Chapter 16
Site Selection, Land & Permitting
Chapter 16: Site Selection, Land & Permitting
16.1 Overview
Before a data center can be built, someone has to find the land, secure the power, navigate the permitting, and convince the community to accept it. This chapter covers the constraints that precede physical construction (Chapter 17) and increasingly determine whether a project happens at all. As of 2025, $156 billion in data center projects have been blocked or delayed globally 1, with $41.7 billion blocked in Q1 2026 alone. Twenty-five data center projects were cancelled in 2025, a fourfold increase from six cancellations in 2024. Roughly 200 community opposition groups are active across 24 or more US states, and 14 states have enacted moratoriums or pauses on new data center construction.
The fundamental constraint is “powered land”: sites with sufficient electrical capacity, grid interconnection agreements, appropriate zoning, and community acceptance. Each of these requirements has its own timeline, stakeholders, and failure modes. Grid interconnection alone can take 4-10 years in primary markets. PJM, the largest US regional transmission organization, has 190 GW of pending interconnection requests 2. ERCOT in Texas has 226 GW of large load requests, 77% of which are data centers 3.
This is a constraint that money cannot accelerate. A hyperscaler with $200 billion in annual capex 4 still cannot make a permitting process move faster than the regulatory machinery allows. This makes land and permitting fundamentally different from every other bottleneck in the supply chain, where capital can at least partially substitute for time.
The chapter connects to power generation (Chapter 13) and power distribution (Chapter 14) through the grid interconnection process, to physical construction (Chapter 17) as the prerequisite step, and to data center operators (Chapter 17) as the demand signal.
16.2 Market Sizing
The data center site selection advisory market was valued at approximately $750 million in 2025, projected to reach $3.17 billion by 2035 678. This is a narrow market that belies the scale of capital it gatekeeps: the $156 billion in blocked or delayed projects 1 represents more than 200x the advisory market’s annual revenue. The powered land developers (companies that acquire land, secure power, and obtain permits for sale to hyperscalers) are scaling rapidly, with companies like Stream Data Centers (Apollo-owned, 4+ GW portfolio), PowerHouse Data Centers (4.1 GW, 57 buildings), and Lancium (1.2 GW Stargate site partner) commanding premium valuations for permitted, powered sites.
The legislative environment is complex and shifting. Virginia’s data center tax exemption cost the state $1.6 billion in FY2025 5, sixteen times original projections. Texas exceeded $1 billion. Georgia reached $625 million, up from $10 million in 2020. This escalating fiscal cost is driving legislative pushback. Over 300 data center-related bills were filed across 30+ states in 2026 alone.
16.3 Key Companies
16.3.1 Site Selection Advisory & Commercial Real Estate
| Company | Ticker | Exchange | Approx. Mkt Cap | Role | Key Metric |
|---|---|---|---|---|---|
| CBRE Group | CBRE | NYSE | ~$40.0B | Largest commercial real estate services firm; dedicated DC advisory practice | Global site selection, lease negotiation, market intelligence for hyperscalers |
| JLL (Jones Lang LaSalle) | JLL | NYSE | ~$13.0B | #2 commercial real estate services; data center advisory | Competing with CBRE for hyperscaler mandates |
| Cushman & Wakefield | CWK | NYSE | ~$4.0B | Real estate advisory including DC site selection | Smaller DC practice than CBRE/JLL but growing |
| Colliers International | CIGI | NASDAQ | ~$7.0B | Global real estate services with DC practice | Data center advisory in North America and Europe |
16.3.2 Environmental & Permitting Consultants
| Company | Ticker | Exchange | Approx. Mkt Cap | Role | Key Metric |
|---|---|---|---|---|---|
| AECOM | ACM | NYSE | ~$14.0B | Environmental consulting, permitting, engineering for DC projects | Major DC permitting advisor in Virginia, Texas, Ohio |
| Tetra Tech | TTEK | NASDAQ | ~$10.0B | Environmental and water consulting; permitting services | Growing DC practice; water management critical for cooling-related permitting |
| WSP Global | WSP | TSX | ~$25.0B (CAD) | Engineering and environmental consulting; global presence | European and North American DC permitting |
| Stantec | STN | TSX | ~$12.0B (CAD) | Environmental consulting and engineering | North American DC site assessment and permitting |
16.3.3 Powered Land Developers
| Company | Ticker | Exchange | Approx. Mkt Cap | Role | Key Metric |
|---|---|---|---|---|---|
| Stream Data Centers | Private | Private (Apollo) | Private | Powered land developer and DC operator | 4+ GW portfolio. Apollo-backed. Securing powered sites ahead of construction. |
