SK Group Chairman Chey Tae-won has announced a plan to build a next-generation AI data center in Japan between 2028 and 2029.
This plan is more than just constructing a new building; it signals SK's major strategic shift from a memory chip supplier to a 'full-stack AI infrastructure provider.' This decision was largely triggered by a confluence of recent, rapid changes in Japan's AI market.
First, the demand for AI infrastructure in Japan is skyrocketing. Microsoft recently announced a massive investment of around $10 billion in the country, while telecom giant NTT has unveiled its own large-scale AI data center roadmap. SoftBank is also exploring the development of its own "Made in Japan" AI servers. As major Japanese companies increase their local investments, it has become crucial for SK to establish a direct presence to capture this growing demand.
Second, SK Group's own technological capabilities are at their peak. SK Hynix is a leader in the HBM (High Bandwidth Memory) market, an essential component for the AI era. The company recently strengthened its partnership with NVIDIA through a multi-year deal for joint memory development and supply. The planned data center in Japan will act as a "showcase," allowing SK to demonstrate its complete stack of cutting-edge memory solutions—including HBM, SSDs, and CXL—directly to potential customers.
Finally, geopolitics and energy considerations form a critical backdrop. As the U.S. restricts exports of advanced AI chips to China, global companies are increasingly looking to build their AI infrastructure in allied nations like South Korea and Japan. Furthermore, the Japanese government is fostering an AI-friendly environment with subsidies for green data centers, making it an attractive investment destination. AI data centers consume immense amounts of power, and Japan's efforts to ensure a stable grid is another positive factor.
In essence, SK's Japan data center plan is a multi-faceted strategic decision. It's about securing a foothold in a booming market, proving its technological prowess, and securing a stable operational environment.
[Glossary]
- HBM (High Bandwidth Memory): A high-performance memory chip made by vertically stacking multiple DRAM chips, drastically increasing data processing speed. It is essential for AI computing.
- Full-Stack AI Infrastructure Provider: A company that offers a comprehensive range of solutions for AI development, from hardware (chips, servers, data centers) to software and cloud services.
- Geopolitics: The study of how geography influences politics and international relations. In this context, it refers to how global events, like the US-China tech rivalry, affect corporate investment decisions.