- A Nikkei survey shows that Japan plans to increase its AI data center capacity from 1.1 GW at the end of 2025 to 4.9 GW by 2033, a more than fourfold increase.
- Nikkei surveyed 26 data center operators, with 22 confirming they will continue to build more AI facilities in Japan.
- According to Cushman & Wakefield, average construction costs are around $16 billion per GW, bringing the total construction capital for these projects to approximately $60 billion.
- When including servers and semiconductors, total investment could reach hundreds of billions of dollars.
- SoftBank estimates that Japan needs 4.2 GW of data center capacity by 2030 to meet AI processing demands.
- NTT aims to reach 2 GW by fiscal year 2033, doubling its previous plan and nearly seven times the capacity at the end of 2025.
- SoftBank plans to increase its data center capacity to 255 MW by fiscal year 2027, about four times its current level.
- More than 10 foreign companies are participating in investments, with Colt Data Centre Services targeting 370 MW by 2030 and GLP Japan expecting to operate 570 MW during 2027–2028.
- A 500 MW data center project in Akita Prefecture with a maximum capital of 2 trillion yen (approx. $13 billion) is expected to be one of Japan’s largest facilities.
- Projects are driven by demand from Microsoft and Amazon Web Services but still face challenges in power supply, community opposition regarding noise and water, and the risk of overcapacity if AI demand is lower than expected.
📌 Japan is accelerating its strategy to build AI infrastructure to move closer to AI autonomy and become one of the world’s largest data center hubs after the US and China. With a projected capacity of 4.9 GW by 2033 and construction capital of about $60 billion, the country expects to meet the exploding AI demand. However, power supply, community approval, and the risk of overinvestment will be major challenges during implementation.
