Georgia is #2 in data centers. Your bill is paying for it.

Georgia is the second-largest data-center market in the country.

Hyperscale data centers — the warehouse-scale facilities that run AI training, cloud services, and content delivery — have clustered in Georgia at a pace that briefly made the state #1 nationally and currently puts it second to Northern Virginia. Atlanta metro, Douglasville, Newnan, the I-20 corridor.

The footprint is real. The demand is real. The demand is growing. Microsoft, Google, Meta, Amazon, QTS, and a longer tail of operators have all announced or activated facilities here within the last five years.

The forecast that rebuilt the grid.

In December 2024, Georgia Power filed its 2024 Integrated Resource Plan with the Public Service Commission. The headline number: 8,200 megawatts of new electric load expected by winter 2030/31. The 2023 IRP, filed a single year earlier, had projected 5,400 MW.

Between the two filings, the utility's modeling absorbed a wave of public announcements from the named hyperscalers above. Roughly 75% of the new load is attributable directly to hyperscale data-center buildouts. The longer arc the utility projects publicly: approximately 47% load increase by 2031, the largest in three decades.

Most facilities arrived with a tax holiday.

The footprint did not appear by accident. In 2008, the Georgia legislature passed HB 1192, exempting high-technology data-center equipment purchases from state sales and use tax. The exemption has been periodically reauthorized — most recently in 2018 with a sunset extended into the 2030s — and is the foundational fiscal incentive that brought the cluster to Georgia.

The fiscal cost to the state is largely undisclosed at the individual-facility level. Aggregate exemption value is recorded in Georgia Department of Revenue annual tax-expenditure reports; the per-facility breakdown is treated as proprietary. The structural fact: the same legislature that approved the exemption does not require it to publish the size.

Coal stayed on.
Gas was added.

To meet a forecast that effectively triples winter peak demand by 2030, the utility proposed keeping three coal plants online past their planned retirement, building new gas peaker capacity, and accelerating transmission buildout across north Georgia.

The PSC approved the plan with modifications in April 2025. The contested question, pressed by Southern Alliance for Clean Energy, Georgia Watch, and the Sierra Club, was not whether the data centers were arriving. They were. It was whether the forecast was accurate, and who would pay if it wasn't.

Twelve hyperscale sites are approved or announced in Georgia by 2030. Three counties — Douglas, Coweta, and Newton — account for over 60% of the announced megawatt capacity. The case made to county commissioners during incentive negotiations frequently implies thousands of permanent jobs. The typical permanent staff at a hyperscale data center after construction is roughly forty.

The grid built for them.
The bill paid by you.

The hyperscale data centers driving 75% of the new load growth pay roughly 24% of the buildout cost. The remaining 76% is absorbed by residential customers, small commercial accounts, and existing industrial users, most of whom never agreed to the build.

The estimated incremental monthly residential bill impact from data-center-driven grid buildout: approximately +$8.50. Small per household. Large per state. Larger still when stacked beside the $420/year already arriving through Vogtle. The two costs do not sit on separate bills — they sit on the same one, approved by the same Public Service Commission, using the same rate-base recovery mechanism.

This is the second financial layer the publication has now documented. The first — S1·06 The Vogtle Bill — traced $35 billion in nuclear cost overrun from a Georgia kitchen table to an institutional dividend. The same mechanism is now socializing the AI build. The integration of both stacks plus a third (donor influence) lives in S1·09 The Capital Stack.

The political ask is simple. The political will is not.

The reform agenda is publicly named by SACE, Georgia Watch, and the Sierra Club, and is not technically contested. A dedicated hyperscale tariff would allocate the buildout cost to the customer that triggered the load. A sunset on HB 1192 would let the legislature decide annually whether the exemption is still earning its keep. Per-facility disclosure of the tax expenditure would let the public read the same balance sheet the utility and the Commission already see.

The bills exist. They are introduced each session by a small bipartisan caucus and die in committee. The PSC commissioners who would have to authorize a hyperscale tariff are elected statewide in low-turnout cycles. The donor architecture that shapes those races is the subject of the next piece in this season.

The ask is simple. The will is not. The grid is still being rebuilt for someone who is not you.