Arizona Attracts Billions in Data Center Investment as Scrutiny Rises

Key Takeaways

• Arizona’s data center growth is increasing pressure on electricity, water, and local infrastructure.
• Performance standards and customer-funded grid upgrades could address those pressures without a blanket moratorium.
• PUE, WUE, independent audits, and public reporting can help distinguish efficient projects from resource-intensive developments.

Arizona has attracted billions of dollars in data center investment, bringing cloud infrastructure, artificial intelligence capacity, construction spending, and long-term technology operations into the state. The expansion also presents a practical policy challenge: how can Arizona capture those benefits without shifting excessive power, water, and infrastructure costs to residents and other businesses?

A blanket moratorium would offer a simple answer, but perhaps too simple. It could stop inefficient projects, yet it could also block facilities using closed-loop cooling, lower-water designs, cleaner electricity, and improved energy management. A more targeted approach would connect permits and utility approvals to measurable performance, transparent demand forecasts, and enforceable infrastructure commitments.

The electricity issue is becoming harder to separate from economic development. According to the International Energy Agency, global data center electricity consumption rose 17% in 2025 and is expected to double by 2030. More than 85% of new capacity over the next decade is expected to be concentrated in the United States, China, and the EU, often around clusters where grids are already constrained.

For Arizona, that outlook supports project-specific grid-impact studies rather than optimistic assumptions about future supply. Developers seeking large interconnections should provide credible load forecasts, explain when facilities will reach full capacity, and identify firm power arrangements. Utilities and regulators could also structure tariffs so that data center customers fund the transmission, substations, and generation capacity primarily built to serve them. Otherwise, households and smaller commercial users may carry part of the bill.

A data center’s announced capacity is not the same as its real-time demand. Loads can ramp over several years, fluctuate with computing activity, or change as customers install denser AI hardware. That makes periodic reporting useful. It also gives utilities a better basis for planning than a one-time estimate submitted early in development.

Water creates a different, distinctly Arizona concern. A 1-MW facility can consume more than 25 million liters annually, although consumption varies substantially by cooling technology, climate, workload, and operating practices. Closed-loop, air, hybrid, and immersion cooling can reduce direct water use. Permitting standards therefore should focus on outcomes, including total withdrawals, potable-water dependence, seasonal consumption, and watershed conditions, rather than prescribing one cooling system.

Could stricter standards actually encourage investment? In many cases, yes. Clear thresholds provide developers with predictable requirements before they purchase land or enter an interconnection queue. They also reward projects that invest in efficient designs instead of treating every proposed facility as environmentally equivalent.

Research summarized by Cornell found that smart siting, grid decarbonization, and operational efficiency could reduce projected AI-related water impacts by up to 86% by 2030. That finding strengthens the case for siting and performance rules. Location matters, as do the source of electricity and the cooling method selected for local conditions.

Common measurements already exist. Power Usage Effectiveness, or PUE, compares a facility’s total electricity consumption with the power delivered to computing equipment. Water Usage Effectiveness, or WUE, measures water consumption relative to IT energy. ISO 50001 can provide an energy-management structure around those operational metrics. Arizona could use these tools while requiring third-party verification and public annual reporting.

There are signs that lower-water models are commercially viable. QTS reported a 2025 portfolio PUE of 1.38 and WUE of 0.44, while saying its newer closed-loop facilities need no ongoing cooling water. Those claims offer a useful benchmark, but project approvals should rely on independently verified design and operating data. Digital Realty and Microsoft Azure also illustrate the range of large operators whose design choices can shape regional resource demand.

That said, metrics alone will not repair roads, expand emergency services, or resolve transmission congestion. Local agreements can address those costs through impact fees, construction milestones, water contingencies, and financial assurances if a project is delayed or abandoned. Arizona’s opportunity is to make infrastructure responsibility part of the investment model. Guardrails built around measurable performance could preserve digital growth while giving communities firmer control over its costs.