ICE model gauges U.S. data center permit disruption risk
Rising scrutiny over power, water and zoning is increasingly complicating U.S. data center construction as developers commit larger sums to AI-linked capacity. ICE says its Construction Disruption Model estimates the probability that a data center building permit is halted or remains incomplete within 24 months, with 77% accuracy in historical back-testing.
Highlights
- ICE's Construction Disruption Model estimates site-specific data center permit disruption risk using 6,914 historical building permits across 41 states since 1982.
- Data Center Tracker records 1,486 legislative actions against U.S. data center projects as of October 2026, impacting $2.6 trillion in announced capital expenditure and 253 gigawatts of load.
- Only 5.3% of opposition involves AI, with most objections targeting zoning, community impact, grid capacity and water use, underscoring the importance of localized disruption analysis.
Model targets permitting and delay exposure
As reported by Intercontinental Exchange, citing ICE, the Construction Disruption Model is part of the company’s Data Center Site Analytics tool and is designed to estimate site-specific disruption risk before developers, investors and lenders commit capital. The model draws on historical project outcomes, local affordability measures, socioeconomic conditions and market saturation data to quantify the likelihood that a permitted project is stopped or left incomplete within two years of filing.ICE says the model is trained on 6,914 building permits for data center projects across 41 states since 1982. It groups inputs into permit and project history, site-specific characteristics and broader market conditions, aiming to give municipalities and private-sector participants a common framework for pricing delay risk and comparing commitments.
The company positions the tool as a way to improve negotiations around community benefit packages, which can vary widely by location. ICE says delay costs often exceed incentive packages, while talks between host communities and developers may proceed without location-specific estimates of the financial impact of permitting setbacks.
Community opposition shapes project economics
Affordability patterns are a major factor in local objections, according to the analysis. Recorded opposition tends to cluster in lower-income, lower-density census tracts where energy and water costs account for a larger share of household income, suggesting resistance is often tied to resource pressure rather than broad ideological concerns.ICE says data centers remain one of the largest components of U.S. private-sector investment, with technology investment, including data centers, accounting for about one-third of year-on-year U.S. GDP growth in the second quarter of 2026. At the same time, the Data Center Tracker records 1,486 legislative actions against U.S. data center projects as of October 2026, covering $2.6 trillion in announced capital expenditure and 253 gigawatts of announced load.
Only 79 of those actions, or 5.3%, involve general objections to AI, ICE says, while most are focused on zoning, community impact, grid capacity and water use. The company argues that better site-specific data can help developers, investors and local governments assess whether projected benefits outweigh likely disruption costs as global AI-related investment rises.
As highlighted earlier, JBIC’s push for project selection under the $550 billion U.S.-Japan investment plan underscored how large-scale infrastructure commitments are being structured.