AI data centers employ very few people: What the numbers how

The article analyzes employment and fiscal impacts of hyperscale AI data centers, finding they create very few permanent jobs relative to capital deployed. Examples: Meta’s $10bn Indiana campus will support ~300 jobs (~$33m per job); TSMC’s $165bn U.S. buildout implies ~12,000 jobs (~$14m per job). State incentives often aren’t tied to job creation, producing high costs per promised job and significant tax exemptions (Virginia lost ~$1.6bn in FY2025). Research shows modest indirect gains — treated counties may see private employment rise 4–5% over 5–6 years and information-sector growth of ~22% — but results vary and can be concentrated only where multiple facilities cluster. Market impact: policymakers may rethink subsidy frameworks and fiscal treatment of data centers; capital-intensive, low-labor profiles reinforce demand for AI hardware (benefiting chip suppliers) while raising scrutiny over taxpayer returns on incentives.