Meta description: Data center leases are priced by the megawatt, not the square foot. A practical guide to what that does to Yardi, your ERP, your controls, and your M&A integrations, drawn from how Prologis, Goodman, and hyperscale developers are handling it.

Outside Chicago, a building used to be a warehouse. Trucks backed into it. Forklifts ran inside it. Today it is a 50-megawatt turnkey data center, owned by Prologis, the largest logistics landlord in the world, and leased to a company that measures its needs in power, not pallets.

That building captures what is happening across industrial real estate right now. Prologis told investors last week that its data center pipeline has reached 5.8 gigawatts, with 1.7 gigawatts of secured power and about a gigawatt already under letters of intent. Meanwhile, Goodman Group, the Sydney-based logistics developer, reported that data centers now make up 78% of its $19.7 billion development book, up from 57% a year earlier, with a 6.4-gigawatt power bank across 16 cities. In about 24 months, two of the world's biggest industrial landlords have become two of the world's biggest data center developers.

Most of the coverage is about power, land, and hyperscalers. Almost none of it is about the part that lands on the CFO's desk the morning after the lease is signed: the property management system now has to bill a tenant by the megawatt, recover a $400-million fit-out through rent, pass through a utility bill the size of a small town's, and roll all of that into a set of accounts that a corporate ERP, an auditor and a joint-venture partner can trust.

This is a guide to that part. It is written for the finance and technology people at owners, developers, and investors who are discovering that "we'll just add a property type" is not how it works, and for the advisers helping them. Examples are public where they can be and anonymized where they cannot.

A logistics lease is simple at heart. The tenant rents square feet, pays base rent plus a share of taxes, insurance, and maintenance, and the landlord's job is to keep the roof on. The property system has been built around that shape for thirty years.

A data center lease starts from a different question: how much power can you deliver, and when? Everything else follows from the answer. Four things change at once.

The unit of measure. Rent is quoted per kilowatt or megawatt of critical IT load per month, not per square foot. A 50-megawatt building might have less rentable floor area than a mid-sized warehouse, and far more revenue. Square footage still exists for tax and insurance, but it is no longer what the tenant is buying.

The two delivery models. Prologis's own language, repeated on every earnings call, is powered shell versus turnkey. In a powered shell, the landlord delivers the building, the substation, and the grid connection; the tenant installs its own cooling, backup power, and racks. It looks a lot like a long triple-net industrial lease. In a turnkey, the landlord delivers all of that mechanical and electrical fit-out as well, and the lease price includes a recovery of that capital over the term. Prologis has said its pipeline represents roughly $17 billion in powered-shell investment, or up to $87 billion on a turnkey basis, which shows how different the two economics are. Many portfolios will end up with both models on the same campus.

The pass-throughs. Electricity is the highest operating cost and is usually metered and passed through at cost, sometimes with a margin, sometimes at a blended rate the tenant has negotiated. That makes utility billing a core revenue process rather than a year-end reconciliation.

The obligations. Data center leases carry service-level commitments on uptime, temperature, and power availability, with financial penalties if they are missed. A warehouse landlord has never had to book a credit because the building was too warm.

None of this is exotic to a data center REIT that has done it for twenty years. It is new to a logistics landlord, an industrial fund manager, or a developer who built the business on square feet, and that is exactly who is entering the market now.

A recurring misunderstanding in large real estate organizations is worth clearing up because it shapes how people budget, staff, and buy advice.

In most large REITs and funds, the corporate ERP (Oracle, SAP, or Workday) is the general ledger of record for the group. This is where consolidation, intercompany, treasury, and statutory reporting happen. The property management system (Yardi Voyager for most North American and APAC industrial owners, MRI for many in the UK) is where the leases, the tenants, the recoveries, the work orders, and the property-level accounting live. Prologis, by public account, has run Yardi Voyager across more than 4,000 properties and 20,000 leases for years. The two systems are not competing for the same job. They are layers.

The ERP cannot bill a tenant for 42 megawatts of contracted capacity at a stepped rate with a true-up against metered draw. It was never meant to. The property system cannot produce the group's consolidated multi-currency financial statements or run the board-level planning model. It was never meant to either. The value sits in the hand-off: the property layer has to produce clean, auditable, property-level numbers that the corporate layer can consume without a small army re-keying and reconciling in spreadsheets.

This matters more for data centers than for warehouses, for three reasons. Revenue per asset is higher, so errors are larger. The capital structures are more complex, with joint ventures and capital partners on most campuses because, as Prologis has said, it will not hold 100% of its data centers on the balance sheet. And the lease-and-recovery logic is new, so the property layer is where configuration decisions are made, and those decisions determine whether the numbers flowing up to the ERP are right.

For an ERP-centred finance leader, the practical implication is this: the Yardi or MRI layer is small in license dollars and large in financial exposure. Producing trustworthy megawatt-based numbers is specialized work; it rarely sits inside the ERP practice, and it sits on the critical path of every data center consolidation, SOX program, and acquisition integration the group will run in the next five years.

