Phase A — Understand the business
Lens 1 · Company Overview
Southern Company is a regulated electric-and-gas utility holding company headquartered at 30 Ivan Allen Jr. Blvd, Atlanta, incorporated in Delaware, par value $5/share, trading as SO on the NYSE. It is not one business — it is a portfolio of rate-regulated monopolies plus a small merchant arm:
- Georgia Power — the crown jewel. FY2025 revenue $12,631M, net income $2.9B (+12.1% YoY). Serves metro Atlanta, which has become the densest AI-data-center cluster in the US.
- Alabama Power — revenue $8,235M, net income $1.5B (+8.1%).
- Mississippi Power — revenue $1,695M, net income $215M (+8%).
- Southern Company Gas — natural-gas distribution (Atlanta Gas Light et al.), revenue $5,044M, net income $732M (−1.1%).
- Southern Power — the only non-regulated piece: wholesale generation (gas + renewables) sold under long-term PPAs. Revenue $2,198M, net income $125M (down $203M on accelerated wind-repowering depreciation).
The business model in plain terms: SO spends capital building power plants, wires, and pipes; state regulators (the Georgia PSC, Alabama PSC, Mississippi PSC) let it earn an authorized return on that rate base; customers pay a tariff. Earnings grow when rate base grows and regulators bless the return. The contract structure with end customers is the regulated tariff (not take-or-pay), but the new growth layer is large-load electric service agreements with data centers — typically carrying early-termination payments, minimum bills, and financial security. Consolidated FY2025: revenue $29,553M, net income to Southern $4,341M, diluted EPS $3.92.
The key structural fact: the two integrated electric utilities (Georgia + Alabama Power) generated ~$4.4B of the $4.34B consolidated net income. The regulated electric franchise is the company; gas and merchant are rounding error and, in Southern Power's case, a drag.
Lens 2 · Supply Chain
Map: fuel & equipment suppliers → SO's generation fleet → transmission/distribution → end customer, with the data-center buyers now the marginal demand.
- Upstream fuel: natural gas (the largest cost line — Fuel $4,897M FY2025 ), nuclear fuel (Vogtle/Hatch), coal (declining — Plant Barry Unit 5 converting coal→gas), and renewables. Cost of natural gas at the gas utility was $1,599M. SO is exposed to gas-price volatility, partly passed through via fuel-cost-recovery mechanisms.
- Generation assets: the crown asset is now Plant Vogtle Units 3 & 4 — the first new US nuclear units in decades, the largest nuclear plant in the US. Unit 3 entered commercial operation July 2023; Unit 4 April 2024.
- Equipment / construction chain: EPC contractors, turbine OEMs (GE Vernova-class), transformers and HV equipment — a sector-wide chokepoint in 2025-26 as every utility builds at once. Long-term service agreements (LTSAs) appear as committed cash outflows.
- End customers / the new demand: ~9 GW of large-load contracts signed since 2023, each ≥100 MW, scaled at the Q1 2026 call to 28 projects / 11 GW fully contracted, 23 GW contracted-or-late-stage, pipeline >75 GW. Named hyperscaler counterparties: Google, Meta, Microsoft.
Chokepoint / single-source dependency: the geographic concentration is the supply chain. Georgia Power is the single assigned electric supplier across most of its territory (Territorial Electric Service Act of 1973) — a legal monopoly, which is both the moat (Lens 3) and the regulatory exposure (Lens 10). The data-center demand is concentrated in one service territory under one PSC.
Lens 3 · Competitive Advantages (moats)
- Legal monopoly (the deepest moat). Each operating company holds an exclusive, regulator-assigned service territory. There is no retail competition for the bulk of load. This is a structural moat no amount of capital can replicate.
- Vertically integrated + constructive regulation. SO owns generation, transmission, and distribution in Southeastern states historically among the most utility-constructive in the country. That integration lets it offer hyperscalers a single-counterparty solution (capacity + delivery) — a genuine selling advantage over restructured markets.
