The MOX Facility: How a Plant to Burn Bomb Plutonium Be...

In 2000 the United States made a serious promise. It signed a treaty with Russia in which each country agreed to dispose of 34 metric tons of surplus weapons-grade plutonium, roughly 17,000 warheads' worth of bomb cores between them, in a way that could never be reversed (U.S. State Department archive). To keep its half of that promise, the United States chose to build a single, first-of-a-kind factory in South Carolina that would take the plutonium out of the weapons and blend it into fuel for commercial nuclear reactors. The plant was pitched at about 1.6 billion dollars and a three-year build. It was cancelled eighteen years later, in 2018, after consuming about 7.6 billion dollars, having never processed a single gram of plutonium into fuel. This is the story of the MOX Fuel Fabrication Facility, and it is the rare public-money case where the goal was genuinely important, the execution was a genuine fiasco, and both of those things stayed true from beginning to end.

What it was, and who was building it

The commitment came from the Plutonium Management and Disposition Agreement, signed on September 1, 2000 and amended in 2010, under which the United States and Russia each pledged to render 34 metric tons of surplus weapons plutonium permanently unusable for arms (U.S. State Department archive). Nonproliferation, not electricity, was the entire point. The plutonium is real and dangerous, and getting it into a form that a state or a thief could never turn back into a bomb is a legitimate national interest.

The method the United States chose was mixed-oxide fuel, or MOX: chemically process the weapons plutonium, blend it with uranium oxide, and fabricate it into fuel assemblies that commercial reactors could irradiate, leaving the plutonium locked inside intensely radioactive spent fuel. To do it, the National Nuclear Security Administration set out to build the MOX Fuel Fabrication Facility at the Savannah River Site in Aiken County, South Carolina (Congressional Research Service). Construction began in 2007, carried out by a contractor consortium first called Shaw Areva MOX Services and later, after a corporate acquisition, CB&I Areva MOX Services, the same entity under a new name (Government Accountability Office).

The cost, in four numbers that must not be confused

The MOX story is told in dollar figures, and the single most important discipline in reading it is to keep four different numbers apart, because they get conflated constantly and they mean entirely different things.

The first is the original pitch: roughly 1.6 billion dollars, on a build of about three years, with a later revised baseline near 4.9 billion (Project On Government Oversight). The second is the construction estimate as it ballooned: by 2012 the Government Accountability Office reported the cost to build the plant had climbed to about 7.7 billion dollars, and it faulted the Energy Department for failing to analyze why its costs kept rising (GAO-14-231). The third is the cost to finish from where they stood: in August 2016 the Energy Department estimated it would take roughly 17.2 billion dollars more just to complete construction, with the plant not finished until around 2048 (GAO-17-390). And the fourth is the lifecycle cost, construction plus decades of operation, which the NNSA put at about 56 billion dollars, a figure Energy Secretary Rick Perry later cited at around 49.4 billion (GAO-17-390).

A one-plant nonproliferation project pitched at 1.6 billion dollars had become, on the government's own numbers, a 50-billion-dollar lifetime commitment that would not deliver a finished plant until the middle of the century. And a warning about arithmetic: the roughly 7.7 billion dollar construction estimate from 2012 is numerically almost identical to the roughly 7.6 billion actually spent by the 2018 cancellation, but they are different quantities from different years, a 2012 forecast and a 2018 sunk cost, and treating the coincidence as confirmation of either is exactly the kind of error the program's own paperwork invites.

The percentage that shows how sunk costs get weaponized

The most revealing single detail in the whole saga is a fight over one number: how complete the plant actually was. It is worth dwelling on, because it is a clean illustration of how the logic of "we have already spent so much, we cannot stop now" gets manufactured.

The contractor said the facility was about 74 percent complete. The Energy Department said it was about 30 percent complete. Both numbers were real, and the Government Accountability Office laid out exactly why they differed: it was a dispute about the denominator (GAO-18-122r). The contractor's 74 percent was measured against its own 2012 cost estimate, which assumed a total project budget of about 6.6 billion dollars, an estimate the GAO had deemed unreliable back in 2014. The Energy Department's 30 percent was measured against its 2016 estimate, which assumed a total budget of about 16.7 billion dollars, an estimate the GAO had deemed reliable in 2017. Divide the work done by a small, unrealistic total and you are almost finished. Divide the same work by the real total and you are barely a third of the way there. The higher number made "we are almost done, do not quit now" sound like common sense. The lower one told the truth, which is that the finish line was still more than a decade and 17 billion dollars away.

The cancellation, and the cheaper road taken instead

By 2018 the Energy Department had seen enough. Secretary Perry notified Congress of the decision to terminate in a waiver letter dated May 10, 2018, and the NNSA issued the formal contract termination notice on October 10, 2018 (Arms Control Association). About 7.6 billion dollars had been spent on the program by then, for a plant that was never finished and never made a gram of fuel (Post and Courier). The precise spent figure carries a caveat worth keeping: depending on the year and whether you count the plant alone or the whole plutonium-disposition program, sources report anywhere from about 5 billion to about 8 billion, with 7.6 billion the commonly cited program total at termination.

