Research, analysis, and perspectives on the built world.
Data center permitting spans multiple simultaneous approval tracks — special use permits, environmental reviews, noise ordinances, and utility coordination — each with its own timeline and interdependencies. AI monitoring systems track permit status across jurisdictions, parse regulatory changes, and generate probability-weighted schedule models for portfolio visibility. This post explains where delays actually come from and what AI can and cannot do about them.
April 12, 2026
Wholesale and colocation data center facilities look similar from the outside but represent fundamentally different development businesses. This post breaks down how lease structure, tenant credit, power density, capex, and exit pricing differ between the two product types, and provides a decision framework for developers evaluating which to build at a given site.
April 12, 2026 ·5 min read
Multimodal AI — models that process images, maps, and documents alongside text — extends AI capabilities into the visual layer of real estate development: site plans, surveys, construction photos, and zoning maps. This post covers the highest-value use cases in development workflows today, accuracy considerations for production deployment, and the implementation pattern that works.
April 12, 2026 ·5 min read
Full due diligence on a data center site spans ten domains, from power availability and grid study risk to title, environmental, and water feasibility. This checklist covers the go/no-go criteria for each, what AI automates, and what still requires experienced human judgment. Designed for institutional developers evaluating multiple sites simultaneously.
April 11, 2026 ·5 min read
Data center site screening covers 30 or more variables across power, land, zoning, fiber, and water before a site earns a full diligence budget. This guide breaks down the go/no-go thresholds for each domain and explains how AI-assisted screening lets development teams evaluate 10x more candidates in the same timeframe. The developer who screens more sites wins more deals.
April 11, 2026 ·5 min read
GPU rack densities from AI compute workloads have outpaced air cooling's limits, forcing data center developers to choose between direct liquid cooling, immersion, and evaporative systems before a building is designed. This post covers the capabilities and site constraints of each approach, explains the water availability issue reshaping market selection, and shows where AI-assisted feasibility modeling changes the economics of the decision.
April 11, 2026 ·5 min read
An analysis of data center supply and demand dynamics across primary and secondary U.S. markets in 2026, covering vacancy rates, hyperscale absorption, power-constrained market dynamics, construction costs, and the tier-2 markets attracting the most institutional capital.
April 10, 2026 ·4 min read
A technical breakdown of the five components of power analysis for data center site selection, covering utility capacity, transmission constraints, interconnection queue timelines, rate structures, and PPA viability, with a guide to public data sources and where AI accelerates the process.
April 10, 2026 ·4 min read
A step-by-step workflow guide to AI-driven data center site sourcing across six stages, from criteria definition through scored shortlist delivery, covering what manual search misses including off-market parcels and early power screening.
April 10, 2026 ·4 min read
Reasoning-capable AI models like OpenAI o3 and Claude with extended thinking are changing complex analytical tasks in real estate development. This post explains what makes reasoning models different from standard LLMs, where they have traction in feasibility analysis and capital stack modeling, and how development teams should deploy them selectively alongside faster models.
April 9, 2026 ·4 min read