- The Vogtle base. ~2.2 GW of incremental carbon-free baseload that competitors cannot build (no one else has finished a US AP1000). For hyperscalers chasing 24/7 clean power, this is a differentiated supply.
- Scale & cost of capital. $156B balance sheet, investment-grade access to debt markets, and a 25-year dividend record give it the cheapest funding among integrated peers — decisive in a capex super-cycle.
Bargaining power: over customers, near-total (monopoly). Over suppliers, moderate-and-weakening — in a sector-wide build-out, turbine/transformer makers and EPC labor hold pricing power. The newer tension: bargaining power over the regulator is eroding as data-center cost-allocation becomes politically charged (Lens 10). The moat is in the franchise, not in pricing freedom.
Lens 4 · Segments
Revenue by operating company, FY2025 vs FY2024 vs FY2023:
| Segment | 2025 rev ($M) | 2024 | 2023 | 2025 net income | Trend / cause |
|---|
| Georgia Power | 12,631 | 11,331 | 10,118 | $2.9B (+12.1%) | Accelerating — data-center sales growth + rate/pricing. The engine. |
| Alabama Power | 8,235 | 7,554 | 7,050 | $1.5B (+8.1%) | Steady — Rate RSE + Plant Barry; solid. |
| Southern Company Gas | 5,044 | 4,456 | 4,702 | $732M (−1.1%) | Revenue up on cold weather/cost recovery; earnings flat-to-down. |
| Southern Power | 2,198 | 2,014 | 2,189 | $125M (−$203M) | Decelerating — accelerated depreciation on wind repowering. |
| Mississippi Power | 1,695 | 1,463 | 1,474 | $215M (+8%) | Steady-small. |
| Consolidated | 29,553 | 26,724 | 25,253 | $4,341M | +10.6% revenue; EPS down (see Lens 5). |
Revenue by type (consolidated): Retail electric $16,343M, Wholesale electric $2,941M, Other electric $953M, Natural gas $5,044M, Other $1,284M. The story is unambiguous: Georgia Power retail electric is where the growth concentrates, driven by data centers; everything else is mature single-digit.
Phase B — Measure performance
Lens 5 · Earnings Result
FY2025 (the 10-K print):
- Revenue $29,553M, +10.6% YoY — strong, well above the utility norm, data-center and rate-driven.
- Operating income $7,285M (vs $7,068M) — up only ~3%, because operating expenses rose faster: D&A $5,501M (+15.7%) as Vogtle and new plant entered service, and O&M $7,066M.
- Interest expense, net $3,238M (vs $2,743M, +18%) — the capex super-cycle is being debt-funded into a higher-rate world; this is eating the topline gains.
- Net income to Southern $4,341M (−1.4%); diluted EPS $3.92 vs $3.99 — a GAAP decline despite +10.6% revenue. The squeeze: D&A + interest + a higher share count outrunning operating growth. (Management frames the growth on an adjusted EPS basis, which excludes some of this; see guidance below.)
Q1 2026 (the 10-Q print):
- Revenue $8,397M vs $7,775M (+8.0%); operating income $2,018M (~flat vs $2,010M); net income to Southern $1,356M (+1.6%).
- Diluted EPS $1.20 vs $1.21 — a slight decline, and the reason matters: diluted share count rose from 1,105M to 1,128M (+2.1%). SO is issuing equity to fund the build; per-share growth lags net-income growth. On an adjusted basis the Street recorded $1.32 vs $1.25 consensus, a beat.
- Data-center usage +42% YoY in Q1.
Balance-sheet flags: Total assets $155,720M; Long-term debt $65,649M (+$6.9B YoY) plus $6,220M current maturities ≈ ~$72B total debt; total common equity $36,016M → debt/common-equity ~2.0x and climbing. Operating cash flow $9,802M did NOT cover capex of $12,737M — the first time the gap has opened (2024: OCF $9,788M > capex $8,955M). Free cash flow is structurally negative and the dividend ($2.94/sh) sits on top of that gap, funded by debt + equity issuance.
