Arizona’s AI Infrastructure Reality Check

Arizona has the land, investment and technology ecosystem. Now comes the harder question: Can the infrastructure keep up?

From the outside, Arizona looks like a clear winner in the AI infrastructure race.

Meta’s Mesa data center is online and expected to represent more than $1 billion in investment when fully built. Google broke ground on a $600 million air-cooled data center nearby. TSMC’s announced Arizona investment has climbed to $265 billion, creating an advanced semiconductor cluster built partly around demand for AI and high-performance computing. Meta Arizona Commerce Authority TSMC

Add available land, a relatively low natural-disaster risk, established fiber infrastructure, major universities and proximity to California, and the Arizona pitch writes itself.

But announcements are the easy part.

The real work begins when hundreds of megawatts, new substations, transmission lines, fiber routes, cooling systems, water strategies, construction workers and public approvals must arrive at the same place at roughly the same time.

Here’s the reality check:

Arizona is no longer competing for attention.

It is competing for capacity.

The Signal

Arizona’s AI infrastructure market is entering a new phase.

For more than a decade, the state focused on attracting data centers and advanced manufacturers. Tax incentives, available land and relatively affordable electricity helped build a fast-growing technology market around Phoenix and Mesa.

Now utilities, regulators and policymakers are asking a different set of questions.

How many projects can the grid realistically support? Who pays for the infrastructure required to serve them? How should Arizona evaluate water use? How much economic value does each project create? And what protections are needed for existing customers and communities?

Those questions moved directly into public policy in 2026.

Governor Katie Hobbs signed a three-year moratorium on new data-center tax incentives beginning July 1, 2026. The move does not prohibit new data centers, but it pauses new applications for the tax relief that helped attract them. Office of the Arizona Governor

That is not necessarily a sign that Arizona is turning against AI infrastructure.

It is a sign that the market is maturing.

The conversation is shifting from attraction to accountability.

The Number Worth Knowing

19 gigawatts

In a November 2025 planning presentation, Arizona Public Service reported nearly 19 gigawatts of uncommitted large-load requests in its queue.

For perspective, APS listed its 2025 peak demand at approximately 8.7 gigawatts.

Serving every request would more than double the utility’s current peak demand. APS also acknowledged that the queue contains duplicate requests and that not every proposed project will materialize. A hyperscaler may approach the utility directly while several data-center developers submit separate requests for the same eventual customer. APS Resource Planning Presentation

That distinction matters.

A queue is not a construction schedule.

It is a signal of interest, not proof that every project has financing, customers, equipment, permits or a realistic path to power.

Arizona has significant AI infrastructure demand. The harder job is separating credible projects from speculative capacity requests.

Arizona’s Advantages Are Real

The reality check should not erase Arizona’s strengths.

The state has developed one of the most significant advanced-technology ecosystems in the country. TSMC, Intel, Amkor and a growing network of suppliers are creating semiconductor manufacturing and packaging capacity directly connected to the future of AI computing.

Phoenix and Mesa already have data-center clusters, established carrier infrastructure and utilities experienced in serving large technology customers.

Arizona also offers something many traditional technology markets do not: room to expand.

Large sites near major transportation corridors can support data centers, semiconductor facilities, energy projects, manufacturing operations and supplier campuses. That creates the potential for integrated technology regions rather than isolated facilities.

The issue is not whether Arizona has opportunity.

The issue is whether the eight physical systems beneath that opportunity are ready to scale together.

The Eight-Pillar Reality Check

Power: Strong Foundation, Tightening Capacity

Power is Arizona’s most important advantage and its most immediate constraint.

APS says an average data center in its territory can require as much electricity as approximately 64,000 homes. The utility has proposed increasing rates for extra-high-load customers by more than 45%, along with long-term contracts and customer contributions intended to prevent infrastructure costs from shifting to households and smaller businesses. APS

SRP reports that large data centers represented 5.1% of its summer 2025 peak demand but are projected to become its fastest-growing customer segment. Some individual facilities in its territory can reach approximately 200 megawatts. SRP

Arizona has generation assets, utility expertise and access to solar, nuclear, natural gas and battery storage.

What it does not have is unlimited power available everywhere, immediately.

A project may find land in months. Building the generation, transmission and substations needed to serve it can take years.

Connectivity: Strong in the Core, Uneven Beyond It

Phoenix and Mesa benefit from established fiber routes, carrier networks and proximity to West Coast markets.