| Lancium | Private | Private | Private | AI data center campus developer; Stargate site partner | 1.2 GW Abilene, TX site (SoftBank/OpenAI Stargate project). Energy flexibility technology. |
| PowerHouse Data Centers | Private | Private | Private | Powered land developer | 4.1 GW, 57 buildings in development. Securing grid interconnections at scale. |
16.3.4 Grid Operators (Non-investable but Critical)
| Entity | Type | Region | Key Metric |
|---|---|---|---|
| PJM Interconnection | ISO | US Mid-Atlantic/Midwest | 190+ GW interconnection queue. 46 GW approved, 63 GW under review through 2026. 4-10 year wait times in primary markets. |
| ERCOT | ISO | Texas | 226 GW large load requests (77% data centers). 432 GW generation queue. Texas is attractive because ERCOT is a deregulated market with faster interconnection (12-18 months vs 4-10 years in PJM). |
| CAISO | ISO | California | Stringent environmental requirements. Limited data center development due to water scarcity and energy policy. |
| MISO | ISO | US Central | Emerging data center market. Less congested queue than PJM. Iowa, Nebraska gaining DC activity. |
16.4 Bottleneck Analysis
Grid interconnection queue (SEVERE, RPN 112). This is the single bottleneck in the supply chain that capital cannot solve. PJM’s 190+ GW queue means that even with unlimited money, a project in northern Virginia will wait years for grid connection approval. The regulatory process involves transmission impact studies, system upgrade assessments, and coordination with incumbent utilities and generators. No amount of spending accelerates this. The workaround (behind-the-meter generation using natural gas or fuel cells) adds cost and complexity but bypasses the queue for some configurations. ERCOT in Texas is faster (12-18 months) because it is a deregulated market, which explains why Texas is capturing a disproportionate share of new data center development.
Community opposition ($156B blocked/delayed) 1. This is an emerging constraint that did not exist at this scale five years ago. Data centers consume enormous amounts of power and water, generate noise, offer few permanent jobs relative to their footprint, and provide limited tax revenue in jurisdictions that offer tax exemptions. The political dynamics have shifted: Virginia, once the most data-center-friendly state in the US, now faces legislative efforts to cap or roll back exemptions because the $1.6 billion annual tax cost 5 is politically unsustainable. Similar dynamics are playing out in Texas, Georgia, and Ohio.
Water consumption (MODERATE, location-dependent). The IEA estimates global data center water consumption at 560 billion liters per year (2023), projected to reach 1.2 trillion liters by 2030, equivalent to the annual consumption of more than four million US households. A single 100 MW facility can consume 2.5 billion liters annually. Water is already a blocking factor in permitting decisions in water-stressed regions (Phoenix, parts of California, northern Texas). As liquid cooling becomes standard for AI racks (Chapter 15), water consumption per facility increases. Several jurisdictions now require water impact assessments as part of data center permitting, adding both timeline and uncertainty 9.
Powered land scarcity (HIGH). The combination of grid queue delays, permitting timelines, and community opposition means that sites with all three requirements (power, permits, community acceptance) are scarce. This scarcity is creating a secondary market for pre-permitted, powered sites where developers like Stream, PowerHouse, and Lancium acquire land, secure interconnections, and obtain permits, then sell or lease the “powered land” to hyperscalers at a premium.
16.5 Risks
Legislative risk. The tax exemption model that fueled US data center growth is under political pressure. If states significantly curtail exemptions, the economics of new US data center construction could shift, redirecting investment to jurisdictions with more favorable policies (Nordics, Middle East, Southeast Asia).
Technology risk. Improvements in AI inference efficiency (fewer GPUs per unit of useful work) could reduce the pace of data center construction, easing permitting pressure. The DeepSeek efficiency thesis suggests this is plausible.
Behind-the-meter workarounds. If hyperscalers increasingly generate their own power on-site (nuclear microreactors, fuel cells, dedicated gas turbines), the grid interconnection bottleneck becomes less binding. This is already happening at some scale with Bloom Energy fuel cell deployments and the Constellation/Microsoft TMI restart.
International diversification. The Nordics (Iceland, Norway, Sweden, Finland), the Middle East (Saudi Arabia, UAE), and parts of Asia (Singapore, Japan) offer alternative permitting environments with different tradeoffs. This diversification reduces US-specific permitting risk but introduces new geopolitical and operational risks.