Brownfield Data Centers: Reuse Without Assumptions

Brownfield data centers are moving from niche conversions to a core development strategy. Former mills, power stations, logistics sites and heavy-industrial campuses can offer something greenfield sites often cannot: existing infrastructure, established land use, grid proximity and a community familiar with large industrial operations.

But “brownfield” is not a synonym for ready-to-build. An old substation is not a deliverable data-center connection. A large industrial water system may not suit modern cooling. A building that once housed machinery may not carry the structural loads, clear heights or security model of an AI facility.

The investment case therefore rests on disciplined reuse: identify what is genuinely transferable, price what must be replaced, and keep historical capacity separate from future IT load.

Why brownfield interest is rising

The pressure is straightforward. AI infrastructure needs land, power and schedule certainty at a scale that is increasingly difficult to assemble. Industrial closures can release sites where those three ingredients appear to coexist.

In Finland, Sesterce announced a proposed AI campus at the former Kaipola paper mill. The developer described a 200 MW first phase and 600 MW second phase, while also pointing to existing grid, water, buildings and logistics infrastructure. Those figures are reported campus capacity; the source does not establish that they are IT MW.

In Germany, Goodman’s FRA02 development in Neu-Isenburg shows a different form of reuse. The project is being developed on a vacant industrial site and is designed to supply up to 50 MW of recovered heat to a municipal network. The source separately states 54 MW of IT capacity, making the distinction between IT load and heat recovery explicit.

Japan offers the power-station model. JERA, Dell and RHAELM plan a 400 MW-class AI campus beside JERA’s Chiba thermal power station. JERA is expected to provide land and long-term power capacity, while RHAELM develops, operates and finances the facility. This is not the same as converting an old factory, but it reflects the same logic: position compute where energy infrastructure and industrial control already exist.

These examples reinforce a principle built into PowerlandMap’s market intelligence: inherited infrastructure is an input to diligence, not a conclusion.

What a brownfield site may genuinely provide

A strong industrial reuse candidate can compress several workstreams.

First, site control may be clearer. A single industrial owner can hold a large contiguous parcel, existing access roads and utility corridors. That can be easier to diligence than assembling multiple greenfield plots.

Second, the site may already have an industrial planning history. Existing zoning, noise expectations and heavy-vehicle access can reduce some community and permitting friction. They do not eliminate environmental review or a change-of-use process.

Third, the location may sit close to transmission, generation or a mature distribution network. Power proximity can improve optionality, especially where an existing industrial load has been retired. The key word is proximity. PowerlandMap’s coverage and product separate location evidence from a bankable connection.

Fourth, industrial utilities can create value beyond electricity. Water systems, heat networks, rail access, fibre routes and existing substations may support a phased build. Goodman’s district-heating agreement demonstrates how a project can turn waste heat into local infrastructure rather than treating it as an abstract sustainability claim.

The five assumptions that fail most often

1. Historical load equals available capacity

A paper mill or smelter may once have consumed substantial electricity. That does not prove the capacity is still reserved, firm, energised or transferable. The connection agreement may have expired. Network conditions may have changed. Equipment may require replacement.

The right question is not “How much power used to be here?” It is “What capacity can the utility deliver, under which agreement, by which date, and with what curtailment conditions?” Our market studies use that evidence hierarchy.

2. Campus MW equals IT MW

Reported capacity may describe grid import, facility power, generation, a full campus, a phase or an ambition. None should be silently converted into IT load. Even when a development has a credible power route, redundancy, cooling and auxiliary systems determine how much can support racks.

That distinction also matters for matching intelligence. A buyer seeking 50 MW of IT capacity cannot be matched responsibly to a site with an undefined 50 MW headline.

3. Existing buildings are automatically reusable

Industrial buildings can have generous floorplates, but data centers require specific structural loading, vibration control, fire separation, roof capacity, security zones and equipment logistics. Sometimes the most valuable inherited asset is the serviced land, not the structure.

A reuse plan should therefore compare three options: retain and retrofit, retain only selected infrastructure, or clear and rebuild. The lowest demolition cost is not necessarily the lowest total project cost.

4. Environmental history is a solved problem

Industrial sites can carry contamination, asbestos, buried utilities, legacy fuel storage and incomplete records. Remediation can affect financing, insurance and schedule. A strong brownfield thesis prices those risks before a customer or lender is asked to rely on the programme.

Community expectations also matter. Replacing a major employer with a highly automated facility can create political tension even when the capital investment is large. Developers should connect jobs, tax revenues, heat recovery and local procurement to measurable commitments. PowerlandMap’s partnership approach is designed to bring market, technical and stakeholder evidence together early.

5. One successful phase proves the full campus

Large reuse projects are usually phased. The first energised building may depend on existing infrastructure while later phases require new substations, transmission works or generation. The Core Stack and Green Arrow plan for 13 Italian data centers illustrates why portfolio sequencing matters: capital allocation, customer demand and power delivery must align site by site.

A development model should therefore show live capacity, committed construction, permitted expansion and longer-term potential as separate categories. Our contribution process preserves those distinctions when new evidence is added.

A practical brownfield diligence sequence

A disciplined screen can be completed in five passes.

1. Establish rights. Confirm ownership, leases, easements, access, environmental liabilities and any restrictions attached to former industrial use. 2. Rebuild the power case. Obtain the current connection position, voltage, firmness, upgrade scope, delivery programme and responsibility for network costs. 3. Test physical reuse. Survey structures, drainage, cooling resources, fibre, security, geotechnical conditions and construction logistics. 4. Separate the phases. Model which assets support phase one and which require new infrastructure. Do not allocate future capacity to the opening phase. 5. Underwrite demand and capital together. Compare the delivery schedule with customer requirements, procurement lead times and financing conditions.

This is also why visible construction matters. At Hut 8’s River Bend project, recent site reporting showed the first building under construction, with a stated first-quarter 2027 operating target. River Bend is not presented as a brownfield conversion; it is a useful counterpoint. Greenfield projects trade inherited complexity for the challenge of building almost every enabling system from scratch.

The decision rule

Brownfield value is real when inherited assets reduce the critical path after verification. It is illusory when the investment case merely repeats historical power, old permits or legacy buildings without confirming present rights and performance.

Developers should value a site by evidence that survives technical, utility, legal and customer diligence. Investors should discount any capacity whose basis is unclear. Buyers should compare deliverable IT load and date, not promotional campus totals.

PowerlandMap helps teams make that comparison across markets, projects and counterparties. Explore pricing, review the platform, or request access to evaluate a brownfield opportunity against verified supply, power, demand and capital signals.

*Matthieu Gallego · Founder, PowerlandMap*