Yardi Voyager can run a data center portfolio well. What it cannot do is guess how you want it run. Below are the configuration decisions that separate a clean data center book from one that gets rebuilt eighteen months after go-live. They apply, with different menu names, to MRI as well.

Charge codes built around capacity, not area. Contracted critical load, actual metered draw, reserved-capacity charges, and ramp schedules each need their own charge code and billing logic. Forcing them into a square-foot rent template produces numbers that reconcile to nothing. Treat kilowatts as the primary unit and carry area as an attribute.

Metered power as a revenue stream. Utility recovery isn't a once-a-year CAM reconciliation; it is a monthly billing cycle with meter reads, changing rates, and true-ups. Decide early whether meter data flows in from the building management system automatically or is keyed, and who owns the reconciliation when the utility bill and the tenant bills do not match.

Capital recovery as its own stream. In a turnkey lease, part of the rent is recovery of fit-out capital over the term. If that is blended into base rent, the asset-level return, the lease accounting, and the partner reporting all go wrong. Separate it, tie it to the asset it recovers, and make the amortization visible.

Job cost for build-to-suit. Prologis has described more than $2 billion of data center starts this year, all build-to-suit. That capital must land on the right asset through the construction and job-cost modules, with the turnkey fit-out separated from the shell. Hence, depreciation, recovery, and valuation all start from the same cost base.

Capital partners and valuation. Most campuses will sit in a joint venture or a managed fund. The Investment Manager and the budgeting and valuation modules must model the same building two ways, at the asset and at the partner share, and the waterfalls must match the partnership agreement, not a logistics-side template.

Service levels as liabilities. Uptime credits and penalties need a home in the system, with a workflow that captures the incident, calculates the credit, and posts it against the right tenant and period. Spreadsheets here are an audit finding waiting to happen.

None of this is a reason to delay the first lease. It is a reason to make these six decisions deliberately, with finance in the room, before the first invoice goes out.

The data center land grab is being done partly by building and partly by buying, and buying is where the systems work gets hardest.

On 28 September, SEGRO shareholders approved the combination with Prologis. This roughly £14 billion deal will give Prologis a 363-million-square-foot European operating portfolio plus SEGRO's own data center sites, with closing expected in the first half of 2027. Prologis's European book has been on Yardi Voyager International for well over a decade. Whatever SEGRO runs today, those leases, tenants, recoveries, and histories will land in Prologis's Yardi instance, in multiple currencies, under UK and continental lease conventions, with a data center layer on top.

That is a data conversion, and data conversions are where integration timelines quietly die. The pattern repeats on every large deal, including Prologis's $23 billion absorption of Duke Realty in 2022: the deal team models synergies in months, the systems team discovers that the target's lease data has thirty years of inconsistent coding, and the first consolidated close slips.

Three things make data center conversions harder than logistics ones.

• The lease data is richer and stranger. Capacity commitments, ramp schedules, power rates, SLA terms, and fit-out recovery schedules do not map to the standard lease fields a logistics conversion expects. They have to be found, often in PDFs, and modeled before they can be loaded.

• The history matters. A hyperscaler tenant with a twenty-year lease and five amendments needs its full billing and recovery history to survive the move, because that history drives true-ups and disputes years later.

• The source systems vary. Targets may be on MRI, SAP, a bespoke platform, or spreadsheets. Every source needs its own extraction and mapping, and doing that by hand for each deal is slow and error-prone.

Owners who handle this well treat conversion as an engineering problem rather than a data-entry problem: a repeatable middleware layer that extracts from the source, validates against rules, maps to the target schema, and reconciles before load, so the second acquisition is faster than the first. Firms that have built that tooling because they got tired of doing MRI-to-Yardi and SAP-to-Yardi conversions by hand are, not coincidentally, the ones getting called when the next deal closes.

Goodman's numbers show how fast this is moving. In its FY2025 results, data centers were 57% of development work in progress. A year later, they made up 78% of a $19.7 billion book, with more than $15 billion in data center projects underway and a 90-megawatt build started in Sydney. Prologis went from launching a data center segment in mid-2024 to a 5.8-gigawatt pipeline today. They told analysts that data center supply-chain customers now account for 17% of its US leasing, in addition to the data centers themselves.

Neither announcement shows what happened inside the finance and systems teams, which in both cases were built for logistics. From working with owners going through the same shift, the pattern looks like this.

Year one: the new asset type lands in the old chart of accounts. The first data center is set up like any other property, with a few custom charge codes. It works, roughly, because there is one tenant and one building.

Year two: the exceptions multiply. Three campuses, two delivery models, a capital partner on one of them, utility true-ups that nobody owns, and an auditor asking how the fit-out recovery is being recognized. The monthly close for the data center book now takes longer than closing a thousand warehouses.

Year three: the rebuild. Someone commissions a proper design: a data center property type with its own charge code set, metered utility workflow, capital recovery stream, job-cost structure and partner reporting, with the ERP interface rebuilt to match. This is the work that should have happened in year one.

