Samsung Electronics and Naver have unveiled their jointly developed artificial intelligence (AI) semiconductor, which has been in development over the past year. Known for its power efficiency being approximately eight times greater than chips from competitors like Nvidia, this product is expected to be used to power Naver’s mega-scale AI model, HyperCLOVA X.
The Ministry of Science and ICT announced that it held the “4th Top-Level Strategic Dialogue on AI Semiconductors” on Dec. 19 at a hotel in the Seocho district of Seoul, showcasing the achievements of domestic AI semiconductor companies.
The AI semiconductor developed by Naver and Samsung Electronics was revealed in the form of a Field-Programmable Gate Array (FPGA). FPGA is a semiconductor that allows developers to modify the design primarily used in the production of prototypes before mass production. This semiconductor is specialized for “inference,” the process of producing logical outcomes using new data after an AI model has completed its “learning” phase.
Naver stated that this AI semiconductor demonstrated eight times greater power efficiency than competing products from companies like Nvidia. This was achieved by integrating low-power, compact Double Data Rate (LPDDR) DRAM to improve efficiency. Higher power efficiency means that the same computational performance can be achieved with less power, allowing large-scale computational tasks to be performed at lower cost.
Previously, the two companies had signed a memorandum of understanding (MOU) in December last year for the development of AI semiconductor solutions and launched a practical task force. The completed AI semiconductor is expected to be used to run Naver’s HyperCLOVA X.
Meanwhile, the Ministry of Science and ICT introduced the status of the first phase of the “K-Cloud Project” on the same day. The K-Cloud Project is a project to develop ultra-high-speed, low-power domestic AI semiconductors and apply them to data centers.
The government plans to enhance domestic AI semiconductors in three stages by 2030. The first phase involves applying domestic Neural Processing Units (NPUs) to data centers and providing cloud-based AI services.
Companies participating in the first phase, such as NHN Cloud, Sapeon Korea, and NOTA, showcased their “Intelligent Traffic Control Service,” which identifies vehicles and pedestrians in crosswalk CCTV footage. KT Cloud, Rebellion, and Superb AI presented a solution for calculating nutritional intake, while Naver Cloud, Furiosa AI, and Simple Platform revealed technology for identifying workers and measuring risk levels using construction site CCTV footage.