When a developer announces a new data center, the jobs number in the press release is usually the biggest, most optimistic version of the truth. Residents in Arizona have heard these figures often enough to be skeptical, and that skepticism is reasonable. So let's be straight about what the research actually shows.

The honest answer has three parts, and they're each a different kind of number.

Part One: Construction Jobs Are Real, Substantial, and Temporary

The first burst of employment from any data center is in construction. It's also the largest. A 100-megawatt hyperscale campus typically employs 800 to 1,200 workers at peak activity over an 18-to-36-month build period, according to CBRE's 2024 North America Data Center Trends report.

These jobs include electricians, pipefitters, ironworkers, concrete crews, HVAC technicians, and commissioning agents. The trades involved are skilled and well-paid. In a market like metro Phoenix, where 400-plus megawatts of new capacity was under construction in recent years, that translates to an estimated 2,800 or more construction jobs across multiple active projects simultaneously.

The Phoenix metro data center buildout has functioned as a sustained construction employment program for skilled tradespeople. Because new projects start continuously, electricians and mechanical contractors can move from project to project without leaving the region. In that sense, what looks like a series of temporary jobs adds up to a semi-permanent career track for the trades in this market.

But be honest with yourself: when a project wraps, those workers move on. Construction jobs are real economic value delivered during the build. They are not permanent positions at the facility.

Part Two: Permanent On-Site Headcount Is Modest

This is the number that surprises people the most.

A fully built 100-megawatt hyperscale campus, the kind operated by Amazon, Google, or Microsoft, typically has 100 to 200 permanent on-site staff. A 50-megawatt campus might have 50 to 80. The reason is automation: modern hyperscale facilities are designed to run with minimal human intervention. Uptime Institute data puts hyperscale staffing at roughly 1 to 2 employees per megawatt of capacity.

Colocation facilities, the kind that lease rack space to multiple tenants, staff at higher ratios: 4 to 8 employees per megawatt is typical, because they support many different customers and their equipment. A 20-megawatt colocation data center might have 80 to 150 permanent staff.

These are real, full-time jobs. Permanent data center technicians earn $55,000 to $95,000 depending on experience and location, based on cross-referenced figures from BLS, Glassdoor, and Indeed. Senior engineers and commissioning specialists earn $100,000 to $160,000. Those wages sit at or well above Arizona median household income. The career pathways are accessible too: Maricopa Community Colleges has a partnership with AWS that trains people for entry-level roles starting at $45,000 to $65,000, with a defined track upward.

What the headcount is not is a factory. A data center with 150 permanent employees serving hundreds of megawatts of compute is not a wrong number or a broken promise. It's what a highly automated facility looks like. The economic case for data centers rests on different things than raw headcount, which brings us to the third part.

Part Three: Indirect Jobs and Tax Base Are Where the Real Story Lives

A Brookings Institution study published in May 2026 is the most rigorous analysis to date on what data centers actually do to local labor markets. Researchers Dany Bahar and Greg Wright tracked 93 U.S. counties that received their first large data center between 2008 and 2024, comparing them against roughly 3,000 control counties using a synthetic control method designed to avoid the inflated claims common in industry-sponsored reports.

Their findings: counties that receive a large data center see total private employment rise by 4% to 5% over five to six years. Construction employment jumps 11%. Information sector employment (IT services, telecom, software) grows by 22% in hyperscale counties specifically. At a typical county with about 98,000 workers, that translates to 2,000 to 4,000 additional jobs after six years.

The Brookings researchers also found that naive estimates, the kind that don't account for the fact that high-growth counties were already growing before a data center arrived, overstate the employment effect by a factor of three. So the inflated numbers in some press releases are not just hype; they're a methodological error. The real effect is meaningful but smaller.

For Arizona specifically, a 2023 analysis by PwC for the Data Center Coalition estimated that the state's data center industry supported approximately 81,000 jobs statewide when indirect and induced effects were included, an industry-funded figure republished by the Arizona Technology Council. The same PwC analysis cited more than $863 million in state and local tax revenue from data centers in 2023, a figure republished by the Arizona Technology Council in June 2025.

Tax revenue is where the economic case for data centers is hardest to dispute. These facilities carry enormous assessed values, and Arizona's transaction privilege tax structure for data center equipment means ongoing equipment investment flows into the state tax base. Counties with major data center campuses collect property and equipment taxes that fund schools, roads, and public services, without the facility generating heavy demand on those services the way a large residential development would. The equipment depreciates and gets replaced on regular cycles, keeping the tax flow active. See our deeper dive at The Tax Base Nobody Talks About.

The Cluster Effect Is Real

One critical nuance from the Brookings study: a single data center produces modest employment effects. The IT-sector growth, the kind that creates an ecosystem of fiber installers, network operations centers, managed service providers, and local IT contractors, only really develops at scale. Counties with four or more facilities see a 23% increase in information-sector employment. A single facility sees modest total employment gains but no statistically significant information-sector growth.

This matters for how Arizona thinks about data center policy. The economic case for the third or fourth campus in a cluster is meaningfully stronger than the case for the first. And Phoenix, which already has one of the largest data center concentrations in the U.S., is positioned to capture those compounding returns in a way that a state hosting its first facility is not.

What to Do With Inflated Numbers

When you see a headline claiming a proposed data center will create 5,000 jobs, read it carefully. Ask: is that construction jobs over the build period, or permanent operations jobs? Is it direct on-site headcount, or does it include indirect and induced effects across the broader economy? Is it job-years (workers times years) or permanent positions? Is the study independent, or industry-commissioned?

None of those questions invalidate the economic case. They just describe what kind of benefit you're actually getting. A data center with 150 permanent technicians, 1,000 construction jobs over two years, $25 million in annual property taxes, and a supply chain pulling in local vendors is a genuine economic contributor. It just looks different from a semiconductor fab or a distribution warehouse. The argument on its own terms is strong enough that it doesn't need to be inflated.

For more on what's actually inside a data center campus and who runs it, see Who Builds and Runs Them. For what one of these facilities actually costs to build, see What a Hyperscale Campus Costs.

Want to ask questions like these directly to people who work in this industry? Join us at the next community forum. Register at azdatacenter.org/sept-17-register.html.

Sources

Related Articles