Bright

BRIGHT EVIDENCE PACK / Emerging

Japan’s planned 400 MW AI campus puts power access first.

JERA, Dell and RHAELM signed an October 1 memorandum to coordinate power and AI infrastructure in Japan. Their first planned Chiba project would have power capacity of up to 400 MW. The campus is a development plan; commissioned computing capacity and customer outcomes are still ahead.

Canonical Bright record · JSON evidence pack · Key-facts embed

Dates and assessment

Source published
2026-10-01
Bright published
2026-10-02
Substantive update
None recorded
Evidence state
Emerging
Independent verification
Not established by this source review
Last source review
2026-10-02

The claim in context

The human problem

Companies and researchers need computing capacity that can be powered, cooled and operated reliably when they need it. Buying servers is only one step. Delays in energy supply, electrical systems and facility construction can leave a computing investment unable to do useful work.

The prior constraint

Building each campus as separate power, cooling and computing projects creates interfaces that must all be ready at the same time. A promising machine does little for its users while the rest of its operating environment remains unfinished.

AI’s actual role

The planned campus would host AI computing infrastructure. The MoU coordinates its physical support systems and rack-scale compute; it does not report a new AI model, improved model performance or a deployed customer result.

The documented result

The October 1 JERA and RHAELM announcements describe a repeatable infrastructure framework, starting at JERA’s Chiba Thermal Power Station. JERA would provide the site and power, Dell standardized compute infrastructure, and RHAELM project delivery. Power capacity is planned at up to 400 MW. RHAELM expects total capital deployment above $15 billion across all phases, including land, power systems, construction and AI compute. Apollo intends to be a strategic investment and financing partner; that does not establish closed financing. RHAELM targets operations around 2028. Reuters reports phased operation in 2028 and full capacity in 2029; these are targets.

Why it may matter

Bright’s analysis: the useful idea is coordination before construction turns every interface into a scheduling problem. Power availability, electrical equipment, cooling and computing need to work as one system. A repeatable plan could help operators recognize missing pieces sooner and give customers a clearer delivery path. The proof will be capacity commissioned, reliable workloads and customer needs met. A project budget describes intended expenditure; megawatts describe a power constraint; neither measures the quality or usefulness of an AI service. Customers should ask when usable capacity is available, what happens during outages and which costs they must carry. Communities should ask how demand, emissions and local impacts will be reported as the plan becomes an operating facility.

Limitations

Original evidence

Attribution

Credit Bright AI Future and link the canonical Bright record.

Linked source material, quotations, trademarks and media remain subject to their owners’ terms. No reuse right is granted for third-party media.