J-POWER to Build and Operate Data Center in Kamikuni, Hokkaido

Automation-news.jp reported on November 15, 2026, that J-POWER will build and operate the tentatively named Kamikuni Data Center in Kamikuni, Hokkaido. The facility is planned to be equipped with the latest GPUs, with computing capacity sold under the company’s own brand. Services are scheduled to launch in summer 2027.

J-POWER is reportedly planning not only to build and operate the facility but also to sell GPU computing capacity under its own brand, making it a service provider as well as an owner of the infrastructure. The report did not specify the scale of the GPU deployment, pricing or operating specifications, so the service’s capacity and price competitiveness cannot yet be assessed.

The key point in the report is that J-POWER plans to handle not only construction of the data center but also its operation and the sale of computing capacity. Rather than building the facility and leasing it out, the company plans to offer GPU-equipped computing resources as a service under its own brand, combining infrastructure ownership with service provision. However, the report provides only an outline of the plan; it does not say which tasks J-POWER will handle itself or which functions it will outsource. Under this model, GPU selection and procurement, facility operations—including power and cooling—and the provision of computing resources are linked. The plan’s execution will therefore depend not only on completing the facility but also on keeping the computing resources running reliably and supplying them as a service. The report, however, does not disclose the GPU models or number of units, computing performance, facility size, power capacity or service fees. Without these details, it is not possible to compare the service’s performance or pricing with other computing platforms or assess its intended uses. By selling computing capacity under its own brand, J-POWER would expand its role from an infrastructure owner to a provider of computing services. Recovery of the capital investment will depend on facility utilization and service demand, but the report does not identify prospective users or contract terms, leaving no basis for a detailed discussion of the revenue model. The planned service launch in summer 2027 was disclosed, but the construction schedule and whether the rollout will be phased are unknown. The target date alone does not establish when services will actually begin or at what scale. Key points to watch are: (1) the GPU models and number of units to be deployed, and the computing capacity to be offered; (2) operating specifications, including the facility’s power capacity and cooling system; and (3) the construction schedule through the service launch and the scale of computing resources available at launch. Once these details are disclosed, it will be possible to assess how fully the build-and-operate model translates into actual service provision.