Friday, October 16, 2026
AI Infrastructure · News & Analysis
Home › Data Centers › Report
Data Centers · Report

JLL says strong AI demand is not enough to protect data center developer margins for projects that lack power certainty, anchor tenants or credible delivery timelines as construction costs rise.

Developers without secured power, a committed anchor tenant or a credible schedule face squeezed returns, so capital is likely to concentrate on already-contracted campuses.
Trade pressSlicast · October 15, 2026 at 15:00 UTC · Global · Source: Data Center Knowledge
importance 50

Data center developers may be able to pass higher construction costs on to tenants, but only when projects have power access, committed customers, and a credible path to delivery, according to JLL.

In its Global Construction Costs Index 2026, the real estate services firm said annual construction cost escalation of 3% to 6% has become the new baseline, up from the 2% to 3% range common before 2020. JLL also forecasts another 3.1% to 4.2% increase in global tender prices in 2027, including 3.2% to 4.2% in North America. Those forecasts apply to construction broadly rather than the data center sector alone.

The JLL report notes that rising rents and asset values have helped data center projects absorb some of that inflation. However, Matthew Giddings, executive director of technical services with JLL's Data Centers and Critical Environments team, said stronger market fundamentals do not automatically protect individual projects.

"The economics break down when costs outpace contracted revenue or power delays postpone income while financing costs continue to accumulate," Giddings told Data Center Knowledge. "Higher asset values alone cannot close that gap."

That risk is most acute for speculative campuses without anchor tenants, projects with uncertain power delivery, and developments tied to fixed-rent agreements that leave little room to absorb construction escalation. Older facilities being repositioned for AI workloads may also face difficult economics if upgrades become too extensive.

"Strong demand supports the market, but it does not make every project viable," Giddings said.

Developers should reassess projects when rising capital costs erode expected returns, utility schedules slip beyond committed occupancy dates, or tenant demand no longer supports planned capacity, he said. Reduced power availability and delays in anchor-tenant commitments are also signs that a project may need to be reworked.

One way to manage that risk is through phased construction. By aligning spending with confirmed electricity supply and customer demand, developers can limit the amount of capital they commit before fundamentals are in place.

Labor constraints are adding to the pressure. Giddings said the electrical and mechanical trades are creating the greatest cost and schedule challenges, particularly electricians, high-voltage specialists, pipefitters, and advanced cooling technicians.

"Controls, commissioning, and startup teams have also become critical bottlenecks, as their availability often determines when a completed facility can be placed into service," he noted.

Competition for utility and substation construction crews is also intensifying, adding risks that extend beyond the data center site itself. To reduce exposure, Giddings recommended that developers bring in trade partners early and test project schedules against real workforce availability rather than assumed labor capacity.

He also pointed to standardized designs, off-site prefabrication, and phased construction as ways to reduce peak labor requirements. In remote markets, developers may need to offer housing, transportation, and travel incentives to attract workers. In the longer term, investment in trade schools, apprenticeships, and union training facilities could help support future construction phases while expanding local employment opportunities, he said.

Procurement strategy is becoming just as important. Long lead times for critical equipment have pushed many developers to place orders earlier, but doing so can create problems if tenant requirements shift, particularly as AI infrastructure continues to evolve.

"The key is to secure manufacturing capacity early without freezing every design decision," Giddings said.

He said developers should reserve production slots for transformers, switchgear, generators, and cooling equipment while negotiating flexibility to modify specifications before fabrication begins. Orders should advance in line with validated tenant requirements and clear decision deadlines.

That approach is especially important as operators try to account for shifting rack densities and growing interest in liquid cooling. Modular power and cooling systems, built-in expansion capacity, and design provisions for liquid cooling can help facilities adapt without undermining the overall project.

"Early procurement protects the schedule, while flexible design protects the long-term value of the asset," Giddings said.

Retrofitting older facilities can offer another path to faster delivery and lower upfront costs, particularly when a site already has available power, sufficient structural capacity, and infrastructure that can be reused. However, those advantages can disappear quickly if a building requires major mechanical upgrades, higher electrical distribution capacity, heavier floor loading, or large-scale liquid-cooling retrofits.

"The economics ultimately depend on how much existing infrastructure can be retained while still meeting current and future performance requirements," Giddings said.

Read the original
JLL says strong AI demand is not enough to… · Slicast