Market reaction: muted — the stock trades ~$93, near analyst targets, with the data-center narrative already largely priced. The market is rewarding the adjusted growth story and looking through the GAAP/share-count drag.
Lens 6 · Earnings Calls (sentiment trend)
No transcripts on disk; synthesized from web coverage of the Q1 2026 call and prior commentary.
- Dominant theme (rising in intensity): data-center load. The contracted pipeline narrative has escalated each quarter — from "~9 GW since 2023" in the 10-K to "11 GW contracted / 23 GW late-stage / >75 GW pipeline" at Q1. Management leans hard on high-credit hyperscaler names (Google, Meta, Microsoft) to signal demand durability.
- Affordability is the new defensive talking point. Management now front-runs the cost-shift critique by quantifying that large-load customers create ≥$556M/yr of downward rate pressure (≈$102/yr per residential customer) in 2029-2031. The fact they volunteer this number is a tell that the political pressure is real.
- What they stopped saying: Vogtle construction risk. With both units in service, the decade-long "will-they-finish" overhang is gone from the script — replaced entirely by growth.
- Tone: confident, growth-forward, guidance raised. The risk is that confidence is the consensus.
Lens 7 · Comps
Integrated/large-cap regulated peers, as of 2026-06-21:
| Company | Ticker | Mkt cap | EV/EBITDA | EV/Sales | Trailing P/E | Fwd P/E | Div yield | ROE |
|---|
| Southern | SO | $104.9B | 12.9x | 5.96x | 23.8x | 20.2x | 3.27% | 11.0% |
| NextEra | NEE | $180.9B | 20.0x | 10.2x | 22.0x | 21.4x | 2.87% | 10.3% |
| Duke | DUK | $96.6B | 11.3x | 5.67x | 19.0x | 18.4x | 3.44% | 9.7% |
| Amer. Electric Power | AEP | $69.5B | 13.2x | 5.38x | 18.9x | 19.9x | 2.98% | 12.6% |
5-yr avg ROE: n/a (single-point ROE only; SO's ~11% is in line with authorized-return norms).
Read: SO trades at a premium to the pure regulated peers (DUK/AEP) on P/E (23.8 vs ~19) and roughly in line on EV/EBITDA, and below NextEra (which carries a renewables-developer growth premium). SO's PEG ~2.5. The premium is the market paying up for the Georgia data-center growth optionality — justified directionally, but it leaves little margin for error. SO has the lowest beta of the group (0.34) — defensively owned, which cuts both ways (low volatility, but crowded).
Lens 8 · Stock-Price Catalysts (5-yr pattern)
What has actually moved SO >5%:
- Vogtle milestones (2021-2024) — every cost overrun, delay, and finally the Unit 3 (Jul 2023) / Unit 4 (Apr 2024) commercial-operation dates. The completion de-risked the stock (Goldman/Morgan Stanley framing: removed a decade of execution risk).
- Data-center load-growth revisions (2024-2026) — upward revisions to Georgia Power load forecasts; Wells Fargo flagged them among the highest in the sector. This is now the #1 swing factor.
- Georgia PSC decisions (2025-2026) — the July 2025 rate freeze, the December 2025 ~$16.3B / 9,885 MW certification, and the 2025 PSC elections (Democrats won seats). Regulatory outcomes now move the stock.
- Rates / macro — as a 3.3%-yield, 0.34-beta bond proxy, SO is sensitive to the 10-year Treasury. Rising long rates pressure the multiple and raise its $3.2B+ interest bill.
The pattern reveals: the market reacts to (a) Georgia regulatory outcomes and (b) the marginal data-center datapoint far more than to a quarterly EPS penny. This is a regulatory-and-demand stock, not an earnings-surprise stock.
Phase C — Judge people & books
Lens 9 · Management
- Chris Womack — Chairman, President & CEO. President Mar 2023, CEO May 2023, Chairman Dec 2023; 35+ years inside Southern, former CEO of Georgia Power (its largest, now fastest-growing subsidiary). A relatively new CEO (~3 years) but the opposite of an outsider — he ran the crown asset.
