The pitch that brings a data center to a community almost always includes a jobs number. The number is usually large, sometimes in the hundreds or thousands, and it is almost always presented in a way that conflates temporary construction employment with permanent operational employment, regional economic multiplier effects with direct local hiring, and industry-wide employment growth with the specific employment that any single facility generates. The communities that approved data center developments in the past five years on the basis of those jobs numbers are now able to compare the promises with the outcomes, and a growing body of research confirms that the permanent employment generated by large data center facilities is systematically smaller than what the incentive applications represent.
The numbers are not hard to find. A Virginia legislative study found that a typical 250,000-square-foot data center may have approximately 50 full-time workers on staff, about half of which are contract workers. A Vantage Data Centers facility outside Reno spanning 1.1 million square feet was projected to create just 73 permanent jobs over the next decade, even as more than 4,000 temporary construction positions, cycled through the site. From 2020 through late 2025, Virginia’s data centers created 1 direct permanent job for every $54 million invested, 168 times more expensive per job than non-data center economic development in the state. In Virginia, the state with the largest and longest-established data center industry in the world, an estimated 7,600 people work in data centers across the entire Northern Virginia cluster. For context, a single Amazon fulfilment warehouse employs 1,000 to 1,500 permanent workers per facility.
The Incentive Structure That Created the Mismatch
The job creation gap is not primarily a consequence of data center operators misleading communities, though misleading claims do occur. It is a consequence of incentive structures that were designed for manufacturing facilities and applied uncritically to data centers whose operational economics are fundamentally different. Manufacturing incentives work when the subsidy lowers the cost of building a facility that then employs a large number of local workers for the life of the facility, generating wages, local spending, and tax revenue that justify the initial incentive cost. Data centers are capital-intensive, labour-light facilities that generate significant property and equipment tax revenue but employ very few people relative to the capital invested.
In Virginia alone, the data center sales tax exemption cost an estimated $1.6 billion in fiscal year 2025, a figure equal to the median income of nearly 8,000 Virginia households, compared to the estimated 7,600 jobs at Virginia data centers. The incentive cost per job ratio is not a marginal policy inefficiency. It is a structural misalignment between the incentive design and the economic reality of the industry it is subsidising.
What the Research Now Shows
Brookings Institution research published in May 2026, examining approximately 770 US data center facilities linked to county-level employment data from 2003 to 2024, found that in hyperscale counties, incentives represent about 2% of total construction investment, meaning subsidies may simply be subsidising investments that would have happened anyway, while in colocation counties, incentives represent a much larger share of total investment at 62%, meaning subsidies may matter more for precisely the facilities that generate the smallest employment benefits. The policy implication is direct. The data center facilities that generate the most local employment benefit, the smaller colocation facilities, are the ones that probably do not need incentives to locate in a given market. The facilities that need incentives the most, hyperscale campuses that would not locate in a specific market without significant tax abatements, generate the least employment per dollar of incentive cost.
The Construction Employment Illusion That Inflates the Job Count
The construction employment figure is the primary mechanism through which data center job creation claims reach numbers that bear no relationship to permanent local employment outcomes. A large data center campus employs thousands of construction workers at peak activity — electricians, ironworkers, pipefitters, and control system technicians who work on the facility for 12 to 24 months and then move to the next project. Many data center employment impact studies count those workers as evidence of significant local job creation even though most are not local workers, many are union craftspeople who travel from project to project across the country, and none will remain employed at the facility once it opens.
Construction jobs at a Cologix data center in Columbus, Ohio lasted on average six and a half weeks, with about 146 workers on site at a time. The electrical workers who build data centers are genuinely skilled, well-paid workers. But from a local economic development perspective, whether they live in the county where the data center is being built determines whether their wages circulate in the local economy or simply pass through it on the way back to where the workers actually live. Data center operators are not responsible for the transient nature of construction employment, but the incentive applications that cite thousands of construction jobs as evidence of local economic impact are presenting those jobs in a way that misrepresents the local benefit they actually deliver.
Why Incentive Structures Are Under Review
A local government that approves $500 million in tax incentives for a data center on the basis of 2,000 construction jobs and 50 permanent jobs is making a fundamentally different economic development decision from one that has those numbers presented honestly and in their correct temporal context.
The distinction matters practically because it determines whether policymakers structure the incentive correctly. Policymakers should design incentives justified by temporary construction activity as construction-period benefits that terminate when the permanent employment phase begins. They should tie incentives justified by permanent employment to permanent headcount verified annually. The data center industry has benefited from incentive structures that fit neither model, using construction employment narratives to justify permanent tax abatements that run for decades regardless of how many people the facility actually employs on an ongoing basis. States and localities reviewing their data center incentive programmes are not acting against technology. They are conducting the basic policy analysis policymakers should have completed before granting the incentives.
Why Communities Are Now Pushing Back on the Incentive Bargain
The evidence gap between promised and actual employment has reached enough communities at enough scale that it is now producing legislative and regulatory responses rather than continued acceptance. Illinois suspended state tax incentives for new data center developments for two years after a study found that hyperscale facilities were receiving hundreds of millions in tax breaks against commitments to create only dozens of permanent jobs. Georgia, Washington, and Connecticut have advanced bills to curtail data center incentive programmes. In New York, one data center promised 125 jobs in exchange for $1.4 billion in incentives, a ratio of $11 million per permanent job that attracted significant legislative scrutiny. Nevada’s $457 million in approved data center incentives came with the promise of approximately 300 total jobs across multiple facilities, a cost-per-job ratio that state legislators are now actively questioning.
The Industry’s Defence of the Incentive Model
The industry’s response to the employment criticism has been to point to construction employment, indirect economic multiplier effects, and the property and equipment tax revenue that data centers generate. Those arguments have genuine merit. Construction employment is real economic activity. Tax revenue funds public services. The economic multipliers of a large capital investment do produce broader regional economic benefits. The problem is that those arguments do not address the specific commitment that the incentive applications made, which was to create a specified number of permanent local jobs in exchange for a specified tax benefit. The communities that approved incentives based on permanent employment promises and received construction booms followed by buildings with 50 employees are not wrong to question whether the bargain they accepted reflects the actual value the data center delivers to their community.
The industry that wants to continue accessing generous tax incentives needs to be honest about what those incentives buy and to structure community benefit agreements that reflect the actual labour economics of its facilities rather than the labour economics of manufacturing facilities that data centers have never resembled. The data is no longer theoretical, it is in the tax records, the employment studies, and the utility bills of every community that has been hosting these facilities for more than three years.
