The Zhitong Finance App learned that Ubitus K.K, a cloud service company supported by NVDA.US (NVDA.US), plans to build a new data center in Japan. The company specifically emphasized that the new data center must be close to a nuclear power plant. Ubitus CEO Wesley Kuo (Wesley Kuo) said in an interview that the company is considering buying land in Kyoto, Shimane, or Kyushu because these regions are rich in nuclear energy resources. Guo Weisi believes that despite the controversy over nuclear power, nuclear power is still the most competitive energy option in terms of cost and supply scale, especially in the field of artificial intelligence that requires a continuous, high-capacity power supply.
Ubitus already has two gaming data centers and plans to build a third to serve generative artificial intelligence. Guo Weisi pointed out that nuclear energy is more stable than renewable energy such as solar energy and wind energy, and is essential for industrial use, especially in the field of artificial intelligence. As the demand for zero-emission, stable electricity from artificial intelligence and data centers increases, Ubitus has joined the ranks of technology companies such as Nvidia, which are actively exploring the use of nuclear energy.
Although Japan's nuclear safety faced challenges after the Fukushima nuclear accident in 2011, Ubitus plans to select a location for the new data center in the first quarter of next year and has visited several potential sites. Guo Weisi said that regions with data centers will be able to use nearby nuclear facilities to obtain cheap and stable electricity. Unlike the practice of American tech giants directly purchasing nuclear power or building nuclear power plants, Ubitus is more inclined to use existing nuclear energy resources.
According to information, electricity prices are usually lower in regions with nuclear power plants, such as the Kansai region in western Japan and Kyushu, which provides a cost advantage for data center operations. Ubitus' new data center is expected to have 2-3 megawatts of electricity reception and plans to expand to 50 megawatts. As the electricity consumption of artificial intelligence models continues to increase, nuclear energy is becoming increasingly popular due to its ability to provide stable and large amounts of electricity that can be supplied on demand.
It is worth mentioning that recently, there have been more and more frequent collaborations between utility companies and tech giants. They have signed a series of nuclear energy agreements aimed at achieving clean energy goals while advancing the ambition of artificial intelligence (AI) technology.
Artificial intelligence models are extremely energy intensive because they need to process massive amounts of data to generate responses. Because of this, nuclear energy has become the preferred energy option over wind or solar energy, and can provide a stable and continuously available amount of electricity.
In this area, Amazon recently announced that it will invest $500 million in the nuclear energy industry, including an agreement with Dominion Resources (D.US) involving discussions to build a small modular nuclear reactor near North Anna Station.
Google is not far behind, and has reached a power purchase agreement with nuclear power startup Kairos Power to buy electricity generated by several small modular reactors built by the company. Microsoft, on the other hand, collaborated with Constellation Energy (CEG.US) to restart a unit at the Three Mile Island nuclear power plant in Pennsylvania.