NextFin News - Southern Company Chief Executive Chris Womack says artificial-intelligence data-center demand is not slowing, pushing back against a wave of market anxiety that the AI infrastructure boom may be peaking. His confidence rests on hard numbers: the utility has more than 17 gigawatts of contracted large-load projects on its books, added 6 gigawatts of new contracts since the first quarter of 2026 alone, and is seeing data-center electricity consumption jump 55 percent year over year. The question is not whether hyperscalers are still building. It is whether the power grid — and the ratepayers who fund it — can keep up without breaking.
Womack's comments, made in a television interview on August 26, 2026, land at a moment when the AI trade is fracturing. Chipmakers face questions about whether soaring capital spending will ever pay for itself, while utilities sit on the other side of the same equation as the unavoidable toll collectors. Southern Company, the Atlanta-based parent of Georgia Power, Alabama Power and Mississippi Power, is making an $81 billion bet through 2030 that the answer is yes. Roughly $68 billion of that plan flows into its regulated electric utilities, where 10 gigawatts of new generation is already under construction and more than 500 miles of new transmission lines are being added.
The centerpiece is Georgia Power's 25-year agreement to serve OpenAI's planned "Project Camellia" data-center campus in Effingham County, near Savannah. The facility is expected to need about 3.2 gigawatts of power delivered in phases from 2028 to 2032 — enough electricity for more than 2 million homes — and it carries a provision that lets the utility cut supply to the campus by up to 1 gigawatt during peak periods to protect grid stability. That is the first such flexibility clause Southern has codified with a data-center customer, and it is the template the entire industry will now be measured against.
The Numbers Behind the Claim
Womack's argument is quantitative before it is rhetorical. Southern told investors on its second-quarter earnings call that it has more than 1.2 gigawatts of data-center load already drawing power on its system, with consumption up 49 percent year to date. Beyond the OpenAI anchor, management said it is pursuing an additional 8 gigawatts of projects in advanced stages of contracting, including 3 gigawatts it expects to finalize in the near term. The OpenAI contract alone pushes the utility roughly 1 gigawatt beyond its recently approved capacity, Womack said in response to an analyst question.
The financial results are already reflecting the shift. Southern reported adjusted second-quarter earnings of $1.13 a share and $2.46 for the first half of the year, ahead of company expectations. Chief Financial Officer David Poroch attributed the improvement to higher usage, customer growth and a rising Allowance for Funds Used During Construction — the accounting mechanism that lets utilities earn a return on capital while new plants and lines are still being built. In other words, the buildout itself is profitable before a single kilowatt-hour is sold to a hyperscaler.
The company's long-term guidance frames the scale of the transformation: management is guiding for roughly 10 percent electric sales growth from 2026 through 2030 and 7 to 9 percent adjusted earnings-per-share growth through 2028 and beyond. For a sector that spent the 2010s and early 2020s managing flat or declining load growth as energy efficiency ate into demand, a decade of double-digit sales growth is not a cycle. It is a new operating regime.
Independent forecasters see the same step-change. Fitch Ratings, in an August 17, 2026 report that affirmed Southern's BBB+ issuer rating, noted that large-load customers are the primary growth driver across the utility's territories through 2030, with more than 17 gigawatts already contracted, 8 gigawatts in late-stage negotiations and more than 75 gigawatts in the prospective pipeline. Southern projects roughly 10 percent sales growth from 2026 to 2030, Fitch said — a number that matches management's own guidance.
The Mechanism: Why Utilities Capture AI Demand Differently Than Chipmakers
The critical distinction between Southern's exposure to AI and Nvidia's is the business model that sits underneath it. A chipmaker must win each generation of product against rivals, defend gross margins, and hope that the revenue from its customers' AI services eventually exceeds the tens of billions they are spending on its hardware. A regulated utility faces no such uncertainty about who pays. It signs a long-term contract, builds rate-base assets to serve the load, and earns an authorized return on that capital — with the costs of serving the new customer borne by the new customer, not by existing households.
