Japan is attracting more greenfield projects even while some FDI flow metrics cooled — and data centers sit in the middle of that paradox. JETRO’s Invest Japan Report 2025 counts 223 greenfield investments in Japan in 2024 (announcement-date basis), up 15.5% year-on-year, while global greenfield counts rose only 2.9%. Communications ranked among the fastest-growing sectors by project count (+84.6%). From early 2024 through the third quarter of 2025, JETRO’s project digest highlights a run of data-center and logistics announcements — Ada Infrastructure (GLP), EdgeConneX, CyrusOne with Kansai Electric Power, GDS, and others. For a foreign operator, hyperscaler partner, or infrastructure investor, the decision is not “Is Japan hot?” It is which constraint binds first: grid connection, land, latency to Tokyo/Osaka, or the Japanese-language commercial surface that wins enterprise and wholesale customers.
The bottom line — projects up; equity caution; power is the real shortlist filter
Headline FDI net flows into Japan fell 13.6% in 2024 to 2.5 trillion yen on a MOF/BOJ basis, yet greenfield announcements and cross-border M&A counts turned up. Data-center entry lives on the announcement side of that split. Capital can be global. Megawatts and Japanese procurement trust are local. Teams that model Japan as “another APAC metro” usually under-price interconnection risk and over-price brand awareness.
What 2024–2025 greenfield data shows
JETRO reports 223 greenfield projects in 2024. Software and IT services led by count (38 projects), followed by business services (34) and communications (24). Communications grew 84.6% year-on-year; renewable energy (+83.3%) and electronic components (+66.7%) also jumped. The United States led as source country with 71 projects, up 10.9%. Where destination prefecture is known, Tokyo still took a large share of cases — 117 of 175 in 2024, and historically around 60–70% over two decades — even as Kansai, Chubu, Hokkaido, and Tohoku posted year-on-year gains outside the Kanto core.
For a subset of 166 projects with known prefectures, JETRO cites about 21.1 billion US dollars in project value (roughly 3.2 trillion yen) and estimates a further 19.8 billion US dollars (roughly 3.0 trillion yen) of spillover investment induced in regions beyond the primary site. Treat those dollar totals as project-value estimates tied to that subset — not as a substitute for Balance of Payments FDI flows.
Why data centers cluster — and why policy pushes decentralization
Tokyo and Osaka historically concentrate Japanese data-center capacity because that is where enterprise demand, network density, and talent sit. The same concentration now collides with land scarcity and grid constraints. Japanese policy discussion increasingly frames coordination between power infrastructure (“watts”) and digital load (“bits”) — sometimes nicknamed “watt-bit” policy — and explores regional clustering with decarbonized power. Foreign operators should read decentralization not as a PR theme but as a site-selection variable: incentives and interconnection timelines may favor non-core prefectures even when sales teams prefer a Tokyo address.
Historically, clustering also simplified sales: one metro narrative, one set of carrier hotels, one pool of facilities managers. That convenience is eroding. As AI and cloud loads grow, the binding constraint shifts from “are we close to Marunouchi?” to “can we energize on a date a customer will accept?” Operators that still sell only the metro story will compete on rent and brand, while operators that can explain Japanese power paths in plain language will compete on certainty.
Named projects are signals, not a playbook
JETRO’s digest of major projects (January 2024–September 2025) lists concrete moves: Ada Infrastructure’s multi-site plan across Tokyo, Chiba, and the Kansai region, targeting 600MW of total IT power; EdgeConneX’s plan for a 140MW-plus campus in the Osaka/Kyoto area by 2027; CyrusOne’s first Asia hyperscale project, built with Kansai Electric Power in Seika, Kyoto, delivering 48MW in three phases; and GDS’s 40MW Fuchu, Tokyo project targeting operations by the end of 2026. These announcements prove demand and partnership patterns — utility joint ventures, logistics-developer brands, US and Asian operators entering together. They do not prove your interconnection queue, your wholesale pricing, or your enterprise sales cycle. Copying a competitor’s prefecture without copying their power and offtake structure is a common failure mode.
