SK Telecom, a top mobile carrier in South Korea, will carry out joint research with the University of Toronto to develop automatic generation and optimization technology for artificial neural networks based on artificial intelligence semiconductor chips that can be used in many areas such as data centers and autonomous driving.
A memorandum of understanding on joint research in AI technologies has been signed between the University of Toronto, SK Telecom (SKT) and Sapeon Inc., a U.S. entity formed by SK Group units to promote their AI semiconductor business.
SKT and Sapeon will provide neural processing unit (NPU) farms consisting of AI semiconductors chips as well as a dedicated cloud platform for joint research to help AI companies automatically generate artificial neural network models or optimize and operate existing models in Sapeon semiconductors, X220 and X330, that can enhance the performance of data centers through the speedy computation of massive amounts of data.
Ordinary computers and mobile devices that run AI systems consist of a central processing unit (CPU) or an application processor (AP) as well as a graphic processing unit (GPU). As such a combination makes the system bulky, AI developers have created NPUs capable of handling AI learning and reasoning.
SKT has led a state project to develop next-generation AI chips and interfaces that can be utilized for high-performance servers like cloud data centers. The deep learning computation speed of X220, unveiled in November 2020, is faster than that of graphics processing units widely used by AI-service companies.
SKT aims to combine AI chips and 5G edge cloud. The application of an AI chip to a cloud located close to the end-users will enable the provision of high-quality AI services with ultra-low latency regardless of customer devices.
By Lim Chang-won