- Track record: delivered FY2024 adjusted EPS at the top of guidance and completed Vogtle Unit 4 — closing out the most troubled construction project in modern US utility history without further blowups on his watch. TSR outperformed the Philadelphia Utility Index over 1/3/5/25-yr periods through 12/31/2024.
- Capital allocation: classic regulated-utility playbook — reinvest into rate base, grow the dividend (25 straight years), modest M&A. The discipline question is whether the $78.1B 2026-2030 capex plan earns its authorized return given financing costs and regulatory lag (see Lens 11/13).
- Comp & governance: CEO total comp ~$19.3M (2024), 171% operational payout, 95% say-on-pay approval. No related-party red flags surfaced; insider ownership is small in absolute terms (typical for a mega-cap utility) —
insider-transactions.csv not present, so skin-in-the-game is modest, alignment driven by stock awards (~$13.8M of the package).
- Archetype: professional manager / lifer-operator, not founder. Right profile for a capital-intensive regulated build-out — execution and regulatory relationships over visionary risk-taking.
Lens 10 · Forensic Red Flags
Income statement / balance sheet / cash flow:
- Cash flow vs. earnings divergence — the headline flag. GAAP net income $4.34B but OCF $9.8B < capex $12.7B; the company funds the gap + the dividend with new debt ($65.6B LT debt, +$6.9B YoY) and equity issuance (diluted shares 1,109M FY25 → 1,128M by Q1'26). Not fraud — it is the economic reality of a utility in a build super-cycle — but it means per-share growth is being diluted and leverage is rising into higher rates. Watch the equity-issuance cadence: every block of new shares is a drag on the adjusted-EPS growth target.
- Adjusted vs. GAAP gap. Management guides and the Street scores on adjusted EPS ($4.50-4.60 for 2026) while GAAP diluted EPS actually fell in FY2025 and Q1'26. The adjustments (Vogtle items, repowering depreciation, etc.) are utility-standard, but the divergence is wide enough to flag: the "7% growth" is an adjusted figure, not a GAAP one.
- AFUDC / capitalized interest. With $3.2B interest expense and a massive construction program, a meaningful slice of cost is capitalized (Allowance for Equity Funds $340M; debt AFUDC embedded). This flatters reported earnings vs. cash — normal and regulator-sanctioned, but it means earnings quality is "regulatory-accounting" quality, not cash quality, until plants enter rates.
- Asset retirement obligations $8.9B (nuclear/coal-ash) and pension obligations — large deferred liabilities, adequately disclosed.
- Noncontrolling-interest losses (−$170M) at Southern Power tax-equity structures — adds NCI complexity but is disclosed and immaterial to the parent.
Regulatory findings (required):
- SEC: No Litigation Releases, no AAERs naming Southern Company in the 2021-2026 window.
- Non-SEC / state regulatory: No fines/penalties found. The material item is political-regulatory, not legal: the Georgia PSC's July 2025 base-rate freeze through 2028, the contested December 2025 $16.3B data-center certification (approved over public objection), and the disputed Georgia Power fuel deal the PSC backed in May 2026 amid cost-shift concerns. These are rate-case and cost-allocation disputes, not enforcement actions.
- 10-K Item 3 (Legal Proceedings): routine — rate proceedings, environmental matters (Plant Barry water quality, ash ponds), no single material adverse litigation disclosed.
- Conclusion: No material accounting or enforcement red flag. The genuine forensic concerns are (1) the negative-FCF / rising-leverage / dilution funding model, and (2) the adjusted-vs-GAAP EPS gap — both structural, both disclosed, neither fraudulent.
Phase D — Project & stress-test
Lens 11 · Forward Projection
Building from FY2025 actuals + management guidance into a base/bull/bear EPS path. SO guides on adjusted EPS; I project adjusted (the basis the market trades on), flagging the GAAP-vs-adjusted gap.
Anchors +:
- FY2025 GAAP diluted EPS $3.92; FY2025 adjusted ≈ $4.20-4.25.