That connectivity has helped create economic gravity around existing technology corridors. Once data centers, carriers and cloud infrastructure begin clustering in one area, additional projects can connect more easily.

The challenge appears when development moves beyond the established core.

Pinal County and Southern Arizona may offer attractive land, transportation access and room for new campuses, but regional connectivity must be evaluated route by route. A highway corridor does not automatically become a fiber corridor.

The question is not whether fiber appears on a map.

It is whether a project can obtain scalable capacity, multiple carriers, physically diverse routes and reliable access to major cloud and interconnection markets.

Land: Abundant Does Not Mean AI-Ready

Arizona has land.

Powered land is a different asset.

A large desert parcel may look ideal until developers discover that the nearest transmission line lacks available capacity, a substation upgrade will take several years, water access is uncertain or the site has only one usable fiber route.

This is where traditional economic-development marketing can fall short.

The Intelligence Economy does not need the cheapest parcel. It needs land where power, connectivity, water, zoning and infrastructure timelines have already been studied.

Arizona’s next competitive advantage may come from identifying and preparing these locations before a hyperscaler arrives with a deadline.

Water: The Credibility Test

Water is not automatically a deal-breaker for Arizona data centers.

It is a design and transparency test.

Google selected air cooling for its Mesa facility. Meta has emphasized water efficiency and restoration programs around its campus. Other facilities may use closed-loop systems, reclaimed water or cooling designs with very different consumption profiles.

That means “How much water do data centers use?” does not have one universal answer.

The more useful questions are:

• What cooling system will the facility use?

• What is its expected annual and peak-day consumption?

• Will it use potable, reclaimed or groundwater supplies?

• How will that use change as the campus expands?

• What happens during extreme heat?

These questions matter because the broader regional water picture remains constrained. The Arizona Department of Water Resources projects 4.86 million acre-feet of unmet groundwater demand in the Phoenix Active Management Area over 100 years under the conditions modeled. The agency attributes those constraints to decades of groundwater overdraft and continued reliance on groundwater supplies. Arizona Department of Water Resources

Data centers did not create Arizona’s water challenge.

But every major new project must operate within it.

Workforce: Growing, but Competing With Itself

Arizona is building a serious advanced-technology workforce.

Universities, community colleges, apprenticeship programs and the Future48 workforce accelerators are creating new pathways into semiconductor manufacturing, advanced manufacturing, electrical work and infrastructure construction.

TSMC alone reports more than 3,500 Arizona employees, and the state says it has attracted more than 70 semiconductor expansions since 2020. Arizona Commerce Authority

That growth creates momentum, but it also creates competition.

Data centers, semiconductor fabs, utilities, construction firms and manufacturers need many of the same electricians, engineers, technicians, pipefitters and skilled tradespeople.

The workforce question is no longer simply whether Arizona can create jobs.

It is whether the state can train enough people to deliver the projects already announced.

Policy: The Rules Are Being Rewritten

Arizona’s tax-incentive moratorium marks a meaningful change.

The state has moved beyond asking, “How do we attract more data centers?” and is beginning to ask, “What should Arizona receive in return?”

That may lead to new expectations around infrastructure payments, water reporting, efficiency, local tax value, workforce investment and community benefits.

Policy uncertainty can slow projects. Clear standards can do the opposite.

Developers can price requirements into a project when the rules are known. What creates risk is a different process in every city, changing expectations halfway through development or infrastructure commitments that were never clearly assigned.

Arizona does not need a race between communities to offer the largest incentive.

It needs a credible statewide framework for responsible AI infrastructure growth.

Capital: Available, but Becoming More Selective

There is no shortage of capital interested in Arizona’s technology story.

The state is attracting hyperscalers, semiconductor manufacturers, infrastructure funds, developers and global suppliers. Capital is available for projects with credible customers and a realistic path to operation.

But money alone cannot manufacture utility capacity.

Investors are increasingly examining energization dates, transmission plans, water access, equipment lead times, fiber diversity and local approvals before treating a site as viable.

The most valuable Arizona project will not necessarily be the one with the largest announcement.

It will be the one with the fewest unresolved infrastructure questions.

Sustainability: Sunshine Is Not a 24/7 Power Strategy

Arizona’s solar resources create an obvious advantage.

AI facilities, however, operate around the clock.

Meeting that demand reliably will require a mix of solar, storage, nuclear, natural gas, grid modernization and potentially new behind-the-meter generation. The exact mix will vary, but the underlying challenge remains the same: annual renewable-energy matching is not the same as delivering reliable hourly power.

