Japan Partners with Dell and JERA on $15 Billion AI Data Centre

Japan, Oct 02 : Japan is moving to expand its artificial intelligence infrastructure through a major partnership involving power generator JERA, technology company Dell Technologies and UK based AI infrastructure developer RHAELM.

The companies announced on October 1 that they had signed a memorandum of understanding to establish a standardised model for developing large-scale AI infrastructure in Japan. The first project under the partnership is a planned $15 billion hyperscale data centre in Chiba, near Tokyo.

The proposed facility is expected to have a capacity of about 400 megawatts, making it Japan’s largest single-site AI infrastructure project. Construction and development are intended to follow a repeatable model that could eventually be applied to additional locations across the country.

The project reflects the growing amount of computing infrastructure required to support artificial intelligence applications. The expansion of generative AI, large language models and other advanced computing systems has increased demand for data centres equipped with powerful processors, high-speed networking and substantial electricity supplies.

For Japan, the challenge is particularly significant because large AI facilities require dependable access to electricity in addition to suitable land, cooling systems and telecommunications infrastructure.

JERA, Japan’s largest power generator, is expected to play a central role in addressing the energy requirement. Under the proposed arrangement, the company will provide land near its Chiba thermal power station and supply power capacity for between 15 and 25 years.

The location is significant because access to electricity is becoming one of the most important factors in determining where new AI data centres can be built. AI computing facilities consume substantially more power than many conventional data centres, particularly when they operate large numbers of high-performance processors continuously.

RHAELM will be responsible for construction, operation and financing of the project, while Dell will provide standardised rack-scale AI infrastructure.

The companies plan to begin phased operations at the Chiba facility in 2028, with the site expected to reach full capacity in 2029.

The proposed development also represents a shift in the way large computing projects are being planned. Rather than developing each data centre independently, the partnership intends to establish a common model that can be replicated at other power-generation sites.

JERA has access to additional locations that could potentially support similar developments. The company’s existing network of power stations could provide opportunities for further data-centre expansion during the 2030s.

Such an approach could help shorten development timelines because infrastructure requirements could be standardised. Instead of designing every facility from the beginning, developers could use established specifications for computing equipment, power connections, cooling and other essential systems.

Dell’s involvement is focused on the computing infrastructure itself. The company will contribute rack-scale AI systems designed to accommodate the processors and networking equipment needed for large AI workloads.

The increasing demand for these systems is being driven by the rapid adoption of AI across businesses and research institutions. Organisations are deploying AI for software development, data analysis, automation, scientific research and a growing range of consumer applications.

The amount of computing required by these applications can vary considerably. Training large AI models requires massive computing clusters, while operating those models for millions of users requires infrastructure capable of handling large numbers of simultaneous requests.

Data centres therefore need to be designed around both computing performance and energy efficiency.

The Chiba project is being developed at a time when countries across Asia are seeking to increase domestic AI capabilities. Japan has significant technology and industrial expertise but faces geographical and energy constraints that can complicate the construction of very large computing facilities.

The partnership’s proposed standardised approach is intended to address some of those constraints.

The availability of electricity is particularly important because AI data centres can consume enormous amounts of power. The International Energy Agency and other industry groups have warned that data-centre expansion could become an increasingly important factor in electricity demand.

For power companies, the trend creates an opportunity to supply a new category of large industrial customer. For technology companies, it creates the need for long-term agreements that provide predictable access to electricity.

JERA’s plan to supply power capacity over a period of 15 to 25 years provides a long-term framework for the project. The arrangement could also give the data centre greater certainty about one of its largest operating costs.

The Chiba facility’s planned 400-megawatt capacity illustrates the scale of the infrastructure needed for advanced AI computing. As processors become more powerful and AI models grow more complex, data centres are increasingly being developed as large industrial facilities rather than simply buildings filled with conventional servers.

Cooling is another important consideration. High-performance AI processors generate substantial amounts of heat, requiring advanced cooling systems to maintain reliable operation.

The combination of power supply, cooling, networking and computing equipment makes AI data centres significantly more complex than many earlier facilities.

The Japanese partnership aims to address those requirements through a coordinated development model. If the Chiba project progresses as planned, its design could become a template for additional facilities across Japan.

The companies intend to expand AI data-centre capacity into the 2030s using other JERA power stations. This could create a network of large computing sites located near existing energy infrastructure.

The strategy also highlights the increasingly close relationship between the energy and technology industries. The expansion of AI is creating demand for electricity, while energy companies are becoming important participants in the development of digital infrastructure.

Japan’s effort could therefore have implications beyond the individual Chiba facility. A repeatable model for combining power generation with AI computing could help the country expand its infrastructure while reducing some of the challenges associated with finding suitable locations and securing long-term electricity supplies.

For Dell, the project provides an opportunity to deploy its AI infrastructure at significant scale. For JERA, it opens a new avenue for long-term electricity demand. For RHAELM, the partnership provides a framework for developing and operating large AI facilities.

The planned investment demonstrates how the next phase of artificial intelligence development is increasingly dependent on physical infrastructure. Powerful models require not only advanced software and processors but also data centres, electricity, cooling systems and high-speed communications.

With operations targeted to begin in 2028, the Chiba project will take several years to reach completion. Its progress will offer an indication of how quickly Japan can build the large-scale computing infrastructure needed to support the country’s growing AI ambitions.