- FY2026 guidance: adjusted EPS $4.50-$4.60 (~7% growth).
- Management long-term: 8-9% adjusted EPS growth through 2028, then 5-7% (raised toward 8%) through 2030.
- $78.1B capex plan 2026-2030 → rate-base CAGR roughly high-single-digit; financed with debt + equity (dilution headwind ~1-2%/yr).
| Scenario | FY2026E | FY2027E | FY2028E | Logic |
|---|
| Base | $4.55 | $4.95 | $5.35 | Mgmt midpoint 2026; ~8.5% growth '27-'28 as Georgia rate base + data-center load ramp into rates. |
| Bull | $4.60 | $5.10 | $5.65 | High end + faster load ramp, constructive 2026 rate case, dilution managed. |
| Bear | $4.45 | $4.65 | $4.85 | Rate freeze caps recovery, dilution + interest heavier than planned, load slips right; ~4-5% growth. |
Implied valuation at base: at ~$93, FY2027E base $4.95 → ~18.8x forward — reasonable for the growth if it lands, but the trailing 23.8x assumes the ramp is near-certain.
Forecast to log (Brier tracker) — not logged in this watchlist run per skill rules; the scoreable base call would be: "SO FY2026 adjusted diluted EPS ≥ $4.55, p=0.62, resolves 2027-02-28."
Lens 12 · Bull vs Bear
Bull case. SO is the cleanest large-cap pure-play on US AI power demand, and unlike a chip or model name, the demand is contracted: 11 GW signed, 23 GW late-stage, >75 GW pipeline with Google/Meta/Microsoft as counterparties. The growth is wrapped in a legal monopoly with constructive Southeastern regulation, funded at the lowest cost of capital in the integrated peer set, anchored by Vogtle's carbon-free baseload (which hyperscalers specifically want). A 25-year dividend Aristocrat at a 3.3% yield with 0.34 beta — it pays you to wait. Earnings surprise potential: each upward load-forecast revision has re-rated the stock; the pipeline implies more to come. The market is right that this is a structurally advantaged growth utility.
Bear case (permanent-impairment risks).
- The affordability backlash politicizes the best regulatory relationship in the portfolio. Georgia froze base rates through 2028; PSC commissioners approved the $16.3B plan over dozens of objecting ratepayers; Democrats won PSC seats for the first time in years (seated Jan 2026). If cost-allocation turns against the utility — data centers don't fully pay their incremental cost, or rate relief is delayed — the authorized return on $78B of new rate base compresses. This is the thesis-killer.
- The financing model dilutes the thesis. Negative FCF, $72B debt rising, and continuous equity issuance mean adjusted-EPS growth is being partly funded by shrinking each shareholder's slice. GAAP EPS already fell two periods running. In a higher-for-longer rate world, the $3.2B+ interest bill grows.
- Demand-durability tail risk. SO's own 10-K concedes "future technological advances could offset or eliminate" data-center electricity demand. Efficiency gains, on-site generation, or an AI-capex air-pocket would leave SO with stranded build-out and contracts whose minimum-bill protections "may not fully protect" it.
Pre-mortem (18 months out, thesis broke): the Georgia PSC — now with new commissioners and an affordability mandate — forced data-center-specific tariffs that the hyperscalers contested or walked from, a 2026 rate case landed below expectations, and a softer AI-capex cycle slowed the load ramp; meanwhile dilution + interest pushed adjusted growth below 7%, and a 24x stock de-rated to a market-multiple ~18x.
Multiples too high? At 23.8x trailing / 20.2x forward, SO is priced at the top of its regulated peer group for a reason (growth) but with little cushion — a regulated utility rarely sustains a ~24x multiple unless the growth keeps surprising up.
Contrarian view (what the market refuses to see): the consensus treats the Georgia data-center boom as pure upside for SO. The market is underweighting that the same concentration that creates the growth (one territory, one PSC, a handful of hyperscaler counterparties) is also a single, politically-exposed point of failure — and that the growth is, for now, being delivered to adjusted EPS while GAAP EPS and per-share value lag.
