Georgia's data-center tax break: a loss every year through 2030

Georgia's legislative auditor commissioned an independent review of the state's data-center sales-tax exemption, and the numbers that came back undercut two of the case usually made for it.▸ The evaluation⧉, by the University of Georgia's Carl Vinson Institute of Government for the Georgia Department of Audits and Accounts⧉, found the exemption forgave $474.2 million in state sales tax on data-center construction and equipment in fiscal 2025 alone -- and that even crediting it with every dollar of new state tax revenue researchers estimate it generates, the program has never posted a net fiscal gain in any year from 2018 through a 2030 projection. The same review also cut the state's own estimate of how much of Georgia's data-center boom the tax break actually causes, from 90% in the program's first evaluation to just 30% now.

The Georgia Department of Audits and Accounts -- the state legislature's own auditor -- ordered this study under the Tax Expenditures Transparency Act of 2024⧉, a 2024 law requiring the periodic re-evaluation of Georgia's tax breaks.▸ Its subject, O.C.G.A. § 48-8-3(68.1), was enacted in 2018 via House Bill 696⧉ to waive state and local sales tax on construction materials and computer-server purchases for qualifying data centers, with lower investment and job thresholds in smaller counties meant to pull the industry toward rural Georgia.

The Institute's own year-by-year table shows the exemption forgave $474.2 million in state sales tax in fiscal 2025; the construction spending and jobs the state credits to that exemption generated $41.5 million back in new state tax revenue the same year -- leaving a net fiscal loss of $432.6 million⧉ for the year.▸ That gap isn't a one-year anomaly: the Institute's table shows a net fiscal loss in every year it modeled, from 2018 through a 2030 projection in which forgone revenue reaches $866.7 million and the net loss reaches $780.2 million.

The exemption's first evaluation, in 2022, assumed 90% of Georgia's data-center construction was happening because of the tax break -- a 'but for' estimate the state borrowed from a Virginia legislative study⧉, because too few Georgia data centers had used the exemption yet for the Department of Revenue to release real usage data.▸ By 2025, the Institute had actual DOR filings from 34 Georgia data centers plus a nationwide database of active, under-construction, and announced projects, and built an econometric model on that real data instead. The result: only 30% of Georgia's data-center construction is attributable to the exemption⧉ -- the other 70% would likely have happened anyway. Applying that smaller share to all data-center activity, the Institute still credits the exemption with 8,505 construction jobs and 1,641 permanent operations jobs in fiscal 2025 -- but that's the number now being weighed against $474.2 million in forgone revenue, not the much larger activity the 90% assumption used to claim.

The Institute doesn't lead with the net-fiscal-loss framing above -- its preferred metric is broader 'value-added economic impact,' which counts wages, output, and indirect spending, not just money that lands back in the state's own account.▸ On that measure the exemption looks like a good investment: the Institute calculates $2.86 in value-added economic impact for every $1 of forgone revenue⧉ (an overall 186.3% ROI, 2018-2030), versus $1.32 in value-added impact if Georgia had simply collected that same dollar and spent it on schools, health care, and other public services instead. Both numbers can be true at once: the tax break may still be growing Georgia's economy by more than the alternative would, while never once returning more state tax revenue than it costs. Which comparison matters depends on whether the yardstick is the state's own budget or the state's economy at large -- and the report doesn't resolve that tension, it just publishes both metrics side by side.

The exemption's tiered thresholds -- as low as $25 million invested and five jobs created in counties under 30,000 people, versus $250 million and 25 jobs in the biggest counties -- were designed to pull data centers toward rural Georgia.▸ DOAA's own summary of the study says that goal's results are 'mixed,'⧉ even as it credits the program with succeeding at drawing data centers to the state overall.

The only local tax-abatement data the Institute could obtain, notably, came from four large complexes concentrated in metro Atlanta -- a 'representative' three-building complex worth $2.28 billion that would owe $33.6 million a year in local property tax at the sampled jurisdictions' average mil rate.▸ Local governments abated 17.3% of that, about $5.9 million, and collected the remaining $27.8 million -- a local-level tradeoff layered on top of the state's own forgone $474.2 million, negotiated county by county with no statewide accounting of how much abatement exists beyond that one sample.