Samsung Electronics is embarking on the development of next-generation Artificial General Intelligence (AGI)-dedicated semiconductors through its recently established AGI Computing Lab.
Kyung Kye-hyun, president of Samsung Electronics’ Device Solutions Division, expressed his excitement on his social media on March 19, saying, “I am pleased to announce the establishment of Samsung Semiconductor’s AGI Computing Labs in both the United States and Korea.” AGI refers to artificial intelligence with capabilities equal to or surpassing that of humans.
The AGI Computing Lab is currently in the process of recruiting, led by Dr. Woo Dong-hyuk, a former developer of Google’s Tensor Processing Unit (TPU). As one of the three individuals who designed the TPU platform at Google, he stated, “We will create a completely new type of semiconductor specifically designed to meet the astonishing processing requirements of future AGI.”
Initially, the AGI Computing Lab will focus on developing chips for Large Language Models (LLM), targeting inference and service applications.
President Kyung stated, “To develop chips that can significantly reduce the power required for running LLMs, we are revisiting every aspect of chip architecture, including memory design, lightweight model optimization, high-speed interconnects, and advanced packaging.”
He further revealed that Samsung Semiconductor’s plan involves continuously releasing new versions of the AGI Computing Lab’s chip designs. This iterative model will provide strong performance and support for increasingly larger models at reduced power and cost, he emphasized.
President Kyung expressed confidence, saying, “With the establishment of the AGI Computing Lab, we are positioned to solve complex system-level problems inherent in AGI, while contributing to sustainable methods for next-generation advanced Artificial Intelligence/Machine Learning models.”