The AI Boom Is Creating a New Market for Data Center Sale-Leasebacks

As Infrastructure Spending Accelerates, Owning the Real Estate Is No Longer the Only Strategy

The artificial intelligence boom is forcing companies to reconsider one of the most fundamental questions in data center real estate:

Do we actually need to own the facility we operate?

For decades, ownership made strategic sense for many enterprises and technology companies. Mission-critical infrastructure was closely tied to the real estate beneath it, and controlling both offered security, operational flexibility, and long-term certainty.

But the economics surrounding digital infrastructure are changing.

AI requires extraordinary amounts of capital. Companies are simultaneously investing in GPUs, servers, networking equipment, cooling systems, electrical infrastructure, power generation, and new data center capacity.

Worldwide data center capital expenditures continue rising as hyperscale AI deployments accelerate. Meanwhile, the largest technology companies have accumulated enormous long-term commitments related to compute, energy, equipment, and infrastructure.

That creates a simple capital allocation question.

If hundreds of millions of dollars are tied up in the real estate beneath a data center, could that capital generate greater value somewhere else?

For some owners, the answer may increasingly be yes.

And that is putting renewed attention on one of commercial real estate's most established transaction structures:

the sale-leaseback.

Selling the Data Center Doesn't Necessarily Mean Leaving It

A sale-leaseback separates ownership from occupancy.

The owner sells the property to an investor and simultaneously enters into a lease allowing it to continue operating from the facility.

Operationally, relatively little may change.

The servers stay.

The customers stay.

The infrastructure stays.

The company continues occupying the facility.

What changes is the ownership of the real estate—and the balance sheet.

The seller converts an illiquid property asset into capital while retaining long-term use of the facility. The buyer acquires mission-critical real estate backed by contractual rental income.

Sale-leasebacks are already established within data center transactions. JLL, for example, completed a $34 million sale-leaseback of a 103,000-square-foot mission-critical facility in Doral, Florida, where the seller continued leasing a portion of the property following the transaction.

But today's AI investment cycle makes the structure particularly relevant.

AI Has Changed the Capital Allocation Equation

The data center itself is becoming more expensive.

But so is everything inside it.

AI operators need enormous quantities of accelerated computing infrastructure. They also need increasingly sophisticated networking, electrical systems, cooling architecture, and supporting energy infrastructure.

That means companies face competing demands for capital.

Owning valuable real estate may make sense strategically.

But so might selling that property and redeploying the proceeds into the infrastructure directly responsible for generating compute revenue.

The scale of current AI spending makes that tradeoff increasingly important. A United Nations preliminary report published in July estimated that major hyperscaler capital expenditure had risen roughly fivefold since 2023, reaching an annualized level of around $650 billion in early 2026.

The real estate question therefore becomes part of a much larger corporate finance decision.

What assets should the company own, and what assets should it simply control?

Real Estate Can Become a Source of Growth Capital

Data centers are unusually capital-intensive properties.

A mature facility can represent significant embedded real estate value, particularly when it offers characteristics such as:

1. secured power
2. established fiber connectivity
3. strong market positioning
4. long-term operational history
5. expansion potential

Selling that property can unlock capital without requiring the operator to abandon the location.

The proceeds can potentially be redirected toward new campuses, equipment, acquisitions, power infrastructure, or other strategic priorities.

This is why the sale-leaseback should not necessarily be viewed as an exit.

In the right circumstances, it can be a capital recycling strategy.

The owner monetizes an established asset and redirects the capital toward the next phase of growth.

The Buyer Sees a Different Asset

The operator and the real estate investor are evaluating the same facility through different lenses.

The operator sees infrastructure.

The investor sees contractual income backed by mission-critical occupancy.

For an institutional real estate buyer, an attractive sale-leaseback can provide a combination of:

long-term tenancy, predictable income, specialized infrastructure, and exposure to digital infrastructure growth.

The quality of the tenant becomes especially important.

A long-term lease with a financially strong occupant can fundamentally change the investment profile of the property.

This is one reason data center capital markets increasingly involve infrastructure funds, private equity, institutional real estate investors, and other sources of private capital alongside traditional operators.

CBRE describes sale-leasebacks as an increasingly relevant capital markets solution for the scale and capital intensity of data center development, particularly as owners seek to release capital from existing assets while maintaining operations.

The Lease Can Be Almost as Important as the Building

In a conventional property transaction, much of the attention naturally falls on the physical asset.

In a data center sale-leaseback, the lease can be equally important.

Buyers need to understand:

How long will the tenant remain?

What are the rent escalations?

Who is responsible for capital improvements?

Who maintains specialized infrastructure?

What happens if major electrical or cooling upgrades become necessary?

Can the tenant expand?

Can excess capacity be leased to another user?

What happens at the end of the lease?

These questions directly influence value.

A technically impressive facility paired with an unfavorable lease structure can create a very different investment proposition from the same facility under a strong long-term agreement.

Power Changes the Valuation Conversation

There is another reason existing data centers can be particularly interesting sale-leaseback candidates today.

Power.

In many markets, obtaining significant new electrical capacity has become one of the largest obstacles to development.

An existing facility with energized capacity may therefore possess strategic value that goes beyond the physical building.

A buyer isn't simply acquiring square footage.

It may be acquiring real estate connected to infrastructure that could take years to replicate elsewhere.

