinction between general-purpose and specialized chips, Chen Yunfu likened them to a "Swiss Army Knife" versus a "kitchen knife." While a general-purpose processor is versatile, it lacks specialization and efficiency. Conversely, a dedicated AI processor, akin to a kitchen knife, excels in specific tasks.
Chen Yunji, Chief Scientist at ICT-CAS and co-founder of Cambrian, highlighted that the 1A chip addresses two critical issues: first, it dramatically boosts computational efficiency in AI systems, surpassing traditional CPUs and GPUs by two orders of magnitude; second, it enables offline intelligence in terminal devices, keeping user data secure by avoiding uploads to the cloud.
Cambrian's current terminal processor IP offerings include models like 1A and 1H. With this funding, the company plans to accelerate the productization and market penetration of its processor series across terminals and clouds. Their goal is to enhance the智能化 of various devices while providing robust, energy-efficient cloud-based AI solutions.
Cambrian is aggressively pushing its chips into the market, focusing on both terminals and cloud services. Terminal products such as smartphones, smart glasses, and wearable devices require chips capable of identifying images, videos, and texts. On the cloud side, prominent clients like HKUST and Shuguang are already utilizing Cambrian's solutions.
Despite the excitement surrounding China's progress in AI chips, opinions vary within the industry. A researcher from HKUST noted that the Cambrian project benefited from the high starting point provided by ICT-CAS, positioning it as a leader in China's AI hardware sector. Despite some criticism of earlier projects like "Loongson One," these efforts contributed significantly to domestic chip development.
Sheng Linghai, Research Vice President at Gartner, pointed out that China's artificial intelligence chip demand is driven by sectors like security, defense, and video AI algorithms. He believes Cambrian's technology could eventually replace NVIDIA's Jetson development boards if perfected.
However, a representative from NVIDIA argued that Cambrian's ASICs are less comparable to NVIDIA's GPUs due to their specialized nature, similar to Google's TPU. While Cambrian's chips excel in specific tasks, NVIDIA's GPUs offer greater flexibility and adaptability across diverse applications.
Chen Yunxi, echoing the sentiment of the impending smart era, remarked that each epoch has a defining material carrier—steam engines in the industrial age and general-purpose CPUs in the information age. He believes a similar core technology will emerge in the smart era.
Tractica, a market research firm, predicts that the market for AI-related deep learning chipsets will grow from $500 million last year to $12.2 billion by 2025, representing a compound annual growth rate exceeding 40%.
In conclusion, Cambrian Technologies represents a promising step forward in China's pursuit of global AI dominance. While challenges remain, the company's innovative approach and ambitious goals position it at the forefront of a transformative technological shift.
High-frequency Transformer,Switching Power Supply Transformer,Ring Electronic Transformer,High Frequency Electrical Transformer
Xuzhou Jiuli Electronics Co., Ltd , https://www.xzjiulielectronic.com