White paper
The Cost Overrun Facility
Pre-Allocated Funding Structures for New Nuclear Construction Overrun Exposure
July 2026 · Oppenheimer Energy Ventures, Inc.
Model basis: Proprietary Cost Overrun Model
A framework for pre-allocating construction overrun exposure in new nuclear.
01 Problem
Vogtle was certified at $14B and finished near $35B, seven years late. V.C. Summer was abandoned outright. Both had incomplete designs at construction start, weak supply chains, undertrained craft labor, thousands of field design changes and no plan for cost overrun management. The AP1000 could be the lowest-cost generator if the US builds dozens of reactors, and no company has yet invested to scale those learnings across a fleet, even with 20% savings already proven from Vogtle Unit 3 to Vogtle Unit 4.
Vogtle's estimate climbed from certification to reported outturn across a decade:
Figure 1
The Vogtle Units 3 and 4 cost estimate, 2012 to completion
Hover, tap or focus a point for its estimate.
Vogtle Units 3 and 4 were certified at approximately $14 billion in 2012; by 2021 Georgia Power monitoring reports put the estimate at a $28 to 34 billion range, and the reported cost approached $35 billion at completion in 2023, roughly seven years behind schedule.
Roughly seven years behind schedule: the units were planned for 2016-2017 and entered service July 2023 and April 2024.
Source: DOE Idaho National Laboratory, INL/RPT-25-84701 (2025), pp. 8-9. Reported figures mix dollar-year bases; the report notes comparisons are indicative rather than normalized.
The longer record places Vogtle in context. Construction costs by year of commercial operation, from the published source data of Liu, He, Qiu and Kammen (2025), show United States costs rising across five decades while recent Chinese builds move down the curve.
Figure 2
Construction costs of nuclear reactors by year of commercial operation
United States construction costs rise across five decades to about $15 per watt at Vogtle 3 and 4, while recent Chinese builds decline toward $2 to $3 per watt.
Overnight construction cost of nuclear reactors by year of commercial operation, in 2020 dollars per watt.
Source: Liu, He, Qiu and Kammen (2025), Nature. Source data as published by the authors.
The published figure for Flamanville 3 is a lower bound above $4 per watt.
02 Solution
Oppenheimer Energy invests pre-FID capital against a finalized design, DOE-backed LLI SPV procurement strategy, augmented federal tax credits and a Division of Responsibility that aligns incentives for a $27B build. A tranched cost overrun management structure sits above it, guaranteeing delivery, allocating shared opportunity and risk across participants, achieving cost certainty at every phase of the project and allowing dozens of units of a single reactor design to be built in the US over the next 15 years.
Unit 4 is the datum the projections stand on: the measured Unit 3-to-Unit 4 reduction, then the Idaho National Laboratory's projected cost curve for the AP1000 builds that follow.
Figure 3
Overnight capital cost by AP1000 plant, realized and projected
Derived unit-level costs place Vogtle Unit 3 at $15,600-16,000 and Vogtle Unit 4 at $12,800-13,200 per kilowatt; the Idaho National Laboratory's moderate scenario projects $10,000, $7,800 and $6,200 per kilowatt for the next three two-unit plants, approaching the $7,000 Westinghouse first-of-a-kind estimate.
Overnight capital cost by AP1000 plant, realized and projected, in 2024 dollars per kWe. Each projected plant is a two-unit build. Unit-level values for Vogtle Units 3 and 4 are derived from the combined realized figure of $14,409 per kWe and the DOE Liftoff Report reduction of 15 to 20 percent from Unit 3 to Unit 4, assuming equal capacity per unit; the derivation is indicative rather than normalized.
Sources: INL/RPT-25-84701 (2025), pp. iv, 20-22 and 30-32; the Unit 3 to Unit 4 reduction is the DOE Liftoff Report figure as cited by INL. INL describes the projections as preliminary.
The structure orders cost coverage in three layers. The first layer is the base cost of approximately $27 billion, which includes interest during construction (IDC) and carries the expected cost of the project. The second layer is a $2 billion budgeted contingency, committed within the base budget at final investment decision (FID) and drawn first against any cost variance, which serves the ordinary project management function of absorbing routine estimating error inside capital the project already owns.
Figure 4
Cost coverage: base budget, the overrun facility and total contemplated coverage
Hover, tap or focus a band in the overrun stack for that party's terms.
A column chart of cost coverage. A $29 billion column carries the $27 billion base cost (including construction-period interest) and the $2 billion budgeted contingency, with an illustrative potential-savings band if the project is delivered under budget. A floating stack shows the $20 billion overrun facility split across five funding sources: PPA escalation, or the anchor offtaker, $6.8 billion; contingent equity $1.84 billion; contingent debt $7.36 billion; the EPC contractor $2 billion; and reinsurance $2 billion. A third column shows total contemplated coverage of up to $49 billion, above the fourth column, the $35 billion reported Vogtle total.