That structure is why Womack can argue that the AI buildout puts downward pressure on electricity rates for existing customers rather than upward pressure. Georgia Power and Alabama Power operate under multiyear base rate freezes that run through 2028, and Southern has joined the Ratepayer Protection Pledge. The economics run like this: the fixed costs of the transmission and distribution grid are spread over a much larger sales base, so the incremental revenue from a data center covers its own infrastructure and then dilutes the per-unit cost for everyone else. "When growth is done right, everyone benefits," Womack said.
"The extraordinary economic development momentum and demand for power across our Southeast region that we've seen for the past several years continues, particularly from data centers and other large-load customers."
The OpenAI deal illustrates the risk-sharing architecture. The 25-year term locks in a revenue stream that outlasts any single chip generation. The 1-gigawatt flexible demand-response provision means the data center acts partly as a virtual power plant during system stress, reducing the need for Southern to build marginal peaking capacity that would otherwise be paid for by all customers. And under Georgia Public Service Commission rules, OpenAI pays the full infrastructure and electrical-service costs of connecting its campus, preventing those costs from being socialized.
This is the transmission mechanism that the equity market often misses when it rotates out of utilities on fears that AI spending is overheating. The hyperscaler bears the technology risk — whether its models will be monetized. The utility bears only the execution risk of building permitted, rate-approved infrastructure under a contract that pays regardless. That asymmetry is why Southern raised its five-year capital plan to $81 billion from $76 billion and why it secured a Department of Energy loan package of up to $26.5 billion in early 2026 to help fund the buildout at below-market rates.
The Cyclical Wave Riding a Structural Shift
It would be a mistake to treat all of this as purely structural. There is a cyclical leg to the AI power story, and it matters. Hyperscaler capital spending moves with financing conditions, GPU supply, and the perceived pace of AI revenue generation. If chip demand softens or funding costs rise, data-center construction timelines can stretch, and a utility's carefully sequenced buildout can be left temporarily overbuilt. Southern's stock, up roughly 3 percent year to date through August 26 after closing at $89.93, has traded in a wide range this summer — including a 2.8 percent single-day drop on August 21 — reflecting exactly that sensitivity.
But the cyclical wave is riding a structural shift, and the two should not be blended. The structural case rests on three pillars that do not self-correct. First, U.S. electricity load growth has fundamentally changed direction after more than a decade of efficiency-driven stagnation; data centers now account for a material and growing share of incremental demand. Second, the contracts themselves are long-dated and legally binding — a 25-year power purchase agreement does not evaporate because next quarter's AI revenue disappoints. Third, the infrastructure is purpose-built and geographically sticky: once a campus is sited in Effingham County with substations and transmission routed to it, the load does not migrate.
The cyclical case, by contrast, requires a mean-reversion pattern that is not yet visible in Southern's data. A genuinely cyclical slowdown would show up as contract deferrals, shrinking pipeline additions quarter over quarter, and utilization rates falling back toward the historical norm. Instead, Southern added 6 gigawatts of contracted large load in a single quarter and reports data-center usage up 55 percent year over year. That is not the signature of a cycle topping out. It is the signature of a demand curve shifting outward faster than supply can be permitted and built.
The Counter-Case: Leverage, Regulation, and the Hyperscaler Capex Cycle
The strongest argument against Womack's optimism is not economic — it is financial and political, and it comes with a rating-agency stamp. On August 17, 2026, Fitch Ratings revised its outlook on Southern to negative from stable, affirming the company's BBB+ issuer default rating. Fitch expects Southern's funds-from-operations leverage to run in the mid- to high-5x range through 2028, materially above its 5.0x downgrade threshold, driven by the elevated capital program and the delay in investment cost recovery created by the base rate freezes. The agency also cited "increasing regulatory uncertainty in Georgia and Alabama," including upcoming Georgia Public Service Commission elections and new oversight legislation in Alabama.
