JERA unveils AI data center plans worth more than $15 billion in Chiba, Japan, pairing a 400-megawatt power allocation with Dell's rack-scale computing systems. The project is intended to become Japan's largest single-site AI infrastructure deployment and one of the largest in Asia.

 

The Chiba project rests on four commitments:

  • Capital deployment exceeding $15 billion across every phase.
  • Up to 400 MW of dedicated power beside an operating plant.
  • Phased operations beginning around 2028.
  • A repeatable model for multi-gigawatt expansion across Japan.

 

Background Reading

 

JERA Unveils AI Data Center Model for Chiba

JERA's October 1 announcement describes a memorandum of understanding with Dell Technologies and UK-based RHAELM Holdings. The partners will develop a standardized framework that combines generation, electrical infrastructure, cooling and AI compute instead of designing every facility as a separate engineering project.

 

The first deployment will sit beside JERA's Chiba Thermal Power Station near Tokyo. JERA and RHAELM are already developing the site, while Apollo Global Management intends to act as RHAELM's strategic investment and financing partner.

 

Total capital deployment is expected to exceed $15 billion, or 2.3 trillion yen, across land, power infrastructure, facility construction and computing equipment. JERA says the site will support up to 400 MW and begin operating around 2028; Reuters reports full capacity is targeted for 2029.

 

How Behind-the-Meter Power Changes the Schedule

The project's defining feature is its location next to an operating generation asset. A behind-the-meter connection can deliver electricity directly from the power station rather than waiting for a conventional grid connection, which the companies say could bring capacity online years earlier.

 

That design addresses a central constraint in AI infrastructure: acquiring chips is not enough when grid interconnections, substations, cooling and fuel supply can delay a campus. JERA brings an LNG supply chain spanning procurement, shipping, import terminals and generation, plus experience operating large power assets continuously.

 

Reuters says JERA expects to supply 400 MW for 15 to 25 years. The long commitment gives RHAELM and its financing partners a clearer power horizon, while the on-site layout reduces dependence on transmission upgrades that can take years to permit and construct.

 

The approach also places energy policy at the center of Japan's AI buildout. JERA argues that gas-fired generation and a resilient LNG chain are necessary for rapid scaling. Its announcement does not provide a project-level emissions estimate, although the company maintains a broader target of net-zero emissions from domestic and overseas operations by 2050.

 

Dell, RHAELM and Apollo Split the Roles

Dell will supply the standardized compute layer through Dell AI Factory, a pre-integrated rack-scale system intended to shorten deployment time. The announcement does not identify a chip vendor, server count or model customer, so the project remains an infrastructure framework rather than a disclosed cloud service contract.

 

RHAELM will lead project delivery, integrating JERA's site and power with Dell's computing stack. Its responsibilities cover development, construction, operation and financing, creating a single infrastructure operator between the energy source and the eventual AI workloads.

 

Apollo's intended participation matters because the investment spans more than the data hall. The capital estimate includes the land, power equipment, building and compute fleet, making the financing structure unusually broad for a single AI campus.

 

The division of labor gives each company a defined constraint to solve: JERA supplies power and land, Dell standardizes the hardware, RHAELM manages delivery, and Apollo provides capital expertise. Whether the partners can repeat that arrangement will determine if Chiba becomes a national template or remains a one-off project.

 

Japan's Multi-Gigawatt Expansion Depends on Replication

The companies plan to refine their model at Chiba before applying it at other JERA sites. Their stated ambition is multi-gigawatt AI infrastructure capacity across Japan in the 2030s, with possible deployment in other markets later.

 

The plan aligns with Japan's Watt-Bit Collaboration policy, which coordinates electricity and telecommunications development, and its GX Strategic Area program for data center clusters and technology-led industrial hubs. That policy fit could help the project move beyond a private campus into a wider national compute strategy.

 

The next evidence points will be binding project agreements, confirmed financing, named technology suppliers and construction milestones. The memorandum establishes roles and scale, but execution will depend on converting a power-station advantage into operating AI capacity on the announced timeline.