The owners who skip the painful middle year are the ones who treat the first data center as a design exercise, not a property set-up. They bring finance, the property-system specialists, and the ERP team together before the first lease, write down how a megawatt lease will flow from the lease abstract to the consolidated statements, and build it once.

There is a cultural point too. Logistics finance teams are used to many small, similar leases. Data center finance involves a few very large, very different leases, where one misconfigured recovery schedule can be a seven-figure error. People and controls have to change along with the system.

A fast-growing data center developer often ends up in the same place: a public-reporting or capital-markets event is coming, auditors want SOX-style controls over the property system, and a small team configured the property system while moving quickly across several countries.

From fieldwork with a hyperscale developer operating across North America and Europe (anonymized here), three findings come up more often than any others.

The control gap is in the property layer, not the ERP. The ERP has been through SOX before; it has approval workflows, segregation of duties, and audit logs that auditors recognize. The property system has user roles set up at go-live and never reviewed, charge code changes nobody signed off on, and manual journals bridging what the system couldn't do. The remediation plan is a property-system plan.

Multi-currency is where the layers collide. A campus in Frankfurt, a partner in New York, and reporting in US dollars means translation, intercompany, and consolidation logic has to live somewhere. The open question on most of these programs is whether that logic sits in Yardi, in the Oracle EPM or similar corporate planning layer above it, or, as often happens by accident, in both. Deciding that explicitly, and diagnosing where the current numbers actually come from, is a project in its own right and should be scoped separately from the control work.

The evidence has to be reproducible. An auditor does not want to be told the utility true-up was reviewed. They want to see the review, the underlying data, and the sign-off for every period. That means moving the true-up out of spreadsheets and into a workflow with a trail, which in turn means the configuration decisions from earlier in this piece have to have been made.

The lesson is simple to state. Diagnosis comes before remediation, and remediation of the property layer comes before the group can credibly claim control over its data center numbers. Owners who have run the diagnostic early have a roadmap and a defensible timeline when the auditors arrive. Owners who haven't are doing the diagnosis under audit pressure, the most expensive way to do it.

If you are a CFO, controller, or CIO at an owner moving into data centers, these are the questions worth settling before the first invoice, whether you are on Yardi, MRI, or something else.

1. What is our unit of measure for rent, and does every downstream report, from the rent roll to the board pack, use it?

2. Will we deliver powered shell, turnkey, or both, and does the system distinguish them at the lease, the asset, and the capital level?

3. How does metered power get from the building management system into tenant billing, and who reconciles it to the utility invoice each month?

4. Where does fit-out capital recovery live, and can we show the asset-level return with and without it?

5. Is build-to-suit spend flowing through job cost onto the right asset, with shell and fit-out separated?

6. For each campus with a capital partner, can the system produce the asset view and the partner-share view from the same data?

7. Where do uptime credits and SLA penalties get captured, calculated, and posted?

8. Which numbers flow from the property system to the ERP, in what form, and where do multi-currency translation and consolidation actually happen?

9. Who approves a charge code change, a user role change, or a manual journal in the property system, and where is the evidence?

10. When we acquire a portfolio, what is our repeatable process for converting its lease data, and how long did the last one take?

Most organizations can answer three or four of these cleanly. The gaps are the design brief.

Large advisory firms are very good at the corporate layer: ERP transformation, EPM, consolidation, controls, deal integration at the group level. That is where the big programs and the big budgets sit, and rightly so.

The data center shift exposes a gap just underneath it. The property layer, where megawatt leases, recoveries, job cost, and partner reporting are configured, is specialized, platform-specific, and increasingly on the critical path of the corporate programs above it. A SOX program stalls if the property system has no control design. An ERP consolidation produces wrong numbers if the property system is sending it blended rent. An acquisition integration slips if nobody can convert the target's leases.

The firms that handle this best pair the two: a large-firm program lead who owns the corporate outcome, and a specialist Yardi or MRI bench that has done data center configuration, SOX control gap assessments, multi-currency diagnostics, and conversion tooling on these platforms before. The specialist does not need to be big. It needs to have seen a turnkey lease billed by the megawatt, a logistics landlord mid-pivot, and a hyperscale developer three months from an audit, and a delivery team that can work across North American, European, and APAC time zones because that is where the campuses are.

Assetsoft, a Markham, Ontario-based consultancy that has implemented and supported Yardi and MRI for owners in Canada, the United States, Australia, and the Middle East since 2012, is one of a small number of firms doing exactly this kind of work today: data center control gap assessments and multi-currency diagnostics alongside big-firm programs, Yardi data strategy for owners whose property data has to feed an ERP and a data platform, and conversion tooling built after too many MRI-to-Yardi and SAP-to-Yardi migrations done by hand. Its founder has spent 25 years in property systems, starting at Yardi itself, which is roughly how long it takes to stop being surprised by what a lease can contain.

The point is not the firm. The point is that the data center boom is a property-systems story as much as a power story, and the owners who treat it that way will close faster, integrate faster, and pass their audits with less drama than the ones who do not.