Washington Obligated Nearly $24bn on Nuclear-Site Contracts. Only Five of 21 Major Builds Had Separate Project Lines.

Factrail | United States | Investigation and documentary analysis | Research cut-off: 31 October 2026

SEO title: NNSA Nuclear Contracts: $24bn, Project Fees and the Oversight Gap

Dek: Federal auditors found separate contract lines for just five of 21 baselined nuclear construction projects. The question is who controls contractor incentives when billions in overruns accumulate.

The finding: a missing decision rule, not a proven stolen dollar

The United States' nuclear-weapons infrastructure agency spent nearly $24 billion on nine management-and-operating contracts in fiscal 2025. Yet a September 23, 2026 federal audit found that only five of 21 ongoing capital projects with approved performance baselines had separately identified project lines within those umbrella contracts. The National Nuclear Security Administration (NNSA) had not documented a consistent rule for when to separate construction work, even though doing so can expose project-specific costs and permit distinct contractor performance fees. A different federal audit recorded $4.8 billion in cumulative cost overruns across 16 projects in execution as of June 2025. The records do not show that the absence of a contract line caused those overruns, that the full $24 billion was spent on construction, or that any contractor stole public money. They establish a narrower but important accountability failure: the government has not consistently documented why it uses a potentially useful tool to isolate costly construction performance from the rest of a contractor's work. GAO's September contract audit (opens in a new tab) and February project assessment (opens in a new tab) provide the documentary basis.

The immediate decision belongs to federal contracting officials, not a president personally signing every site contract. Energy Secretary Chris Wright and NNSA Administrator Brandon Williams oversee the department and agency during 2026; the September audit specifically directs its two recommendations to NNSA's Associate Administrator for Partnership and Acquisition Services. Congress can demand records, attach conditions to appropriations, and hold hearings. None of those actors can retroactively make a 2014 contract decision into a 2026 political act. The question for current officeholders is whether the next contracting decisions will make fees and deliverables auditable.

How a nuclear facility can be inside a vast contract

The NNSA operates government-owned facilities through private or university-linked management-and-operating contractors. Those organizations do not own the nuclear weapons mission. They run sites, supervise subcontractors, maintain facilities and perform specialized work under federal contracts. The agency's own Partnership and Acquisition Services description (opens in a new tab) explains the trade: long-term operators can preserve highly specialized technical capacity, while the government uses performance fees and recompetition to hold them accountable.

A contract line item, or CLIN, is not a separate public tender and does not necessarily transfer construction risk to the private sector. It is a way to describe and account for a distinct deliverable inside an existing contract. A project can then have its own funding trail, schedule requirements, acceptance criteria and potentially its own incentive or award fee. Without a separate project line, construction may remain inside a broader management line that covers many activities. That can make poor construction performance less visible when a contractor earns good marks for unrelated site operations.

The September GAO review is particularly useful because it did not simply count contracts. It examined all nine management-and-operating contracts and interviewed headquarters, field-office and contractor personnel. It found that every contract had four or fewer lines; the Lawrence Livermore National Laboratory agreement, dating to 2007, had none. Five of 21 projects with approved baselines were covered by separate construction lines as of June 2026. That is 23.8% (5 divided by 21), leaving 16 of 21, or 76.2%, without that particular project-specific structure. This is our calculation from the auditor's denominator, not a GAO estimate of wasted money. Some projects may be overseen effectively by other contractual and administrative controls. GAO's contract-line inventory (opens in a new tab) is the source of the count.

The distinction between a line item and a performance guarantee is central. The audit explicitly says that using separate lines has not by itself produced proven cost savings or improved performance. It is an available management instrument, not a cure. That limitation prevents a misleading headline that treats the 16 projects without lines as 16 cases of misconduct.

The documentary timeline: what the government knew, and when

2014: NNSA began using project-specific contract lines, including a line for the Uranium Processing Facility (UPF) at the Y-12 National Security Complex in Tennessee. Officials later told GAO that earlier contractor-performance problems were one reason to isolate this project's requirements and fee.

2018: The UPF received a $6.5 billion project baseline. The old uranium-processing buildings it was intended to replace remained operational. Subsequent cost and schedule problems were not resolved by the fact that UPF already had a dedicated contract line.

December 2024: DOE approved a UPF re-baseline of $10.35 billion, with project completion expected in January 2032 and full operations in 2034. This revised total should not be confused with the $8.469 billion estimated value of the UPF contract line in GAO's September 2026 table: they describe differently defined financial objects, not two additive expenditures. GAO's September 2025 UPF review (opens in a new tab) documents the re-baseline and its consequences.