Cancelling MOX did not cancel the underlying obligation, and this is the part that keeps the story from being a simple tale of waste. The plutonium still has to go somewhere. The replacement approach is called dilute and dispose: chemically down-blend the plutonium with an inert material and bury it deep underground at the Waste Isolation Pilot Plant in New Mexico. The Energy Department argued this path is far cheaper, on the order of 20 billion dollars against the roughly 49 billion for MOX, with Perry citing a figure of at most 19.9 billion (GAO-17-390). The savings claim was the department's own, with a fully validated NNSA lifecycle estimate still pending at the time, and it carries two real caveats: diluted plutonium is theoretically somewhat easier to reconstitute than plutonium locked in spent reactor fuel, and the disposal site itself has limited capacity, a constraint the GAO warned could bind as early as the mid-2020s. The nonproliferation mission survived the plant's cancellation. It just took a different and cheaper road, one that trades some irreversibility for a great deal of money.

The politics, and a treaty partner who left first

Two footnotes keep the record honest. South Carolina, which stood to lose the jobs at Savannah River and feared becoming a de facto plutonium storage state, sued to block the termination, and the litigation ran until the courts ended the final bid to save MOX in 2020 (Arms Control Association). And Russia, the treaty partner the whole edifice was built to match, had already walked away for its own reasons. On October 3, 2016, before the United States cancelled MOX, Vladimir Putin suspended Russia's participation in the plutonium agreement, citing sanctions and NATO expansion and setting conditions for its resumption (Arms Control Association); Russia moved toward formally exiting the pact in 2025. It is important not to draw the wrong causal line here. Russia's 2016 suspension was a political act that preceded the American decision, but it did not cause it. What cancelled MOX was cost.

The ledger reading

The MOX Fuel Fabrication Facility is the textbook version of a boondoggle, and it earns the label honestly: a plant that quintupled and then decupled past its budget, slipped its finish date by decades, and was killed after 7.6 billion dollars without ever doing the job it was built for. The Government Accountability Office, which flagged the cost growth at every stage, later credited its own oversight of the program with about 13 billion dollars in avoided future costs, a self-reported estimate of savings rather than cash recovered (Post and Courier).

But the ledger has a second column, and it is why this is not a simple story of a government setting fire to money. The goal was never fake. Thirty-four metric tons of weapons plutonium is a real thing that really needs to be disposed of, and the failure here was one of engineering ambition and cost control, not of purpose. The most useful lesson is not "the government wastes money," which explains nothing, but something more specific: a first-of-a-kind megaproject chosen for a good reason can still be the wrong tool, sunk-cost math will always be marshaled to keep it alive, and the discipline to stop, even after 7.6 billion dollars, is worth more than the discipline to start. The plutonium is still waiting. At least now it is waiting on the cheaper plan.

Fact-check notes and sources

Every figure was checked against a primary or authoritative source; links are inline, and dollar figures are attributed to their year and source because the program's different cost measures are easy to confuse.

The treaty and the mission (the 2000 Plutonium Management and Disposition Agreement, amended 2010, committing the United States and Russia each to dispose of 34 metric tons of weapons-grade plutonium, roughly 17,000 weapons' worth combined; the MOX method; and the Savannah River Site location, built by the Shaw/CB&I Areva MOX Services consortium from 2007): the U.S. State Department archive, the Congressional Research Service, and GAO-18-122r.
The four cost figures (the roughly 1.6 billion dollar original pitch on a three-year build with a later 4.9 billion baseline; the roughly 7.7 billion 2012 construction estimate; the roughly 17.2 billion to finish and 2048 completion from the 2016 estimate; and the roughly 56 billion NNSA lifecycle figure, cited by Secretary Perry at about 49.4 billion): POGO, GAO-14-231, and GAO-17-390. These are four distinct quantities and are not additive.
The completion dispute (the contractor's roughly 74 percent against a 6.6 billion dollar total-budget assumption GAO deemed unreliable, versus the Energy Department's roughly 30 percent against a 16.7 billion dollar assumption GAO deemed reliable): GAO-18-122r.
The cancellation and the pivot (the May 10, 2018 notification and October 10, 2018 formal termination; the roughly 7.6 billion dollars spent, with a scope-and-year caveat spanning about 5 to 8 billion; the dilute-and-dispose replacement at roughly 20 billion versus MOX's roughly 49 billion, with reversibility and disposal-capacity caveats; and the GAO's self-credited roughly 13 billion dollars in avoided costs): Arms Control Association, GAO-17-390, and the Post and Courier.
The politics (South Carolina's failed litigation ending in 2020, and Russia's October 3, 2016 suspension of the agreement, which preceded but did not cause the cost-driven U.S. termination): Arms Control Association on the litigation and on the Russian suspension.

This post is informational and journalistic, describing a public program and public records. It is not legal, financial, or policy advice. Figures are drawn from government reports, treaty documents, and reporting, and are attributed to their source and year because the program's cost measures differ by scope; where a figure is a projection or a self-reported estimate, it is labeled as such.

The MOX Facility: How a Plant to Burn Bomb Plutonium Became a $7.6 Billion Hole in the Ground