S1·09 · Financial substrate

The Capital Stack.

Vogtle, data centers, donations. Three financial layers stacked on one ratepayer. The bill arrives in a different envelope every time, and the balance sheet beneath the politics rarely makes a headline. This is the substrate the runoff is being decided on top of.

Read · ~14 min S1·09 · drafted May 22, 2026
You are reading S1·09 The Capital Stack.
Part of The Pressure Series — Season I's ten reported field studies.
Reads cleanly after S1·06 The Vogtle Bill + S1·07 The Data Center Bill. The architecture this exposes is taken apart in EP 03 The Trust Market.
The stack · live composition
$36.8BVogtle base cost
Layer 1 · Vogtle base
~$11B rate base
added by units 3 & 4 build
Layer 2 · Data-center load
~8,200 MW
2024 → 2030 IRP forecast
Layer 3 · Donations
87% · 61%
incumbent PSC funding from regulated parties
The three layers · documented spine

A rate base, a load curve, and a donor list.

Pull each layer apart. Each is published, each is auditable, each is rendered here from sourced numbers, not vibes. The political question is what they look like stacked together. The Capital Stack is not a conspiracy; it is a balance sheet. The argument is the assembly, not the line items.

Layer 01 · Vogtle
The floor of the stack.
Proven · T1
Original budget
$14.0B
Final cost
$36.8B
Rate base added
~$11B
Avg household / yr
~$420

Units 3 & 4 ballooned from a $14B estimate to $36.8B over fifteen years. Unit 3 came online July 2023; Unit 4 in April 2024. Residential rates rose 23.7% across the recovery period. The PSC moved the bulk to ratepayers in December 2023. None of this is in dispute; it is the documented floor.

Drift SRC · GCV · AJC · GA Recorder · Wiki
Layer 02 · Data centers
The load that arrives next.
Strong · T2
2024 IRP growth fcst
+8,200 MW
vs 2023 IRP
+2,200 MW
Demand multiplier
~3× by 2034
New gas + coal kept
multiple

Georgia Power's 2024 IRP projects roughly 8,200 MW of new load by winter 2030/31, layered on top of the Vogtle base. The largest single driver is hyperscale data-center growth. The same plan kept coal units running and added gas capacity to serve it. Whether the forecast holds is contested; the regulatory commitments built on it are not.

Drift SRC · GA Power IRP · Utility Dive · SACE
Layer 03 · Donations
The grease between rate cases.
Proven · T1
Fitz Johnson
87%
Tim Echols
61%
Comms ≥70% utility
4 of 5
Top donor cluster
Southern / GP

Per Energy & Policy Institute filings, Johnson took 87% and Echols 61% of their reported campaign funds from sources tied to the utilities they regulate. Across the broader commission, four of five regulators have accepted at least 70% from regulated parties. The pattern is filed, public, and ongoing. The dispute is what it means, not whether it exists.

Drift SRC · E&PI · NFOIC · AJC
Confidence spectrum · per-claim transparency

Not every claim deserves the same volume.

Most political coverage gives every assertion the same typographic weight. We don't. Each claim below carries an explicit tier from Proven (5 dots, glowing) to Speculative (1 dot, dim). The site's intellectual posture: "we identified mechanisms," not "we proved corruption." Read accordingly.

Proven Strong Correlated Plausible Speculative
Proven · T1
Vogtle Units 3 & 4 finished at roughly $36.8B versus a $14B original budget.
Wikipedia · AJC · GA Recorder · Third Act GA
Proven · T1
PSC moved $7.6B of Vogtle costs into ratepayer recovery in December 2023.
GA Recorder Dec 19 2023 filing
Proven · T1
Incumbent commissioners Johnson and Echols draw 87% / 61% of campaign funds from regulated parties.
Energy & Policy Institute · disclosure filings
Strong · T2
Georgia Power's 2024 IRP projects ~8,200 MW of new load by 2030/31, driven primarily by data centers.
GA Power 2024 IRP · Utility Dive · SACE
Strong · T2
Multiple expert filings argue the data-center load forecast is uncertain at the magnitude claimed.
SACE · SELC Greenlink · IEEFA
Correlated · T3
Donor concentration at the PSC correlates with rate-case votes favorable to Georgia Power.
PSC docket voting record · E&PI cross-tabs
Correlated · T3
Affordability pressure correlates with primary-cycle political volatility in suburban GA.
GA voter file · ACS county-level · prior cycles
Plausible · T4
A meaningful share of new data-center infrastructure is being underwritten by residential ratepayers.
PSC rate-design filings · industry analyst commentary
Plausible · T4
Vogtle-driven bills are a non-trivial factor in 2026 ratepayer political receptivity.
inference from county-level affordability scoring
Speculative · T5
A specific PSC commissioner's runoff fate is being decided by Vogtle resentment.
narrative extension · not testable in current data
Speculative · T5
Hyperscale operators have a deliberate strategy to socialize grid expansion costs.
interpretive · advocate framing · adversarial
Speculative · T5
Donor dependency is corruption.
contested framing · industry would refuse the label
Bridge · cross-mechanism callback

Three of the seven races on the June 16 runoff ballot (the GOP Senate race, the GOP Governor race, the Lt Gov race in both parties) touch the same regulatory geometry as the Capital Stack. The runoff cohort that returns to vote will be whiter, older, wealthier, and more reliably partisan than the primary cohort that put these races into runoff in the first place. That selection compounds with the financial stack: the voters most insulated from rate-burden are structurally over-represented in the cohort that will decide who supervises the rate-base. The Runoff lens and the Capital Stack lens are not separate stories; they are the same balance sheet, read at two different vertical scales.

