Samsung Electronics has announced a significant order from Japanese artificial intelligence company Preferred Networks.
The order involves the production of chips for AI applications using Samsung’s advanced 2-nanometre (nm) foundry process and state-of-the-art chip packaging services.
This marks the first publicized order for Samsung’s cutting-edge 2nm chip manufacturing technology.
Details of the Chip Technology:
The chips will be fabricated using advanced gate-all-around (GAA) architecture. This high-tech design allows for better control of electrical currents in transistors, enhancing performance and energy efficiency.
Additionally, multiple chips will be integrated into a single package, which boosts inter-connection speed and reduces the overall size of the hardware.
South Korea’s Gaonchips Co. developed the chip design. This collaboration highlights the integration of specialized design and manufacturing capabilities to meet the sophisticated needs of AI applications.
Applications in AI:
Samsung’s chips will be utilized in Preferred Networks’ high-performance computing hardware.
These are aimed at generative AI technologies, including large language models (LLMs), which are critical for advanced AI applications such as natural language processing and machine learning.
Statements from Preferred Networks:
Junichiro Makino, vice president and chief technology officer of computing architecture at Preferred Networks, emphasized the importance of these high-performance chips for their generative AI technologies.
The integration of Samsung’s 2nm chips is anticipated to significantly improve the implementation and efficiency of Preferred Networks’ AI hardware.
This order is a milestone for Samsung’s semiconductor division, showcasing the company’s capability to produce advanced chips for high-demand applications. It also reflects the growing collaboration between leading tech firms in South Korea and Japan to drive innovation in AI and semiconductor technologies.