Song Jae-hyeok, President and Chief Technology Officer of Samsung Electronics, is giving a keynote speech at Semicon Korea 2025 held at COEX in Gangnam, Seoul, on Nov. 19./Courtesy of Jeon Byeong-su.

To support key technologies of the post-Artificial Intelligence (AI) era, such as autonomous vehicles, quantum computing, humanoid robots, and biotechnology, semiconductors are essential. A better life for all humanity can be achieved through collaboration within the semiconductor industry.

Song Jae-hyuk, chief technology officer (president) of Samsung Electronics, said this during the keynote speech at 'Semicon Korea 2025' held in the COEX in Gangnam, Seoul, on the 19th. As the chairperson of the preparatory committee for the event, President Song delivered a keynote address on the opening day under the theme of 'Semiconductor innovation for a better life.'

President Song explained that while AI technology is evolving rapidly, it still has limitations in storage capacity, energy efficiency, and processing speed compared to the human brain. He noted, "The accuracy of AI models improved from 32% to 92% over five years, but innovation in semiconductor technology is essential for even higher performance and efficiency."

He compared AI technology to the human brain, stating, "The human brain has evolved over 3.4 billion years, occupies only 2% of body weight, yet consumes 20% of total energy, making it efficient." He added, "While AI's processing speed is faster than that of humans, it cannot match the brain's capacity to understand and adapt to the environment."

President Song also discussed the pace of AI technology development. He said, "From the Turing test in 1950 to ChatGPT-4 in 2023, AI has rapidly advanced over 80 years, and its current accuracy stands at 92.3%. Yet, AI still has limitations in fully mimicking human thought processes and judgment."

President Song emphasized that collaboration in various fields such as materials, equipment, EDA, and design is essential for innovation in the semiconductor industry to keep pace with rapidly evolving AI technology. He stated, "The semiconductor industry has entered an era where individual corporations can no longer achieve innovation alone, highlighting the importance of collaboration within the industry."

In fact, the semiconductor industry landscape is changing rapidly with the advent of the AI era. Demand for semiconductors optimized for AI applications is surging, not only for conventional general-purpose DRAM and NAND flash but also for high-bandwidth memory (HBM). The market is shifting, with an increasing demand for customized products that enhance the performance and features desired by companies designing AI accelerators beyond existing specifications.

However, Samsung Electronics is receiving evaluations that it is lagging behind in the benefits brought about by the emergence of the AI market. Samsung has ceded the HBM market to its competitor SK hynix and is failing to pass quality tests from NVIDIA, a major player in the HBM market. Jae-hyun Jeon, head of Samsung Electronics' DS division (vice chairman), is reportedly exerting full efforts to secure HBM supply, including meeting Jensen Huang, CEO of NVIDIA, but the supply of HBM3E 8-layer products delivered by SK hynix in the second half of last year is also facing difficulties. In the foundry business, challenges in stabilizing yield rates for advanced processes are widening the market share gap with TSMC, the leading company.

Meanwhile, President Song was appointed as an inside director of Samsung Electronics through the board of directors meeting held on the 18th. Samsung Electronics reinforced the board with technology experts in semiconductors, including appointing Professor Lee Hyuk-jae of Seoul National University as a new outside director alongside President Song. This is analyzed as an intent to strengthen the board's technological expertise as the gap with competing corporations continues to expand.