The commercial surface foreign operators underbuild
Data-center marketing in English often leads with megawatts, PUE, and sustainability badges. Japanese enterprise and channel buyers still ask who operates locally, who answers incidents in Japanese, what certifications and audit paths exist, and whether the entity on the contract is findable. A translated brochure that opens with global superlatives under-performs a denser Japanese page that leads with location, capacity roadmap, compliance orientation, and named escalation. Wholesale and retail paths differ — but both punish ambiguity.
There is also a machine-readable layer. Procurement teams and advisors increasingly ask AI systems in Japanese which operators are building where. If your Japan offer exists only in English press releases and PDF decks behind a form, you are harder to cite than a smaller rival with a clear Japanese capacity page. This is not a reason to overclaim megawatts. It is a reason to publish bounded, dated facts that a careful buyer — human or machine — can reuse without embarrassment.
A misconception market-entry teams often hold
“If we sign the land and power, demand will find us” worked better when capacity was scarce and buyers had few alternatives. As announcements multiply, buyers compare operators digitally before the first site tour. Another misconception: treating Japan’s AI demand narrative as automatic offtake. AI load growth is real in policy and industry planning, but contracted megawatts still depend on specific customers, latency needs, and risk allocation. Model offtake scenarios; do not treat the narrative as a signed LOI.
So what for a foreign entrant — three decisions
1. Sequence power diligence ahead of brand launch. Interconnection timing, decarbonized-power eligibility, and utility partnership shape which prefectures are actually available. Marketing cannot fix a queue you have not entered.
2. Decide Tokyo-proximate vs regional cluster on commercial truth, not prestige. Latency-sensitive workloads and national enterprise HQ sales may still need Kanto/Kansai. Cost, land, and policy support may favor regional sites — but only if your Japanese sales motion can sell them.
3. Ship a Japanese evaluation surface before you scale APAC roadshows. Specs, capacity roadmap, local entity identity, and incident-response language beat a translated global brand film. Measure whether Japanese-language search and AI answers can describe your Japan offer accurately.
Branching paths through 2027–2028
One path: core metros remain capacity-constrained; regional projects with utility partners absorb incremental AI and cloud load; foreign operators win on execution and Japanese trust assets. Another path: interconnection delays slip announced COD dates, and buyers re-open vendor comparisons — rewarding operators with transparent Japanese status pages over those with only English press releases. In both paths, greenfield count is a starting signal. Power and local commercial clarity decide outcomes.
📌
Read this first
1. 223 greenfield projects in Japan in 2024 (+15.5%)
2. Communications projects +84.6% YoY
3. Data-center and logistics announcements ran dense from 2024 into 2025
4. Tokyo still dominates known destinations; regions are rising
5. Project-value dollars ≠ BoP FDI flows — keep the metrics separate
Japan’s data-center opportunity is real enough to fill JETRO’s project tables and constrained enough to punish teams that confuse announcement heat with interconnection certainty. For foreign operators, the work is to decide site logic, power path, and Japanese-language commercial proof in one plan — not three disconnected workstreams. Regulatory and energy-policy references are orientation, not legal or engineering advice. Confirm grid, environmental, and foreign-investment procedures with qualified advisors before committing capital.
FAQ
Is Japan’s inward FDI rising overall?
Net FDI flows (MOF/BOJ) fell 13.6% in 2024 to 2.5 trillion yen, while greenfield project counts and some M&A counts rose. Use the metric that matches your decision — flow, stock, or project announcements.
Do we need a Japanese website if we sell wholesale capacity?
Wholesale still involves Japanese-side diligence. Counterparties look for local accountability and clear technical status. English-only materials slow that diligence.
Should we locate only in Tokyo?
Tokyo remains central for many demand and network reasons. Policy and power realities push regional options. Choose based on workload and offtake, not headquarters prestige alone.
How should we treat competitor megawatt announcements?
As demand and partnership signals. Re-validate COD, power, and offtake independently before using them in your board model.
Sources. JETRO, Invest Japan Report 2025, Chapter 2 — greenfield counts, sector/country rankings, major project digest (Jan 2024–Sep 2025), MOF/BOJ flow/stock context; information as of September 2025. Report hub · Chapter 2 PDF. Judgements about site selection and buyer behaviour are practitioner orientation from UDX’s work inside the Japanese market, not survey findings. Last reviewed 5 August 2026.