That can make established facilities especially interesting in constrained markets.

But investors must distinguish between existing power, contracted power, and theoretical future capacity.

The difference can materially affect both valuation and expansion potential.

Expansion Rights Can Create a Second Investment Thesis

The strongest sale-leaseback opportunities may offer more than stable income from an existing facility.

They may also provide future development potential.

Consider an operational data center occupying part of a larger campus.

The seller leases back the existing facility, but the buyer also acquires:

additional land, unused power capacity, development rights, or space capable of supporting another building.

Now the transaction has two components.

The first is the stabilized income generated by the lease.

The second is the potential value of future development.

In a market where development-ready land and power are scarce, that optionality can be significant.

Not Every Data Center Is a Good Sale-Leaseback Candidate

The strategy has limitations.

Older enterprise facilities may require significant modernization.

Some were designed for computing environments that look very different from today's high-density AI workloads.

Others may suffer from limited power scalability, inefficient cooling systems, poor connectivity, or constrained sites.

Lease structure can also create complications.

A seller seeking maximum operational flexibility may resist the long-term commitments investors typically prefer.

Similarly, an investor may be reluctant to acquire a highly specialized facility if its value depends almost entirely on a single occupant.

That creates an important diligence question:

What happens to the property if the current tenant eventually leaves?

The answer matters enormously.

Residual Value Is Becoming a Bigger Question

A warehouse can usually accommodate another logistics tenant.

An office building can potentially attract another company.

A highly specialized data center is different.

Its value may depend heavily on the usefulness of its power, cooling, connectivity, and physical configuration to another operator.

Investors therefore need to underwrite not only today's lease income but the property's future relevance.

Can the facility support higher-density workloads?

Can electrical capacity increase?

Can cooling infrastructure be modernized?

Could another operator use the building?

Does the site itself have redevelopment value?

In some transactions, the underlying land and power position may ultimately provide more downside protection than the existing building.

Long-Term Leasing Is Already Central to the AI Buildout

The broader AI infrastructure market is already demonstrating how important long-term occupancy structures can be.

In August 2026, Goodman Group secured a 20-year hyperscale lease for the first 50 MW phase of its planned 1 GW Tsukuba Tech Central campus near Tokyo. The lease helped validate a major development without requiring the hyperscale customer to own the underlying campus.

Even larger financing structures are emerging around AI infrastructure in the United States. OpenAI has reportedly entered into a long-term arrangement for a massive Ohio campus being developed and operated by SB Energy, with Nvidia providing substantial financial support around the infrastructure.

These are not traditional sale-leasebacks.

But they demonstrate the same broader principle:

The company consuming the computing capacity does not necessarily need to own the underlying real estate and infrastructure.

That separation is becoming increasingly important as AI infrastructure grows more capital-intensive.

Enterprise-Owned Data Centers May Be Especially Interesting

Hyperscalers receive most of the industry's attention, but the sale-leaseback opportunity extends well beyond Big Tech.

Large enterprises still own mission-critical data centers accumulated over decades.

Banks.

Insurance companies.

Healthcare organizations.

Telecommunications companies.

Retailers.

Government contractors.

Other large corporations.

Some of these facilities remain strategically important.

But owning the underlying real estate may no longer be central to the company's business model.

JLL notes that owners of aging data centers increasingly face choices between upgrading, selling, repurposing, partially leasing, or executing sale-leaseback strategies as they reassess legacy infrastructure portfolios.

That creates potential opportunities for both sellers and specialized buyers.

Sale-Leasebacks Could Become Part of the AI Capital Recycling Cycle

The larger trend is not simply about real estate transactions.

It is about how the industry finances growth.

The current AI buildout requires enormous amounts of capital, and companies are increasingly combining corporate balance sheets with leases, joint ventures, private capital, infrastructure financing, and other structures.

Sale-leasebacks fit naturally into that environment.

An owner develops or operates a valuable facility.

The asset matures.

Institutional capital acquires the real estate.

The operator retains occupancy.

Capital is released.

That capital can then be redeployed into additional infrastructure.

The cycle can begin again.

For companies pursuing aggressive expansion, the ability to recycle capital may become almost as important as the ability to develop new facilities.

What This Means for Data Center Real Estate

For owners, the current market creates an opportunity to reconsider whether real estate should remain permanently on the balance sheet.

For investors, it creates access to specialized assets supported by long-term digital infrastructure demand.

For brokers and advisors, it creates increasingly complex transactions where real estate, infrastructure, credit, power, and corporate strategy intersect.

And for operators, it introduces another way to finance expansion without necessarily surrendering operational control.

That is ultimately what makes sale-leasebacks particularly relevant to the current AI cycle.

They separate two things that have historically been closely connected:

using a data center and owning it.

Ownership Is Becoming a Strategic Choice

The AI boom is forcing the data center industry to rethink capital.

Companies need more computing infrastructure, more power, more land, and more facilities—and they need them simultaneously.

Against that backdrop, every dollar tied up in an existing asset deserves scrutiny.

For some organizations, owning the real estate will remain strategically important.

For others, the greater opportunity may be unlocking the value of an existing facility while continuing to operate from it.

That is why sale-leasebacks deserve more attention in the next phase of data center growth.

The question is no longer simply:

What is this data center worth?

It is:

Who should own it, and where could that capital create more value?