Read · S1·10 · The Runoff Tells
Bridge · architecture echo · Season II

The financial instruments described above — rate-base recovery, data-center underwriting, donor-class influence — are the same architectural pattern that gets re-deployed at the data layer in Season II. The composite of you assembled in EP 02 Personhood Inc. is securitized through the scoring architecture in EP 03 The Trust Market, with the same logic — pool exposure, abstract it into a tradable claim, distribute returns upward. The Capital Stack here is the financial mirror of the data stack there. When EP 08 The Class Engine ships, the bowtie that makes the two stacks one story becomes legible. The architecture is reproducible across asset classes; the names are local.

A constellation of competing reads · lens console

One fact about the stack. Four lenses. Four pictures.

Same fact, different argument, different chart. Tabs are asymmetric: stronger-evidence lenses carry a visible stripe. Press A/B/C/D.

The shared fact
"Three financial layers (Vogtle rate base, data-center load growth, regulator donor concentration) are stacked on the same Georgia ratepayer simultaneously."
SRC · PSC dockets · IRP 2024 · E&PI filings
A · Information design
Capital stack · cost flow at scale
three layers, scaled to dollar magnitude · who pays, who profits
B · Information design
Donor constellation · regulator gravity
edge thickness = $ amount · node size = vote influence
C · Information design
Frame slope · growth vs burden
the same fact narrated as opportunity OR extraction
D · Information design
Household spark grid · small multiples
twelve household line items · which moved together, which didn't
A · ECONOMIC LENS
The stack, drawn at scale
"Stacked balance sheets do not feel like decisions. They feel like weather."
Vogtle is the documented floor: $36.8B built, $11B added to rate base, $420 per household per year. The data-center layer is the next 8,200 MW arriving on top. Donations are the lubrication between rate cases. Drawn to scale, the household is the smallest box and the bottom box. The political question is not whether each layer is real, it is what happens when they compound on the same household. People experience politics materially before they experience it ideologically.
capital buildrate basecost recoveryhousehold burden
rate baserecoverycompoundingsubstrate
The distinguishing test
If you remove any one layer from the stack, does ratepayer burden materially decline within 24 months?
What this lens makes visible
stacked cost magnitude
$ scaled across layers
Key shape in the data
household is smallest, lowest
structural geometry
Vocabulary signature
rate base · recovery · burden
regulatory-financial register
What this lens underplays
long-horizon utility value
baseload, decarbonization, growth

What this piece chooses to mark rather than fabricate.

The stack is real in shape and the floor is heavily documented. Several joins between layers (exactly how much data-center build is socialized to residential ratepayers, exactly which donor relationships shaped which votes, exactly how Vogtle bills convert to runoff turnout) are inferred from pattern, not measured to the decimal. Where they are measured, they are sourced. Where they are not, they are marked and tier-tagged above.

Documented · solid

Vogtle cost & recovery structure

$36.8B build, ~$11B rate-base addition, PSC recovery orders. Multi-source, official-filing-backed, not in dispute.

Documented · solid

Donor concentration percentages

87% / 61% figures come from disclosure filings analyzed by Energy & Policy Institute. Methodology and source documents are public.

Documented · contested

Data-center load forecast magnitude

8,200 MW figure is from Georgia Power's filed 2024 IRP. Whether the magnitude materializes is actively contested by multiple intervenors at PSC hearings.

Inferred · plausible

The socialization split

The share of data-center grid build absorbed by residential vs commercial rate classes is still being argued at the PSC. Our framing leans into the residential-burden read; an industry-side analyst would frame it differently.

Available · 2-3 quarters

Per-county rate-burden index

A defensible household-level burden index requires PSC tariff data joined with ACS income at sub-county level. The data exists; the join is not yet published here.

Structurally hard

Did the bill change the vote?

Individual-voter ballot behavior cannot be tied to individual utility bills in any panel data we have access to. We assert mechanism shape, not voter-level causation.

Counter-frame to honor

The utility-executive read

Vogtle provides long-duration baseload generation. Nuclear required massive upfront capital. Long-horizon infrastructure cannot be evaluated solely through near-term rate impacts. This counter-frame is real and we hold it next to the burden frame, not under it.

Permanent caveat

Selection bias risk

We chose this stack after watching this primary. Mechanism narratives constructed in hindsight can produce false coherence. The site's defense against this is the confidence spectrum above and explicit predictions registered with future revisits.

Sources · tier-tagged · verified May 22, 2026

T1 · primary official GA PSC · Data Center Fact Sheet (March 2026) commission's own framing of DC load
T1 · primary official Georgia Power · 2025 Integrated Resource Plan (PDF) load forecast methodology, plant additions
T1 · primary official Georgia Power · IRP filings index multi-cycle IRP history
T2 · investigative analysis Energy & Policy Institute · PSC incumbent donor analysis the 87% / 61% figures
T2 · investigative analysis E&PI · Big-dollar contributors to Pridemore / Eaton cycle-to-cycle donor pattern
T2 · investigative analysis NFOIC · Donors to PSC members cross-source confirmation
T2 · accountability report GCV · $420 / yr per household figure household-impact derivation
T3 · reporting AJC · PSC raises Georgia Power rates for Vogtle December 2023 rate-recovery decision
T3 · reporting GA Recorder · $7.6B tab passed to ratepayers contemporaneous coverage of the order
T3 · reporting GA Recorder · Demand to triple in next decade 10-year load-growth framing
T2 · intervenor analysis SACE · Data center load uncertainty counter-forecast at PSC hearing
T2 · expert analysis IEEFA · Southeast Gas Infrastructure & Data Centers regional infrastructure framing
T2 · structured reference Vogtle Electric Generating Plant · Wikipedia canonical cost-history reference