Marginal-dollar coverage by source: anchor offtaker $0-16B; contingent equity, contingent debt and the EPC contractor $0-20B; reinsurance $10-20B. The $2B budgeted contingency draws before the facility. The slide labels the anchor offtaker's layer “PPA Escalation”.
Source: OE Cost Overrun Model, master slides (July 2026). Cost sharing structures are illustrative and subject to change
The facility draws only after the $2 billion budgeted contingency is exhausted. Total contemplated coverage near $49 billion exceeds the reported reference outturn, and it is defined before construction rather than assembled through distress. The sovereign facility sits above this, available only for exposure beyond the $20 billion.
A project cannot credibly promise that it will stay within budget, but it can decide before construction begins who pays if it does not.
03 How it works
Overruns beyond the budgeted contingency draw through twenty sequential tranches of $1 billion each, and a tranche begins to draw when cumulative overrun crosses its lower boundary and is fully drawn at its upper boundary. Each tranche carries a fixed blend of the five funding sources, and the blend changes deliberately across the sequence. The design premise is that overrun dollars differ economically according to their depth. An early overrun dollar is relatively likely, arrives while delivery momentum matters most and is best carried by parties whose interest lies in completion, while a deep overrun dollar is relatively unlikely, signals severe project distress and is best carried by parties structured to absorb low-probability, high-severity outcomes. A single uniform split across all twenty tranches would treat those dollars as identical, and the shifting blend prevents that error while giving each counterparty an exposure profile matched to the layer it has expressed willingness to occupy.1
Figure 5
Allocation of a total overrun across the five funding sources
Drag the slider or choose a scenario; each lane fills to its share of the total overrun.
The facility's twenty one billion dollar tranches allocate across five funding sources in shares that change with depth: the Anchor Offtaker carries the largest share through the middle tranches and is capped from tranche 16, the EPC contractor's share tapers, reinsurance attaches at tranche 11 and contingent equity and debt carry the final tranches.
Construction equity and debt investors commit $2bn to cover overrun above budget, funded before the cost overrun facility.
Construction advance repaid through a post-COD PPA price adjustment
Pre-committed equity drawn only if required
Pre-agreed debt capacity drawn only if required
Capped reduction in contractor compensation absorbed into margin
Premium-funded transfer of high-severity layers
Cobalt marks the two modeled scenarios.
Cumulative overrun in $ billions, including the $2B budgeted contingency drawn first.
Modeled precisely at named scenarios.
Cost sharing structures are illustrative and subject to change
The full model also includes a sovereign facility layer above the structure shown here. That layer involves a different class of counterparty and a separate set of policy questions, and this article does not cover it. It will be addressed in a later library article.
Four principles organize the allocation. Strategic capital takes early exposure, because an offtaker motivated by completion accepts layers that yield-motivated capital avoids. Contractual capital enters at the first tranche of the facility, after the budgeted contingency has absorbed the first $2 billion of total project overrun, and tapers in the tail. Insurance capital attaches only to far out-of-the-money severity, because that is the layer the reinsurance market has indicated it can underwrite. Committed financial capital scales in the tail, because pre-agreed debt and equity are the instruments suited to severities that other sources have capped out of.
The anchor offtaker's layer is a construction advance. It draws while the plant is built, a price adjustment to the power purchase agreement is set at the commercial operation date and incremental revenue retires the balance across the offtake term.
04 Application and status
Applied to a $14 billion total overrun, the structure draws the $2 billion budgeted contingency first and then tranches 1 through 12 of the facility: $5.5 billion from the anchor offtaker, $3.76 billion from contingent debt, $1.4 billion from the EPC contractor, $0.94 billion from contingent equity and $0.4 billion from reinsurance.
The facility assigns and caps every dollar before construction begins, an outcome neither Vogtle nor V.C. Summer achieved. A sovereign facility backstops any exposure beyond $20 billion.
Members of the Department of Energy, utility executives, and leading investment firms have reviewed this model, confirming it derisks utility-led nuclear construction for the first time in modern U.S. history.
The market those builds serve is set out in the Department of Energy's deployment pathway: 200 gigawatts of new nuclear capacity by 2050, with a steady state of 13 gigawatts per year reached in 2041.
Figure 6
US nuclear deployment to 2050: cumulative capacity and the annual rate
The Department of Energy pathway reaches 200 gigawatts of cumulative new nuclear capacity by 2050, shown as bars, with the annual rate, shown as a line, reaching a steady state of 13 gigawatts per year in 2041.
Projected US nuclear deployment, 2030 to 2050: cumulative capacity as bars, the annual rate as a line.
Source: US DOE, Pathways to Commercial Liftoff: Advanced Nuclear (September 2024), Figure 32. Intermediate annual values are reconstructed from the figure's published endpoints.
Footnotes
-
Measured in overrun dollars beyond the budgeted contingency, the marginal-dollar participation ranges are $0 to $16 billion for the Anchor Offtaker, $0 to $20 billion for both contingent equity and contingent debt, $0 to $20 billion for EPC participation and $10 to $20 billion for reinsurance. ↩