June 2025: The later portfolio audit's measurement date. Across 16 projects in execution, cumulative overruns stood at $4.8 billion and aggregate schedule delay at about 30 years, compared with $2.1 billion and nine years in the earlier assessment. Thirty years is the sum of project delays, not the predicted postponement of one facility. Two UPF projects at Y-12 accounted for about 80% of that cost overrun.

February 26, 2026: GAO published its biennial assessment (opens in a new tab) of 28 major construction projects, 16 in execution and 12 in definition. It identified management weaknesses, vendor performance and higher input costs rather than attributing all overruns to a single contracting format.

July 15, 2026: NNSA announced authorization (opens in a new tab) for the Y-12 Calciner Project. That is a real modernization milestone and a material counterweight to a narrative that no work is progressing. It does not mean the larger UPF has reached full operations.

September 23, 2026: GAO published the contract-line audit. It recommended written selection criteria and documented lessons learned. NNSA boundary the findings and described planned actions. The audit did not certify those actions as implemented. The distinction matters for October 2026 accountability.

The decision and veto map

The President and Congress set national-security priorities and supply the broader statutory and budgetary environment. The Energy Secretary oversees DOE, within which NNSA is a separately organized agency. The NNSA Administrator leads the agency's execution. The Office of Partnership and Acquisition Services advises on acquisition policy and, through contracting officials, the rules for using separate contract lines. Site contracting officers and program managers decide how to structure individual deliverables within the lawful acquisition framework. The management-and-operating contractor performs or subcontracts the work and is evaluated under the relevant fee plan. GAO audits; congressional committees can challenge policy, require reporting or condition future funds. No single line of this map establishes that a named politician personally directed a specific contractor fee.

The specific accountability gap is at the selection stage. Officials described considering project cost, complexity and dedicated appropriations, but GAO found no documented criteria for deciding when those factors require a separate line. Without a written rule, two similar high-value projects may receive different fee visibility without an auditable explanation. A procurement official can reasonably choose different structures for different risks. The government should still be able to show the decision record.

The September report went to congressional committees, including the Senate Armed Services Committee chaired by Roger Wicker with Jack Reed as ranking member, and the House Armed Services Committee chaired by Mike Rogers with Adam Smith as ranking member. That identifies oversight routes, not evidence that those members designed the contracts. The legal and practical vetoes differ: Congress controls appropriations and law; DOE/NNSA can change guidance and future contract structures; contractors can negotiate terms; auditors recommend but cannot themselves rewrite contracts.

Where the money actually goes

The first financial distinction is between an agency-wide spending total and a specific construction project's price. GAO reports nearly $24 billion in fiscal 2025 spending on nine M&O contracts. The agency also projected approximately $193 billion in modernization and supporting activities over the following decade. These are different periods and scopes. Neither number is a measured contractor profit margin.

The second distinction is between a budget estimate and a contract payment. GAO's table of fiscal 2026 site estimates lists about $2.422 billion for Consolidated Nuclear Security's Y-12 contract, $2.996 billion for the Sandia operator, $4.714 billion for Triad National Security at Los Alamos, and $2.547 billion for Lawrence Livermore National Security. These are agency budget estimates associated with the sites, not a list of payments for the five separate construction lines. The DOE Y-12 contract page (opens in a new tab) identifies Consolidated Nuclear Security as the operator under contract DE-NA0001942, with exercised performance through September 2027. It also publishes modifications and performance-evaluation documents. The contract is a real financial relationship; its existence does not establish improper political influence.

Third, a conspicuous figure in the September audit must be kept in its proper box. GAO's table assigns $25 billion to the Savannah River Plutonium Processing Facility line, but explicitly labels it a fiscal 2026 budgetary placeholder for estimated total project cost. It is not proof of $25 billion disbursed, earned or lost. Likewise the UPF line's approximately $8.469 billion estimate should not be added to the $10.35 billion project re-baseline to invent an $18.8 billion bill. A credible financial investigation has to reconcile scope, date and accounting class before using a dramatic number.

Who receives the contracts? The named management operators include Consolidated Nuclear Security at Y-12, Triad National Security at Los Alamos, National Technology and Engineering Solutions of Sandia, Lawrence Livermore National Security and Savannah River Nuclear Solutions. The GAO report identifies these legal contractor entities and their contract arrangements. Independent Engineering News-Record reporting (opens in a new tab) identifies Bechtel National as the primary UPF construction contractor, with Consolidated Nuclear Security operating Y-12. These organizations have identifiable revenue opportunities from federal work. The sources examined here do not establish their net profit on the disputed overruns, ultimate shareholder distributions, campaign donations linked to the contracting decisions, or a corrupt exchange. We do not infer any of those from their names.

