The data center land grab is accelerating. As hyperscalers and colocation providers race to secure sites for AI infrastructure, investors are finding opportunity beyond the usual semiconductor and cloud names — in the land itself.
Back
Data center land grab creates 3 investment plays for AI buildout
The AI infrastructure buildout is driving a land grab for data center sites across the US, creating investment opportunities in real estate, modular infrastructure, and IT services that extend beyond the hyperscaler names dominating headlines.
"The demand for data center capacity is outstripping available power and land in traditional markets like Northern Virginia," Mike Santos, data center infrastructure analyst at Edgen, said. "Investors are now looking at secondary markets and modular approaches as the next wave."
Hyperscale data centers require enormous parcels — often 100 acres or more — along with access to 100-plus megawatts of power and millions of gallons of water for cooling. That scarcity is pushing developers into new regions and alternative models. North Carolina, for example, has emerged as a candidate for modular edge data centers that consume far less land and water, according to a recent analysis from the Carolina Journal. These smaller facilities, optimized for AI inference workloads rather than model training, can be deployed in urban, suburban and rural areas without the massive infrastructure upgrades hyperscale projects require.
Three ways to play the trend
The first play is through data center REITs and land-banking entities that own sites pre-zoned for data center development. As suitable parcels grow scarcer in Northern Virginia, the world's largest data center market, landowners with approved permits and power allocations command significant premiums.
The second is through infrastructure and construction companies building the facilities. Comfort Systems USA, a mechanical contractor specializing in data center HVAC and electrical systems, saw its shares correct 14% from highs earlier this year, creating what some analysts view as an entry point into a secular growth story. The company trades with a fortress balance sheet and benefits from the multiyear CapEx cycle driven by AI demand.
The third is through IT solutions providers helping enterprises deploy and manage the infrastructure. ePlus reported net sales of $2.4 billion for fiscal 2026, up 22.1% year over year, with record gross billings of $3.8 billion. Chief Executive Officer Mark Marron said the company achieved record backlog and continued gaining market share as customers advanced AI-related initiatives. Fourth-quarter revenue of $576.2 million beat consensus estimates of $569.3 million, though management guided for mid-single-digit growth in fiscal 2027 after a difficult comparison.
The modular edge alternative
Not all data center demand requires hyperscale. AI inference — the process of running trained models to generate outputs — benefits from proximity to users. A distributed network of modular edge facilities can serve applications such as fraud detection, remote patient monitoring and real-time traffic optimization without the resource demands of a 200-megawatt training cluster.
Modular designs use efficient or water-free cooling technologies, require far less land and can be built incrementally as demand grows. That model spreads economic benefits across multiple communities rather than concentrating them in a handful of industrial zones, and it reduces the latency that degrades real-time AI applications.
For investors, the distinction matters. Hyperscale construction benefits a concentrated set of players in Northern Virginia, Texas and the West Coast. The modular edge model opens opportunity across a broader geographic and competitive landscape — and for a wider set of companies supplying the physical layer of the AI internet.
This article is for informational purposes only and does not constitute investment advice.