This is the real tension in the bull case, and it deserves more weight than a passing mention. Southern plans to fund all incremental capital spending with approximately 40 percent equity, including an $11 billion equity raise from 2025 through 2030. Equity issuance dilutes existing shareholders, and debt issuance at current yields raises interest costs. With total debt of about $77 billion as of June 2026, the company does not have unlimited room to absorb a demand shock without consequences for its credit profile or for the dividend it has raised for 25 consecutive years. Fitch expects significant credit recovery only beyond 2029, in line with Southern's commitment to an FFO-to-debt ratio of 17 percent that year. In plain terms: the payoff is real, but it is back-loaded.
Data centers are also facing growing public backlash over water consumption, land use, and the perception that they raise household bills. Womack acknowledged this directly: "There's noise all across the country about data centers," he said. "Once again, I'll say we've got to do a better job. Hyperscalers need to do a better job of explaining the benefits and value and dispelling some of the misinformation that's out there on social media."
That backlash can translate into real friction. Regulatory approvals for new generation and transmission can be delayed or conditioned. Environmental litigation remains a live risk: on May 18, the U.S. Court of Appeals for the Eleventh Circuit revived a suit alleging that Alabama Power's coal-ash closure plan at Plant Barry violates federal law, and the case remains pending. Southern also faces the universal utility problem of interconnection queues — a data center can sign a contract, but it cannot draw power until substations, lines, and generation are physically in service, and permitting timelines are the binding constraint, not the utility's willingness to build.
Then there is the hyperscaler capex cycle itself. The 8 gigawatts of projects Southern says are in advanced contracting, and the 3 gigawatts expected to close near term, are not yet signed. If AI monetization lags and technology companies pull back spending, those contracts could slip. Womack's answer is that the contract architecture protects the utility: minimum bills, long-term contracts, termination payments, and high collateral requirements all shift risk to the data-center tenant. The vertical integration that makes Southern a "comprehensive one-stop shop for power solutions," as Womack described it, is also a concentration of execution risk — and Fitch's negative outlook is the clearest signal that the market is not yet convinced the execution will be clean.
What to Watch: The Signals That Would Change the Call
The base case is that Southern converts a large share of its contracted pipeline into energized, rate-base load by 2030, delivering on its 10 percent sales-growth guidance and supporting the upper half of its long-term earnings trajectory. In that scenario, utilities with regulated exposure to AI load — Southern first among them in the Southeast — continue to re-rate as infrastructure compounders rather than bond proxies, and credit metrics recover beyond 2029 as rate cases reset allowed returns.
The upside case is that the 3 gigawatts of near-term contracts close ahead of schedule and the flexible demand-response model pioneered with OpenAI becomes standard across the industry, allowing utilities to serve more load with less marginal generation. That would push earnings growth toward the top of the 7-to-9 percent range and could extend rate stability beyond 2028.
The downside case is that permitting delays, litigation, a hyperscaler capex pullback, or regulatory pushback stalls the pipeline. The specific falsifying signal is concrete: if Southern's contracted large load fails to grow meaningfully beyond 17 gigawatts over the next two quarters, or if the 3 gigawatts of "near-term" contracts are not finalized by the end of 2026, the structural-demand thesis weakens materially. A second signal would be credit deterioration: if FFO leverage does not begin declining toward the company's 17 percent FFO-to-debt target by 2029, the negative rating outlook would likely convert into a downgrade, compressing the stock's multiple.
For investors, the time horizons point in different directions. In the short term, Southern's shares will track interest-rate expectations and the broader AI capex sentiment cycle. Over the medium term, earnings will follow the conversion of contracts into rate base — a visible, quarter-by-quarter process. Over the long term, the question is whether AI-driven electricity demand proves to be a durable step-change in national load growth, and Southern's 25-year contracts are the clearest evidence yet that the industry is betting it is.
The AI boom will have its cyclical corrections, and chipmakers may well see their multiples compress if revenue fails to catch up with spending. But the electricity that powers it is not optional, and the utility that holds the 25-year contract gets paid through every cycle. Southern's bet is that America does not have to choose between winning the AI race and keeping energy affordable — and for now, its contracted pipeline says both are still possible.
Explore more exclusive insights at nextfin.ai.