The fee problem is subtler than the cost overrun

Cost-reimbursement arrangements can be appropriate when requirements are difficult to specify. The government reimburses allowable costs, while fixed, award or incentive fees create different reward structures. In an award-fee system, broader contractor performance is assessed; an incentive-fee component can tie compensation to a more objective project milestone. NNSA's own acquisition-office explanation (opens in a new tab) argues that poor performance can reduce fees or eventually trigger recompetition.

The September GAO audit shows why the details matter. Officials said that strong overall site ratings could obscure poor construction performance, allowing significant management fees despite trouble on a particular project. A separate project line permits a distinct performance evaluation. But GAO also found a concrete example of a badly chosen incentive: a Los Alamos Transuranic Waste Facility contractor received a fee after obtaining a certificate of beneficial occupancy, even though the facility was not yet operational for its intended nuclear work. The milestone was achieved; the government's ultimate desired outcome was not. The report does not say the fee payment was illegal. It demonstrates how a contract can pay for the wrong proxy.

UPF supplies the inverse warning. It has had a separate line and performance plan since 2014, yet still experienced major overruns and delay. That is the strongest empirical objection to any suggestion that simply adding lines to the other 16 projects would have prevented $4.8 billion of overruns. A line can reveal a problem without solving design maturity, staffing shortages, subcontractor failures or price inflation. The right test is whether project-specific measures match actual completion, operational readiness, safety and cost performance.

The human and security cost of waiting

Y-12 illustrates a public consequence that is not captured by an award-fee spreadsheet. The Uranium Processing Facility is intended to replace aging enriched-uranium operations. Until it is fully operational, the government must continue using Building 9212, built in 1945 and not designed to modern earthquake and high-wind safety codes. The 2025 GAO UPF audit (opens in a new tab) records a contractor estimate of roughly $463 million in costs associated with keeping that building safely operational through 2035, including maintenance, operations and deferred work. This is an estimate of an additional legacy-building burden, not a final invoice or an automatic addition to the project construction baseline.

The burden falls first on the federal budget and the workers and communities dependent on safe nuclear operations. A delayed replacement can require additional protective equipment, maintenance and contingency planning. It can also complicate production schedules. That does not mean a particular accident will occur, or that the dollar value of all these risks is known. GAO's recommendation for a comprehensive operating plan boundary the risks while recognizing that NNSA must sustain an essential mission.

NNSA's July 2026 calciner authorization is a meaningful counterexample. Administrator Brandon Williams and Consolidated Nuclear Security chief Todd Ailes described a safer process for handling uranium-bearing solutions and reducing inventories in older buildings. The government says this supports transition to the new facility. It should count as progress, while being measured separately from final UPF completion. Another NNSA announcement in May 2026 (opens in a new tab) reported the first phase of Kansas City's KC NeXT expansion completed ahead of schedule and on budget. That demonstrates the portfolio contains successes as well as failures; a fair investigation does not erase them.

The strongest counter-case: separate lines can be expensive, and they are not magic

NNSA officials and contractor representatives told GAO that creating a separate line requires negotiation, administration, performance plans and additional evaluation resources. Some construction projects are too small, too uncertain or too intertwined with ongoing operations to justify that overhead. A flexible umbrella contract can protect mission continuity and reduce the cost of repeatedly competing specialized work. The agency also told auditors that it was standardizing structures as contracts were recompeted, and had considered adding other project lines.

These are serious objections, not evasions by definition. The government's procurement choice should weigh the value of added visibility against the extra administrative expense. The problem GAO actually identified is that NNSA had not documented the criteria for making that tradeoff consistently and had not fully recorded lessons from its previous use of separate lines. Officials could point to one narrow fixed-price lesson but not a comprehensive evaluation of when the tool improves outcomes.

The independent Breaking Defense assessment (opens in a new tab) emphasized that the broader construction portfolio's cost and schedule problems predated the September line-item review. Arms Control Today's April analysis (opens in a new tab) connected construction delays to ambitious plutonium-production goals and explicitly published a correction to its own account of a production target. That correction is a useful caution: a capability goal, a target date and actual production are not interchangeable. Neither independent account proves that adding contract lines would have eliminated the overruns.

NNSA's formal response to the September audit boundary the findings and outlined actions that GAO said, if implemented, would address its two recommendations. This is the agency's strongest documented institutional response. The material reviewed does not verify implementation by October 31. We have not requested a new right-of-reply response from DOE, the contractors or individual politicians; the report uses their available published statements and records rather than implying direct interviews.

