Goodman Group: How a Warehouse Developer Became a Data Centre Company

The most valuable thing Goodman owns is not buildings. It is land in the right places with electricity attached, at a moment when the constraint on data centre construction is not capital or demand but grid capacity. A logistics developer that accumulated infill sites in Sydney, London, Los Angeles and Tokyo over decades is now sitting on the scarcest input in the AI build-out, largely by accident. This article covers how that happened and what the risks are.

How did a warehouse developer end up here?

By buying the right land for the wrong reason and holding it. Goodman’s logistics strategy was built on infill sites — industrial land close to population centres rather than on distant urban fringes — because e-commerce delivery economics reward proximity to customers. Last-mile logistics needs to be near the last mile.

Those same characteristics turned out to define a good data centre site. Data centres serving cloud and AI workloads need low latency and high connectivity, which means metropolitan locations, and they need enormous electricity, which means existing industrial land with grid connections rather than greenfield sites requiring new transmission.

The company has been explicit that it continues investing in its land bank, acquiring significant sites offering future regeneration potential for high-value logistics, data centres or both. That optionality is the strategic point: a site that can become either is worth more than one that can only become one, and the decision can be deferred until demand is clear.

Why is power the binding constraint?

Because you cannot connect a large data centre to a grid that lacks capacity, and building new transmission takes a decade. Access to grid capacity remains a gating factor for new developments, which means the number of viable data centre sites in any city is limited by electricity infrastructure rather than by land availability or construction capacity.

Goodman has responded by treating power as an asset to accumulate. Its global power bank — sites it owns with secured or allocated power — expanded from 5 gigawatts to 6.4 gigawatts, primarily across Australia and continental Europe, with 3.6 gigawatts secured. Holding allocated grid capacity is now a competitive position in its own right.

The commercial effect is pricing power in a supply-constrained market. Chief executive Greg Goodman has framed the strategy as providing essential digital infrastructure in supply-constrained markets, building into strong demand for metro locations across both logistics and data centres. When customers cannot go elsewhere, the developer captures more of the value.

How does the funds management model work?

Goodman develops assets and then sells stakes into partnerships co-funded with institutional investors, earning development profit, ongoing management fees and performance fees while retaining a share of the asset. That structure allows a development programme far larger than the group’s own balance sheet could support.

The mechanics are visible in recent activity. Goodman established a partnership in Europe reported at around €14 billion for data centre development, launched a data centre partnership in Australia, established a US$2.7 billion Hong Kong data centre partnership, and sold a completed data centre into its Japan data centre partnership. Roughly A$7.2 billion of third-party equity was raised over an eighteen-month period.

For investors the model produces higher returns on Goodman’s own capital and introduces dependence on partner appetite. A development programme funded by rotating capital into partnerships works while institutional investors want the exposure; if they stop, the pipeline must either slow or move onto the balance sheet.

What are the risks?

Concentration in a single thesis is the obvious one. With 73% of work in progress dedicated to data centres, Goodman’s development earnings now depend heavily on continued hyperscale capital expenditure. Morningstar has noted the company has been among the worst performers in its sector coverage recently, with building concerns about a potential artificial intelligence bubble and data centre overspending.

Returns on invested capital are the second. Morningstar expects data centre development and leasing to contribute roughly one-fifth of Goodman’s earnings in the medium term, up from less than 10%, while cautioning that returns could fall as global competition intensifies. Every major developer has noticed the same opportunity, and land with power will attract competing bidders.

Dilution is the third. The A$4 billion capital raising undertaken to strengthen the balance sheet for new developments supports the growth ambition and increases the security count against which earnings are measured. A development programme of this scale requires equity, and equity issued at the wrong point in a cycle is expensive in a way that is only visible later.

What does the outlook say?

Growth, funded and forecast. Goodman is targeting 9% operating earnings per security growth in FY2026, equating to more than A$2.6 billion of operating profit, on the back of FY2025 revenue of A$2.31 billion and net income of A$1.67 billion. Work in progress is expected to rise substantially, with data centre projects under construction growing from around 300 megawatts toward 500 megawatts during 2026.

The Australian dimension matters domestically. Goodman has commenced a 90 megawatt data centre project in Artarmon in metropolitan Sydney, and metro Sydney has become one of the country’s key digital infrastructure hubs driven by cloud and AI workloads — while grid capacity remains the gating factor for further development.

For Australian investors the broader significance is that the largest listed property company in the country is now substantially a digital infrastructure business. That reflects where returns have moved, and it means the ASX property sector’s largest constituent has almost nothing to do with the residential construction crisis that dominates domestic property discussion.

What makes a data centre site valuable?

Four things in descending order of scarcity: available grid power, proximity to network infrastructure, land with appropriate zoning, and water or alternative cooling capacity. Power is the binding constraint almost everywhere, which is why Goodman’s power bank matters more than its land bank.

Latency drives the location requirement. Cloud and AI workloads serving users in a city need to sit near that city, which rules out the cheap remote land where power might be easier to obtain. Metropolitan sites with existing industrial-scale electrical connections are a small and finite set in any market.

Zoning is the quiet obstacle. Industrial land is being rezoned for residential use across most Australian cities because housing is politically urgent, which steadily reduces the supply of sites suitable for either logistics or data centres. A developer holding zoned, powered, metropolitan industrial land is holding something that cannot be replicated by anyone with capital.

How does this compare with the logistics boom?

It is the same land producing a higher-value use, which is a rerating rather than a new business. Goodman’s logistics thesis was that e-commerce required warehouse space near customers, and it was correct – industrial property delivered the strongest rental growth of any Australian property sector for most of the past decade.

The data centre thesis rests on a similar structural argument with a different customer. Hyperscale operators need metropolitan capacity for latency-sensitive workloads, they sign long leases, and they have creditworthiness that a third-party logistics operator does not. On those terms a data centre tenant is a better covenant than a warehouse tenant.

The differences are capital intensity and obsolescence. A warehouse costs a fraction per square metre of a data centre and can be re-tenanted by almost any occupier. A data centre is purpose-built, expensive, and its power and cooling specifications are set by the hardware generation it was designed for. The returns are higher and so is the risk, which is the trade every developer in the sector is now making.

A final note on how to read the disclosures. Goodman reports work in progress, the proportion dedicated to data centres, the power bank in gigawatts and the split between secured and allocated capacity. Of those, secured power is the number that constrains everything else, because allocated capacity can be reduced or delayed by network operators while secured capacity cannot. The gap between the 6.4 gigawatt power bank and the 3.6 gigawatts secured is therefore the most informative figure in the update, and it is the one that determines how much of the pipeline is genuinely deliverable on schedule.

The competitive position is worth stating plainly. Every large industrial developer globally has identified the same opportunity, and hyperscale tenants are sophisticated buyers who will run competitive processes across multiple sites. What differentiates Goodman is not insight – the thesis is widely shared – but the fact that it already owns the land and holds the power allocations, accumulated over decades at logistics prices. That head start is finite and cannot be extended, which is why the pipeline is being committed so quickly.

Frequently Asked Questions