Arizona can become a laboratory for energy-resilient AI infrastructure.

But doing so will require sustainability strategies built around actual grid conditions rather than marketing claims.

The Pinal County Opportunity

The next phase of Arizona’s infrastructure growth may not remain entirely inside Phoenix and Mesa.

Pinal County sits between the state’s largest technology market and Tucson’s research, university and aerospace ecosystem. Casa Grande, Eloy and Florence offer industrial land, transportation access and the potential to support new energy, manufacturing and data infrastructure.

This creates the foundation for what IOI calls the Phoenix–Casa Grande–Tucson Intelligence Corridor™.

But a corridor is not created by drawing a line on a map.

It becomes real when utilities, cities, water providers, fiber carriers, universities, landowners and capital partners coordinate around shared infrastructure.

Pinal County’s opportunity is to prepare before demand arrives at full scale.

That means identifying powered-land opportunities, mapping fiber routes, evaluating water strategies, aligning zoning, building workforce programs and creating a regional infrastructure narrative supported by evidence.

The region should not market itself as the place with empty land.

It should position itself as the place where the next generation of infrastructure can be intentionally built.

Utility to Watch: SRP

SRP sits at the center of Arizona’s AI infrastructure story because it operates across both power and water.

The utility has introduced a Large Customer Integration Process to evaluate the infrastructure required for major new loads. Its updated pricing structure also requires new customers forecasting at least 20 megawatts to pay based on the greater of their actual demand or 80% of their forecast.

That provision matters because overly ambitious requests can cause utilities to reserve capacity and plan infrastructure for projects that never reach their expected size.

SRP is also constructing physical infrastructure around the state’s technology clusters. Project Huckleberry includes a 230-kilovolt transmission line and substation serving Meta’s Mesa campus, while Project Red Hawk connects another planned data-center project to the regional transmission system. SRP Grid Projects

SRP is not simply responding to data-center growth.

It is helping determine where that growth can physically occur.

What Arizona Leaders Should Start Asking

Arizona does not need fewer technology ambitions. It needs better infrastructure questions.

Leaders Should Ask:

• How many requested megawatts are connected to committed projects?

• How much power is available today, in three years and in five years?

• Who will pay for new generation, transmission and substations?

• Are proposed fiber routes scalable and genuinely diverse?

• What cooling technology and water source will each facility use?

• How many permanent jobs will the project create relative to its infrastructure requirements?

• What local tax revenue and community benefits will remain after incentives?

• Can workforce programs support the announced construction pipeline?

• What happens if a project reserves capacity but never fully develops?

• How will communities measure performance after incentives and approvals are granted?

These are not anti-growth questions.

They are what responsible growth looks like.

The Hot Take

Arizona does not have an AI demand problem.

It has an infrastructure coordination problem.

Companies want to build. Developers want to acquire land. Utilities want to serve credible customers. Communities want investment and high-paying jobs.

The friction appears when each participant is working from a different timeline, dataset or definition of readiness.

Arizona’s next competitive advantage will not come from another incentive.

It will come from coordination.

The Golden Nugget

The market no longer rewards a site simply for having land. It rewards a site for having coordinated capacity.

That means available power, accessible fiber, a defensible water strategy, workable zoning, infrastructure financing and a community prepared to support the project.

Anything less is potential, not readiness.

The IOI Take

Arizona is well positioned to lead the physical buildout of the Intelligence Economy.

It has a globally significant semiconductor cluster, an established data-center market, experienced utilities, major research institutions and enough investment momentum to remain on every serious infrastructure watchlist.

But Arizona’s second act must be different from its first.

The first act was about attraction.

The second must be about execution, accountability and regional coordination.

Success should not be measured by the number of projects announced or the total megawatts requested. It should be measured by how many credible projects can operate without undermining grid reliability, water security, affordability or community trust.

Arizona has already proven that it can attract the Intelligence Economy.

Now it must prove that it can build it responsibly.

One Last Thought

The biggest risk to Arizona is not that AI infrastructure goes somewhere else.

It is that the state attempts to move faster than the systems beneath it.

Power takes time. Water requires trust. Fiber follows routes. Workforces must be trained. Communities expect answers.

Arizona can still become one of the most important infrastructure markets of the Intelligence Economy.

But the winner will not be the region with the biggest announcement.

It will be the region with the fewest hidden constraints.

Workforce

Policy

Capital

Sustainability