What the 2026 decision can change—and what it cannot

The actionable decision now is not whether to punish every contractor named in a government report. It is whether NNSA publishes a coherent selection rule and tests its fee structures against project-specific outcomes. A practical audit trail would show the expected cost and risk of each construction project; why a separate line was chosen or rejected; the contractor's fee at risk; the milestone used to earn it; and whether that milestone corresponds to actual operational capability. Where national security limits disclosure, congressional auditors still need enough controlled access to test the decision.

The most useful measurable indicators are the share of baselined projects with project-specific accountability structures, the proportion of incentive fees tied to operational rather than intermediate milestones, cumulative cost growth against approved baselines, aggregate schedule delay, and verified expenditures to sustain legacy buildings. The September audit establishes a 5-of-21 baseline for one structural measure. The February audit supplies a $4.8 billion cost-growth and 30-year aggregate delay baseline as of June 2025. None is an annual contractor-profit series.

There is also a political accountability distinction. Williams and Wright can direct present-day corrective action within their legal powers; Congress can demand reporting and alter future appropriations; contractors must comply with contract terms and can challenge or negotiate modifications. The historical delays and fee choices span multiple administrations. Holding current leaders responsible for the response is justified. Assigning them sole personal responsibility for a decade of decisions is not.

What would change this finding

A new NNSA acquisition directive with explicit thresholds, documented exceptions and an evaluation of prior line-item outcomes would materially answer GAO's process criticism. Contract-level fee records could show whether contractors lost compensation for overruns or were rewarded despite them. A reconciled boundary of obligated, paid and earned amounts could identify actual beneficiaries more precisely than site budget estimates. A new independent portfolio audit could demonstrate improved cost and schedule results, or show that added administrative burden failed to produce value. Updated Building 9212 safety and spending records could quantify the true consequence of continued reliance on old infrastructure.

Until those records are available, the defensible conclusion is specific: NNSA has a large, technically indispensable private-contractor system; documented cost and schedule overruns are substantial; and the government has not established a consistent written method for deciding when high-value construction should receive separately visible contractual treatment. The $24 billion is not a fraud estimate, the $4.8 billion is not a measure of private profit, and the absence of a line item is not proof of wrongdoing. The failure that can be documented today is the missing decision rule and incomplete institutional learning. That is enough to demand an accountable answer without inventing a crime.

For further context on the institutions, decisions and public consequences in this account, Factrail connects this investigation to Related decision-maker (opens in a new tab), Relevant policy mechanism (opens in a new tab), Public impact context (opens in a new tab). These references provide navigation to related subjects, not independent evidence of an unreported financial transfer.

Sources and documentary checks

1. GAO, National Nuclear Security Administration: Use of Line Items in Management and Operating Contracts Could Be Improved, GAO-26-108409, September 23, 2026 (opens in a new tab) — contract inventory, costs, fee mechanisms, recommendations and agency response.
2. GAO, Nuclear Security Enterprise: Assessments of NNSA Major Projects, GAO-26-107777, February 26, 2026 (opens in a new tab) — portfolio cost and schedule denominators.
3. GAO, Modernizing the Nuclear Security Enterprise, GAO-25-107330, September 18, 2025 (opens in a new tab) — UPF re-baseline and Building 9212.
4. DOE/NNSA, Y-12 National Security Complex Contract (opens in a new tab) — legal operator, contract number and modifications.
5. DOE/NNSA, Partnership and Acquisition Services (opens in a new tab) — government's account of the contracting model.
6. DOE/NNSA, NNSA Grants Y-12 Startup Authorization for Calciner Project, July 15, 2026 (opens in a new tab) — documented implementation and official/contractor responses.
7. DOE/NNSA, NNSA Celebrates First-Phase Achievement of Major Infrastructure Project in Kansas City, May 18, 2026 (opens in a new tab) — counterexample of a reported on-time phase.
8. Breaking Defense, Costs and delays on NNSA construction projects more than doubled since 2023, February 26, 2026 (opens in a new tab) — independent defense-industry reporting.
9. Arms Control Today, NNSA Charts Buildup as Delays Mount, April 2026 (opens in a new tab) — independent nuclear-policy context, with published correction.
10. Engineering News-Record, GAO Finds Mounting Costs, Years-Long Delays Across NNSA Nuclear Projects, 2026 (opens in a new tab) — independently indexed contractor attribution; full page blocked on direct access, so not used to substantiate disputed financial numbers.
11. DOE/NNSA, Lawrence Livermore National Laboratory Contract (opens in a new tab) — operator and contract context.

Editorial status: DRAFT_INDEPENDENT_QA_PENDING. No source establishes personal enrichment, illegal procurement, a final project loss, or completed corrective action. Internal Factrail links, image rights, legal review and importer validation remain unverified.