Professor Fang Lu's research group in the Department of Electronic Engineering at Tsinghua University and Academician Dai Qionghai's group in the Department of Automation have taken a new approach to create a fully forward-thinking intelligent optical computing training architecture, developing the Taiji-II optical training chip. Alibaba DAMO Academy believes that the fusion of optoelectronics, silicon photonics, and silicon electronics will drive continued improvement in computing power. Over the next three years, silicon photonics chips will support high-speed information transmission in large data centers.
According to official news from Tsinghua University, Professor Fang Lu's research group in the Department of Electronic Engineering and Academician Dai Qionghai's group in the Department of Automation have taken a new approach to create a fully forward-thinking intelligent optical computing training architecture, developing the Taiji-II optical training chip. They have achieved in-situ optical training of large-scale neural networks, exploring new paths for artificial intelligence (AI) big models to undertake optical training. This research was published online in the journal Nature under the title "Fully Forward Optical Neural Network Training."
Silicon photonics chips are photonic integrated chips developed based on silicon wafers, which use silicon photonic materials and devices to manufacture integrated circuits using special processes. They have the characteristics of high integration, low cost, high transmission bandwidth, and other advantages in terms of size, speed, power consumption, and other aspects. Alibaba DAMO Academy has listed silicon photonics chips as one of the top ten technology trends for 2022. The Academy believes that the fusion of optoelectronics, silicon photonics, and silicon electronics will drive continued improvement in computing power. Over the next three years, silicon photonics chips will support high-speed information transmission in large data centers; and in the next 5-10 years, optical computing based on silicon photonics chips will gradually replace some computing scenarios of electronic chips.
According to the Financial Data Center, among related listed companies:
Accelink Technologies has established Guoxun Quantum Chip Co., Ltd. to carry out integrated optical chip and device-related businesses required for quantum key communication and quantum measurement systems.
Robotechnik Intelligent Technology's subsidiary, ficonTEC, is one of the world's leading manufacturers of automated packaging and testing equipment in the optoelectronics and semiconductor industry. The equipment it produces is mainly used for the micro-assembly and testing of optoelectronic components, including silicon photonics chips, optical modules, lidar, optical sensors, and biological sensors coupling, packaging, testing, and more.