SK会长崔泰源警告：AI正在吞噬一切，今年千亿美元利润或瞬间变巨亏

Core Viewpoint - SK Group Chairman Choi Tae-won warns that the optimistic market expectation of SK Hynix's operating profit exceeding $100 billion this year could also translate into a potential loss of the same magnitude, highlighting unprecedented volatility due to AI-driven technological transformation [1][2] Group 1: Profit Expectations - SK Hynix's profit forecast has dramatically changed from $50 billion to over $100 billion within a few months, driven by strong demand for high-bandwidth memory and robust general DRAM sales [2] - The company achieved record sales of approximately 79 trillion KRW and operating profit of 47 trillion KRW in 2024, marking its best performance to date [2] - Despite the positive outlook, Choi expresses caution, indicating that new technologies could either provide solutions or lead to significant losses [2] Group 2: Market Distortion - Choi points out that AI infrastructure development is causing structural distortions in the semiconductor market, where the profitability of general memory has surpassed that of advanced products like HBM [3] - This distortion reflects the impact of AI demand on the semiconductor supply chain, which exceeds normal market dynamics [3] - Semiconductor companies face a strategic dilemma of needing to invest in cutting-edge technology while the rapid market changes may undermine the value of their technological advantages [3] Group 3: Energy Bottlenecks - Energy supply is identified as one of the most severe challenges in the AI era, with AI consuming nearly all required energy and electricity [4] - Building a data center costs about $50 billion, and the U.S. AI data center capacity demand is around 100 gigawatts, requiring an investment of $5 trillion in infrastructure alone, excluding energy costs [4] - The current situation necessitates matching each data center with a 1-gigawatt power plant, which exceeds traditional industry planning capabilities [4] Group 4: Integrated Solutions - In response to energy challenges, SK is preparing a new solution that integrates AI data centers with power plants, focusing on enhancing efficiency and competitiveness [5] - This strategy aligns with the energy infrastructure initiatives promoted by the Trump administration [5] - Choi emphasizes the mismatch in timelines between rapid AI changes and the long-term nature of energy projects, which typically require five years [5] Group 5: International Cooperation - Choi stresses that the cooperation among South Korea, the U.S., and Japan at this critical juncture will determine future order in the AI and semiconductor industries [6] - The forum, initiated in 2021, gathers officials, scholars, and business leaders from these countries to discuss international issues [6] - Choi's warnings provide a sobering perspective on the opportunities and uncertainties posed by AI, emphasizing the need for technological alliances and supply chain collaboration [6]