Market analysis shows supply constraints and permitting delays for new data center construction are boosting the value and utilization of existing capacity.

Existing data centers, which can often be upgraded to support AI demand, remain largely protected from grid constraints and interconnection moratoriums that impinge on new development, according to industry experts.

Data center capacity is extremely limited. Most tenants today secure space in small, fragmented blocks and are contracting for 2028 deliveries, according to JLL's 2026 North America Data Center Midyear report released Tuesday. North America has roughly doubled net new demand in the first half of 2026 to 25GW, the report states. Vacancy has held at 1% for three consecutive years despite unprecedented construction activity, reflecting demand driven by digital services and AI adoption.

The difficulty in procuring and delivering new capacity is disrupting development pipelines and increasing reliance on existing, stabilized facilities, according to Morningstar. Because these assets already have contractually reserved capacity, they are largely shielded from grid constraints and new interconnection moratoriums. "Once operational, data centers are substantially protected from grid constraints and new interconnection moratoriums through binding utility agreements that fix capacity rights, allocate infrastructure risk, and define service priority," Morningstar noted.

Older facilities retain significant value. "Facilities that are 10, 15, or 20 years old still have power and fiber in place, and still have mechanical plants with many years of useful life," said Sean Farney, vice president of data center strategy at JLL. Hyperscale, colocation, and enterprise organizations are upgrading these older data centers to support AI computing requirements.

Existing data center clusters create a "self-reinforcing concentration effect" as operators seek proximity to dense fiber routes, cloud availability zones, carrier hotels, and pools of skilled labor in major metropolitan areas. This dynamic is supporting "robust operating performance for stabilized assets," Morningstar stated. Data center rents have increased approximately 9% year over year since 2020. Larger deployments exceeding 20 megawatts are now commanding average rents of approximately $141 per kilowatt-hour, excluding energy costs, reflecting a premium for high-density, hyperscaler-ready capacity.

However, these facilities are not wholly insulated from power scarcity. "This insulation is not absolute," Morningstar cautioned. While contractual structures improve cost certainty and ESG alignment, they generally do not displace the utility's role in physical delivery. Utilities and system operators retain control over transmission, balancing, and delivery. As electricity demand from data centers outpaces infrastructure expansion, utilities and regulators are increasingly incorporating curtailment provisions, demand response obligations, and load-shedding mechanisms into large-load frameworks that could force operators to reduce power consumption during grid emergencies.

The Department of Energy authorized PJM Interconnection to divert power from data centers and large industrial facilities to residential households during extreme cold snaps to prevent blackouts. Texas enacted rules requiring data centers with loads exceeding 75MW to participate in mandatory demand management programs.

"Stabilized data centers should be viewed as contractually secured but systemically contingent assets—protected from routine grid constraints under existing agreements, yet still exposed to term and tail-risk scenarios involving emergency curtailment, regulatory intervention, or structural changes to electricity market design," Morningstar said.

Data center landlords are capturing substantial rent increases on renewals, up 70% on average since 2020. Most leases signed today carry annual escalations of 3% or more with no concessions.

Artificial intelligence drives more than 50% of new data center capacity demand. However, even markets with attractive energy prices may remain "commercially unviable if substations are full, transmission upgrades are backlogged, interconnection studies are slow, or utility operators cannot assure adequate generation capacity during peak usage," Morningstar observed.

While existing facilities may not support the liquid cooling required for AI GPU storage compute, they can be retrofitted. "This is happening across hyperscale, co-location, and enterprise facilities," Farney said. Enterprise users are increasingly managing hybrid portfolios of on-premise, colocation, and cloud infrastructure, with individual site requirements ranging from 500 kW to 3 MW.

"Some enterprises are retrofitting their on-premise data centers, which largely stood empty over the past decade, moving compute loads to the cloud," Farney explained. "This leaves stranded power that is perfect for building test and AI environments." These projects offer opportunities to reduce long-term spend versus relying on cloud-based AI while maintaining low-latency access and on-site staff accessibility.

"Running AI inference in the cloud can be expensive over time and may be limited by latency, so enterprises are building their own on-premise AI inference lab environments in formerly abandoned facilities," Farney said.

The primary challenges lie in retrofitting mechanical and electrical infrastructure. However, if systems have been properly maintained and adequately sized with quality equipment, retrofits are manageable.

"Yet even this group is struggling to secure capacity to support business growth," according to JLL.