Summary

  • Georgia Power says uprates at Vogtle and Hatch would add about 96 MW of capacity on its ownership share, while Google would receive the associated zero-emission credits under a proposed NU-1 tariff.
  • The utility projects roughly $900 million in customer benefits over the units’ lives. That is a forward estimate, not a realized saving, final tariff determination or independently reproducible calculation.
  • Regulators still need to expose how the subscription, plant costs, output and customer credits fit together; a 2025 planning table’s 112 MW figure is not reconciled with the newer 96 MW announcement.

The commercial proposition turns on a distinction often compressed into the phrase “nuclear power for a data centre.” Georgia Power says the proposed uprates would put approximately 96 megawatts of additional capacity on the grid for its customers. Google, under the proposed Nuclear Uprate tariff, would receive the zero-emission credits associated with electricity generated by those uprates. Those are related products, but they are not the same product. Capacity is a system resource; energy is generated and dispatched over time; a credit assigns an environmental attribute.

Nothing in the public description says Google receives a physically dedicated 96 MW block, or that its load is matched to the plants hour by hour.

That separation is the deal’s most important design feature—and the first point regulators and customers should be able to audit. If Google pays for the attributes while the incremental capacity serves Georgia Power’s system, the public-interest case depends on the financial bridge between the two: what the subscriber pays, which costs its payment covers, which costs remain in utility rates, and how any excess or shortfall is allocated. The release says the structure is intended to protect customers who do not participate from incremental uprate costs.

It does not disclose the tariff’s full rate design, subscription price, duration, settlement rules or downside-sharing terms. Until those documents are available, “protection” is a proposition to test, not an observed outcome.

Georgia Power and Google announced the agreement on 21 September 2026, subject to approval by the Georgia Public Service Commission. The utility says the proposal covers uprates on its ownership portion of units at Plant Vogtle and Plant Hatch, adds about 96 MW of grid capacity and is expected to enable about $900 million in customer benefits over the units’ lives. It also describes a new NU-1 tariff, a request for an extended power uprate at Hatch Units 1 and 2, and the earlier approval of a Vogtle uprate. These are the company’s descriptions of a pending arrangement, not a Commission finding that the value has been delivered. Georgia Power’s announcement

The $900 million number needs a denominator, a time profile and a counterfactual before it can be evaluated. Is it nominal or discounted? Which customers receive the benefit, and through which rates or credits? Does the estimate compare the proposed structure with a case in which the uprates proceed without Google, with a case in which they do not proceed, or with some other resource plan? Does it include avoided capacity costs, fuel and operating effects, subscription revenue, tax treatment or financing? Public materials reviewed for this commission do not provide a calculation that answers those questions.

The claim may prove sound; it is not independently derivable from the headline.

There is also an unresolved capacity discrepancy. Georgia Power’s 2025 Integrated Resource Plan projected 92 MW of company-owned nuclear uprate capacity in 2030 and 112 MW in 2031–34. The September 2026 announcement instead cites approximately 96 MW. The sources available here do not reconcile the dates, plant scope or measurement basis. It would be wrong to declare either figure erroneous or invent a reason for the gap. The older table is a planning schedule; the newer figure is a company estimate attached to a proposed transaction. A regulator-facing filing should show which units, ownership shares, expected output and schedule produce the 96 MW figure. The utility’s IRP page remains available, although the linked 2025 plan PDF returned an error during this review. Georgia Power’s IRP materials · 2025 IRP PDF

The units do not all stand at the same regulatory starting line. In July 2025, Georgia Power said the Commission approved a Vogtle Units 1 and 2 EPU described as adding 54 MW, while Hatch received only preliminary planning, licensing and engineering approval for potential uprates. The current announcement describes a new Hatch EPU request. The separate PSC filing record numbered 228893 concerns the Hatch Units 1 and 2 request; a filing is evidence that an application entered the process, not that the Commission approved its economics or construction. The company’s 2025 announcement is explicit about the distinction between Vogtle’s approval and Hatch’s preliminary stage. Georgia Power’s 2025 IRP approval announcement · PSC filing record 228893

An EPU is not a new reactor. Georgia Power describes equipment changes—such as turbines, pumps, motors and cooling systems—that allow a unit to operate at a higher licensed thermal power level and produce more electricity. The utility’s share is also not the same as each plant’s total output: Vogtle and Hatch have other owners, as the Plant Hatch and Plant Vogtle pages show. The federal Nuclear Regulatory Commission’s license-amendment process is a different control surface from Georgia’s review of the utility plan, tariff and customer cost allocation. Approval at one level cannot substitute for the other. NRC power-uprate overview · NRC public-involvement guidance

The right test is not whether the arrangement uses existing nuclear sites or carries a large customer’s clean-energy label. It is whether each ledger can be reconciled after the work is defined. The tariff should state the subscriber’s payment and duration, identify which uprate and service costs it covers, specify who bears overruns or underperformance, and describe how ZEC volumes are measured and retired. The utility should report actual unit output against the approved baseline, the attributes transferred, incremental system costs and customer credits by class.

Independent review should be able to reproduce the claimed benefit without treating a forecast as cash already saved.

If the subscriber payment covers costs that would otherwise fall to other customers, and the added capacity reduces system costs without shifting overruns back to them, the model could create genuine shared value. If credits are transferred while cost recovery is broad, or if projected output falls short while the tariff leaves customers exposed, the same headline structure could distribute benefits and risks asymmetrically. No evidence reviewed establishes either result today. That is precisely why the Commission’s review matters.

What to watch

The next disclosures should answer five questions: what tariff and subscription terms the Georgia PSC approves; how the 96 MW estimate maps to units and ownership shares; how it reconciles with the IRP’s 92 MW and 112 MW planning figures; what Hatch licensing, outage and construction milestones follow; and how actual output, costs, credits and customer benefits will be reported against a stated baseline. The $900 million should remain labelled projected until the underlying calculation and realized results can be checked.

Sources: Georgia Power–Google agreement; Georgia Power 2025 IRP page; PSC filing 228893; 2025 IRP approval announcement; NRC uprate guidance.