Lens 13 · Devil's Advocate (short-seller)
Dismantling the bull case:
- Revenue concentration / counterparty risk: the marginal growth depends on a handful of hyperscalers (Google/Meta/Microsoft) in one state. If two of them slow data-center capex or renegotiate, the >75 GW "pipeline" deflates fast — pipelines are not contracts.
- The moat is weaker than bulls think where it matters most — the regulator. A legal monopoly is worthless if the PSC won't grant the return. The 2025 rate freeze + newly-elected Democratic commissioners + active cost-shift litigation mean the regulatory moat is being renegotiated in real time.
- Most dangerous competitor bulls underestimate: not another utility — behind-the-meter / on-site generation. Hyperscalers are increasingly building their own gas turbines and signing direct nuclear/SMR deals to bypass utility queues and rate cases. SO's own 10-K flags this. If hyperscalers self-supply, the 75 GW pipeline never reaches the rate base.
- Worst capital-allocation exposure: committing $78.1B (up 73% from the prior $45.2B plan) on the strength of a demand forecast the company itself says "may vary widely." Over-build into a regulated framework that won't let you fully recover = the classic utility value-destruction pattern (cf. Vogtle's own $14B→$35B history).
- What must hold for $93: ~7-8% adjusted EPS growth sustained, constructive Georgia rate cases through the build, the load ramp landing on schedule, and rates not spiking. Miss the growth by 20-30% (e.g. 5% not 8%) and the multiple compresses from ~24x toward ~18x — a 20-25% de-rate even with flat earnings.
- Single permanent-impairment scenario: an AI-capex air-pocket coincides with a hostile post-2026 Georgia PSC, leaving SO with built-and-unrecovered generation and contracts whose minimum-bill clauses prove unenforceable in practice. Plausibility: low-to-moderate, but non-trivial given the political shift.
Lens 14 · Management Questions (ordered by information value)
- Of the 11 GW fully contracted large load, how much carries unconditional minimum-bill/take-or-pay obligations that survive a hyperscaler cancelling, and what is the weighted-average contract tenor?
- With Georgia base rates frozen through 2028 and new commissioners seated, what specifically protects the authorized ROE on the $78.1B capex plan from regulatory lag or cost-disallowance?
- How much of the 2026-2030 capex will you fund with equity, and what is the expected annual share-count dilution — i.e., what's the bridge from net-income growth to per-share growth?
- What share of the >75 GW pipeline do you assume the hyperscalers will instead self-supply (behind-the-meter gas / direct SMR), and how does that change the plan?
- At what 10-year Treasury level does rising interest expense break the 5-8% adjusted-EPS growth target, given the rolling refinancing of $72B of debt?
- Quantify the gap between adjusted and GAAP EPS through 2028 — what are the specific add-backs and when do they roll off?
- The 10-K warns data-center demand "could be eliminated" by technology — what is your stranded-asset / minimum-recovery protection if load materially under-delivers?
- What is the contingency if the post-2026 Georgia PSC mandates data-center-specific tariffs the hyperscalers reject?
- Is the dividend (now $3.04, ~76% payout) safe through the negative-FCF build years, and what payout ceiling would trigger a growth-rate cut?
- Southern Power's earnings fell on wind repowering — is the merchant arm still strategic, or a candidate for divestiture to fund the regulated build?
- What is the realistic path to a second Vogtle-scale nuclear or SMR program, and have you learned enough from Units 3/4 to not repeat the $14B→$35B overrun?
- How do you keep residential bills politically tolerable while loading $78B of capex — beyond the $556M/yr large-load offset?
- What gas-supply and pipeline-capacity commitments underpin the new gas generation, and what is the price-exposure if gas spikes?
- Where are the binding constraints in the equipment/EPC supply chain (turbines, transformers, labor) on hitting the build schedule and budget?
- What does success look like on ROIC (not adjusted EPS) across 2026-2030, and why should investors believe rate base growth converts to value rather than just to a bigger asset base?