A $10 billion data center campus in Lebanon, Indiana, will employ about 300 people once it is operational. Meta’s facility, the company announced in February, will represent more than $10 billion in regional investment. At peak construction, the project is expected to support more than 4,000 construction jobs. Once operational, the campus will support about 300 jobs.
That works out to one permanent position for every $33 million invested. Compare that to TSMC’s semiconductor complex in Phoenix, Arizona: TSMC’s total investment of $165 billion in the U.S. is expected to directly create 12,000 jobs once all sites are completed and fully operational, according to the company’s president, Rose Castanares, in an interview cited by TrendForce. That is one job for every $14 million, still capital-heavy but more than twice the labor density of Meta’s data center.
The gap gets wider. Virginia data centers generate just one permanent job for every $13 million invested, according to a January 2026 analysis by Food & Water Watch, based on Virginia Economic Development Partnership data dating back to 1990. In contrast, it costs $137,000 to create one job outside of the data center sector, about 100 times less investment.
The disparity sits at the center of an accelerating national debate over what communities should expect when a hyperscale facility lands in their county.
What the facility-level data shows
The most automated hyperscale campuses can run on skeleton crews. Facilities exceeding 100 megawatts can operate with as few as 20 to 30 permanent staff per 100 MW, according to a November 2025 data center workforce forecast from the Hamm Institute. Industry benchmarks put permanent staffing at the most automated campuses at about 25 to 40 operators per 100 megawatts, Latitude Media reported in May 2026.
Specific project announcements confirm the pattern. Amazon Web Services plans to invest $35 billion by 2040 to establish multiple data center campuses across Virginia. This investment will create at least 1,000 total new jobs across the state, according to the Virginia governor’s office. That is 1,000 jobs over 17 years for $35 billion. Ark Data Centers is building a $136 million campus expansion in Ohio. The project’s final job count is exactly 10, according to Futurism, citing public records.
An average retail data center using two to five megawatts employs about 30 permanent workers, according to Built In. Hyperscale facilities create 100 to 1,000 permanent jobs, depending on size. But even at the high end, the numbers are small relative to the capital deployed.
How data centers compare to other developments
Manufacturing plants competing for the same state incentive packages have different labor profiles. Pharmaceutical company Becton, Dickinson and Company is investing $110 million in a manufacturing expansion in Columbus, Nebraska, creating 120 jobs. A new automotive venture in Orangeburg, South Carolina, is investing $120 million in a new plant, bringing in about 400 jobs. Both projects cost less than Ark Data Centers’ Ohio expansion, which promised 10.
TSMC’s Arizona project illustrates the contrast at the largest scale. The initial $65 billion investment in three fabs is projected to generate about 6,000 direct manufacturing jobs, over 20,000 construction jobs, and tens of thousands of indirect jobs. A semiconductor fab of that size requires human operators running equipment around the clock. A data center of equivalent cost does not.
The structural reason is straightforward. Hyperscale facilities are designed to operate with very few people, and most of the capital cost is in hardware that gets replaced every five to seven years rather than in long-lived infrastructure that requires operating crews, as Latitude Media noted.
The subsidy question
State and local governments have offered data center incentive packages built on factory-oriented frameworks. Almost half of state data center subsidies, 16 out of 36, do not require job creation, according to Good Jobs First, the nonprofit subsidy watchdog. States that impose requirements usually set them at 50 or fewer jobs per project.
The cost per job can be extreme. In one case, a data center in New York promised 125 jobs in exchange for $1.4 billion, or $11 million per job, Good Jobs First found. The average cost of data center “megadeals” is $1.95 million per job, according to a Good Jobs First study.
Virginia offers the clearest case study. The state missed more than $1.6 billion in tax revenue in fiscal year 2025 due to data center tax exemptions, an increase of 118% over the previous fiscal year, according to Data Center Dynamics, citing Virginia’s annual financial report. In fiscal year 2025, the data center industry added 1,610 jobs and reported a tax benefit of $1.9 billion, or $1.2 million per new job, according to VPM.
What the research says about broader effects
The picture becomes more complicated when indirect employment is factored in. Economists Dany Bahar and Greg Wright found that counties that receive their first large data center see total private employment rise by 4% to 5% over five to six years. Construction employment jumps 11%, and information sector employment grows by 22%. Their research, published by the Brookings Institution in May 2026, analyzed about 770 U.S. data center facilities.
At a typical treated county with 98,000 workers, these estimates imply about 2,000 to 4,000 additional jobs after six years, depending on facility type. But the gains depend on concentration. Single facilities produce modest employment gains. The information sector benefits require multiple facilities in the same area.
Data centers do create local jobs, though fewer than industry advocates claim. Naive estimates that fail to account for preexisting growth trends overstate the effect by a factor of three. The Brookings research also found that location decisions for hyperscale facilities are driven by power availability, land, and fiber infrastructure, not by tax breaks. In colocation counties, incentives account for a much larger share of total investment (62%), suggesting subsidies may matter more for facilities that generate the smallest employment benefits.
Economist Michael J. Hicks, examining data center development in Texas, reached a starker conclusion. His estimates concluded that the net effect of data center employment within a county is effectively zero, as workers shifted between industry subsectors rather than entering new positions, he wrote in November 2025.
None of this means data centers provide zero economic value to host communities. Property tax revenue can be significant. In Loudoun County, Virginia, data centers generate 38% of the county’s General Fund revenue and almost half of all property tax collections. But property tax revenue and job creation are different metrics, and communities evaluating data center proposals should know which one they are being offered.
