The most repeated criticism of data centers, after water and power, is that they don't create jobs. A hyperscale campus the size of a regional shopping mall, the argument goes, ends up employing 50 people. The economic-impact-per-acre is unfavorable. The community gets the noise and the trucks and the substations, and Amazon gets a building staffed by a dozen technicians.

The argument has been hard to rebut because the industry has been refuting the wrong number. The 50-person ops headcount is roughly correct for steady-state operations of a single building. It is also a small fraction of the workforce attached to the facility. The actual employment story is in the construction phase, in the long-tail vendor ecosystem, and in the trades that turn over crews continuously across multiple campuses for the entire life of the buildout. Those workers are mostly invisible in the public conversation because nobody bothers to count them.

What a Construction Phase Actually Looks Like

A 100-megawatt data center building, the kind currently being built across the West Valley and East Valley, takes 18 to 30 months from groundbreaking to commissioning. During peak construction, the site has between 1,200 and 2,000 workers on it daily. A single hyperscale campus typically includes four to eight buildings, often built in overlapping phases. The campus-level headcount during multi-building construction can sit at 3,000 to 5,000 workers for several years running.

These are not low-wage jobs. The trades on a data center site include electricians, mechanical fitters, sheet metal workers, ironworkers, controls technicians, fire protection specialists, low-voltage cabling crews, civil contractors, equipment operators, and project management staff. Arizona electricians earn a median around $59,000 a year, with the top tier above $81,000, and overtime on data center work pushes skilled-trade compensation higher still. Most data center work pays prevailing wage by contract requirement, even on private projects, because the developers and general contractors compete for skilled crews.

The construction headcount is not a one-time event. Metro Phoenix ranks #2 in North America for planned data center development, with roughly 1,300 megawatts under construction and another 4,154 megawatts planned as of mid-2025, and individual projects like the Pinal County and Buckeye campuses running into the tens of billions of dollars each. At the construction labor intensities the industry uses, that pipeline keeps thousands of trade workers attached to data center construction in the metro at any given moment, for years to come.

That is not 50 jobs.

The Long-Tail Vendor Ecosystem

The construction crews are the biggest visible category. They are not the only one. Once a campus is operational, the day-to-day vendor footprint is what most observers miss.

Generators need monthly testing, oil changes, and overhauls. The companies that do that work employ field service technicians who drive between sites every week. A campus with 60 generators (a typical hyperscale ratio) supports several full-time equivalents of generator service work just from one customer.

Switchgear and electrical infrastructure require periodic inspection, infrared thermal scanning, breaker testing, and recalibration. Specialized contractors handle that work and rotate through multiple sites monthly.

Cooling systems need water treatment, refrigerant management, vibration analysis, and bearing service. The mechanical service vendors maintain crews dedicated to data center work because the volume in the metro now justifies it.

Fire protection systems require quarterly inspections, annual valve testing, and periodic full-system flow tests. The pre-action and gaseous suppression systems used in data centers require specialized technicians.

Security systems, controls integration, structured cabling, building management systems, leak detection, environmental monitoring. Each of these is a separate vendor with a separate field crew.

The on-site headcount of the data center operator is perhaps 50 people. The off-site, vendor-employed headcount supporting the same building is roughly five to ten times that, distributed across companies that serve dozens of sites. None of those workers show up in the data center's employment number, but they exist because the data center exists.

The Apprentice and Pre-Apprentice Pipeline

The construction labor demand is large enough that it has restructured the local trades training ecosystem. Maricopa Community Colleges, Mesa Community College, Estrella Mountain Community College, GateWay Community College, and the East Valley Institute of Technology all run programs that feed into data center work directly or indirectly. Apprenticeship programs run through the Arizona Builders Alliance, the IBEW Local 640, the SMACNA Phoenix chapter, the Arizona Pipe Trades, and the operating engineers locals are all enrolled at or near capacity.

The math on a trades career is favorable. An apprentice electrician in a registered program earns wages plus benefits during a 4 to 5 year apprenticeship, completes the program with no debt, and emerges as a journeyman earning a solid middle-class wage, with overtime on data center work pushing total compensation considerably higher. Compare to a four-year university path with debt service and a starting salary frequently below the journeyman number. The trades are not a fallback option in this market. They are the better option for many high school graduates, and the data center buildout is the largest single demand source pulling them in.

Microsoft's Datacenter Academy, Aligned's career pathways, AWS's Spark series, the AZDCA Foundation's scholarship programs, and the AFCOM intern program are all components of an emerging workforce pipeline that didn't exist in 2018. The companies building data centers know they cannot hire faster than the local talent pool grows, so they are funding the pool's expansion. That is also a real economic outcome that doesn't show up in the headcount number for any single facility.

The Operations Workforce Itself

The 50-person figure for steady-state operations is roughly correct, but the composition is different from what the criticism implies.

A typical hyperscale facility operations team includes critical facilities engineers (mechanical and electrical), facilities technicians, security staff, IT operations technicians, network engineers (when present on-site), site reliability roles, and management. Average compensation on these teams runs from $70,000 a year for entry-level technicians to over $200,000 a year for senior critical facilities engineers and site managers. Most positions require either a trade certification, a two-year associate degree in a relevant technical field, or a four-year engineering degree. The training pathway is well-defined and accessible. People without bachelor's degrees can and do enter the field at $70,000 to $90,000 a year and grow from there.

A campus with eight buildings has 8 times that headcount. A campus with 20 buildings has 20 times. The bigger the campus, the larger the operations team, and the more concentrated the on-site employment becomes. Mesa, Goodyear, and El Mirage are heading toward the operational scales where the steady-state employment per campus crosses into the hundreds.

The Multiplier That Economists Already Counted

Economic impact analyses commissioned by utilities and municipalities routinely run multiplier math on data center construction and operations. Standard regional models put each direct job as supporting additional indirect jobs in the surrounding economy, with construction multipliers running higher because the construction workforce is larger and more wage-intensive.

Add up the construction workforce, the on-site operations teams, the long-tail vendors, and the indirect employment, and the data center attributable footprint across the metro runs well into the tens of thousands of jobs. The exact number depends on the model and the assumptions, which vary by source. The point is the order of magnitude.

The order of magnitude is not 50.

What's Hard to See

The reason this story is hard to tell is that the workforce is distributed. There is no single payroll list to point at. Workers move from campus to campus. Contractors come and go. The trades crews on any given building represent 50 different employers, all of whom are also working on other projects, hospitals and chip fabs and freeway work, in addition to the data center.

The data center critic who counts the operations team is asking the wrong question. The right question is what the regional industrial economy looks like with the buildout and what it would look like without it. Take the construction crews off the books, take the vendor ecosystem off the books, take the apprenticeship pipeline funding off the books, take the operations teams off the books, and what remains is a metro with significantly less skilled trade employment, fewer career pathways for non-college-bound workers, and a thinner industrial base.

That is the actual workforce story. It is more complicated than a single number, which is why nobody has been telling it. But it is also more accurate, and it is the version that holds up under scrutiny.

Sources