Nepes Arc is showing strong performance. It appears that the news of Samsung Electronics starting the development of next-generation AI semiconductors, which will serve as the brain for artificial intelligence (AI) computations, is influencing the stock price. To develop ‘Artificial General Intelligence’ (AGI) that thinks and acts like a human, Samsung Electronics has established a special research organization in Silicon Valley, USA, and recruited top experts in the field. Recently, Samsung Electronics appointed Professor Ham Donhee as the new deputy director of SAIT, a previously vacant position. In 2021, Deputy Director Ham, along with Harvard Professor Park Honggeun, Samsung SDS President Hwang Seongwoo, and Samsung Electronics Advisor Kim Kinam, published a paper on 'Neuromorphic' in the world-renowned journal Nature Electronics. 'Neuromorphic semiconductors' are next-generation semiconductors that implement technology combining AI and brain science principles.
As of 10:06 AM on the 21st, Nepes Arc recorded a 29.93% increase from the previous day, reaching 36,900 KRW.
The previous day, a media outlet reported that Samsung Electronics recently established an AGI semiconductor development team in Silicon Valley to create AGI-dedicated semiconductors. It is known that Woo Donghyuk, former senior vice president and developer of Google's Tensor Processing Unit (TPU), has joined the 'AGI Computing Lab.'
Samsung Electronics’ full-scale entry into AGI semiconductor development is interpreted as a response to the anticipation that the AI singularity will arrive sooner than expected. Global big tech companies such as Google, OpenAI, and SoftBank are fiercely competing for market dominance, investing hundreds of trillions of won in this 'war of money.'
Samsung maintains close relationships with the 'Big Two'?OpenAI and SoftBank. There is even speculation about the formation of an anti-NVIDIA triangular alliance involving OpenAI, SoftBank, and Samsung Electronics.
Nepes Arc was established in April 2019 as a spin-off from Nepes and specializes in testing PMIC, SoC, and RF system semiconductors. Due to the overall increase in volume, including the expansion of high-end Exynos smartphone integration and the re-release of the FE series after three years, its performance is expected to improve this year. Minjung Kwak, a researcher at Hyundai Motor Securities, analyzed, "As on-device AI increases test time, Nepes Arc’s performance will improve," and added, "With the increase in PMIC and AP volumes, it will turn profitable."
Nepes Arc has developed testing for neuromorphic AI chips. Neuromorphic chips refer to semiconductors where AI algorithms are integrated into both the processing logic and memory parts of the chip itself. The product tested by Nepes Arc was developed by Nepes’ Future Intelligence Division and includes all cognitive functions learned and processed within the chip itself. It is an AI semiconductor integrating 576 artificial neurons, each composed of memory and logic areas for AI computation. The architecture allows easy expansion of the number of neurons, enabling low power consumption and miniaturization while supporting high-speed parallel processing similar to the human brain.
This is the first domestic test implementing test algorithms for control logic, learning, storage, and recognition functions, and it has been applied to mass production. The semiconductor market used in global AI applications is expected to grow to $128.9 billion (approximately 153 trillion KRW) by 2025. Nepes Arc explained that it is currently accumulating know-how to become a technology-leading company in the AI industry, which is expected to become more intelligent in performance.
Neuromorphic semiconductors refer to AI-type semiconductors that mimic the structure and characteristics of the human brain or nerve cells to enhance efficiency through parallel computation. The related industry expects that once neuromorphic chips implementing spiking neural networks are released, they will be a 'game changer.' These chips are expected to consume less power, deliver much better performance, and operate efficiently in a manner similar to humans.
Recently, Professor Park Junhong’s research team (Biomimetic Semiconductor Laboratory) at Gyeongsang National University in Jinju, Gyeongnam, developed neuromorphic semiconductor technology that mimics humans by recognizing objects through multiple senses and learning signals, processing recognized information through multi-domain signals at the single-device level.
Professor Kim Kyungmin’s research team in the Department of Materials Science and Engineering at KAIST succeeded in developing an intelligent motion recognition device that mimics visual intelligence in insect optic nerves by integrating various memristor devices. They designed a neuromorphic computing system and applied it to the motion recognition device. Compared to existing technology, they verified a 92.9% reduction in energy consumption and a 15.0% improvement in the accuracy of predicting object movement.
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
