Shrinking something that already works is supposed to be the easy version of engineering. That is the entire pitch behind the Westinghouse AP300: take the AP1000 reactor running at Plant Vogtle in Georgia, cut it down to a single coolant loop and roughly 300 megawatts, keep the same fuel, the same passive safety systems and the same control gear, and skip the part where a regulator has to evaluate a machine nobody has ever built.<\/p>\n
Westinghouse launched it on May 4, 2023, and said at the time that design certification should arrive by 2027.<\/p>\n
It is now August 2026. According to the Nuclear Regulatory Commission’s own project page for the AP300<\/a>, the reactor is still in pre-application activities toward a future design certification application. The application has not been filed.<\/p>\n On August 4, Westinghouse announced who it is bringing in to change that.<\/p>\n Westinghouse and Amentum signed a series of agreements supporting what Westinghouse calls its APX platform<\/a>, announced from Cranberry Township, Pennsylvania. APX is the company’s name for treating the AP1000 and the AP300 as one technology stack rather than two separate reactors, sharing engineering, components, licensing basis, fuel design and supply chain.<\/p>\n The operative line in the announcement is short. Under the most recent of the agreements, the two companies will work together to obtain NRC approval for the AP300.<\/p>\n Dan Sumner, president and chief executive of Westinghouse, said the collaboration is “expected to accelerate the licensing of the AP300 SMR with the NRC”<\/a>, and described a separate track aimed at engineering and execution for gigawatt-scale reactors. Amentum chief executive John Heller framed it around growing demand for firm baseload power, including from AI data centers.<\/p>\n Read the nouns rather than the adjectives and the deal is unusually specific. Nobody is buying forging capacity here. Nobody is buying cranes, modules or craft labor. Westinghouse is buying engineering hours and licensing hours, which is a different scarcity entirely.<\/p>\n The NRC docket is public, and it is a fair record of where the design actually stands. Eleven white papers have been submitted and returned with staff feedback, covering core design and fuel cycle, instrumentation and control, probabilistic risk assessment, emergency planning zone definition, seismic categorization and the rest.<\/p>\n One planned white paper never arrived. The row on the NRC table titled “Licensing Application Approach” is marked as no longer planned.<\/p>\n The regulatory engagement plan has been updated twice since the original 2023 filing. The February 2024 version is public. The June 11, 2026 version is not: Westinghouse filed it with an affidavit requesting that it be withheld in full under 10 CFR 2.390, signed by Anthony Schoedel, the company’s director of global licensing, on the grounds that disclosure would help competitors meet NRC documentation requirements without paying for it.<\/p>\n That is a legitimate and routine use of trade-secret protection. It also means the current public record of how Westinghouse plans to license its small reactor stops in early 2024.<\/p>\n The AP300 has had a smoother run abroad. It entered the UK’s Generic Design Assessment in 2024, and a privately financed developer has picked it for four units in northeast England<\/a>. The American certification is the one still outstanding.<\/p>\n Amentum is not a household name and does not build reactors. It is a Chantilly, Virginia engineering and technical services contractor with more than 50,000 employees across roughly 70 countries, and a backlog it reported at $47.8 billion as of April 3, 2026.<\/p>\n Its relevant credential is not concrete. It is regulatory paperwork. The company says it has supported clients across every UK Generic Design Assessment since that process was created, for both large reactors and small modular ones.<\/p>\n Which is why the recent contract list is worth reading side by side. Rolls-Royce picked Amentum as its global program delivery partner for SMR deployments in the UK and Czech Republic. EDF awarded it a ten-year, $730 million professional services contract for British stations. Then Great British Energy \u2013 Nuclear awarded an Amentum-led joint venture a $406 million, 14-year contract as owner’s engineer<\/a> on the UK’s SMR program in Wales.<\/p>\n Now it works on the AP300’s American licensing too. The same firm is being rented by reactor vendors who are competing against each other, which tells you the shortage is real and that it sits upstream of anyone’s brand.<\/p>\n Four days before the Amentum announcement, Westinghouse co-owner Cameco filed its second-quarter results and disclosed the pipeline the company is now planning against.<\/p>\n The 91 units break down into tiers ordered by how close each is to a final investment decision. Ten come through the Department of Energy loan program that is buying long-lead components including the reactor pressure vessels<\/a>. Ten more come through the Cameco, Brookfield and Commerce Department partnership. Two are the restart of the abandoned V.C. Summer units in South Carolina<\/a>. Three are Poland’s first nuclear plant on the Baltic coast<\/a>. Two each go to Bulgaria and Ukraine.<\/p>\n Then eleven units sit in a tier Cameco labels FEED-stage, meaning front-end engineering and design, and 51 more sit in early origination across Canada, India, Saudi Arabia, Slovakia and elsewhere. Cameco’s filing is explicit that projects may stall or leave the pipeline at any point.<\/p>\n Here is the part that makes the Amentum deal legible. The cost and schedule improvements in that disclosure are not free. They arrive only at what the industry calls Nth-of-a-kind, defined in Cameco’s own terms as five separate two-unit projects completed and a sustained order rate of at least two reactors a year. Overnight capital costs fall from $20\u201326 billion to $14\u201317 billion under those conditions.<\/p>\n Two reactors a year, every year, forever. Not two reactors built in sequence by one team, the way Vogtle 3 handed its lessons to Vogtle 4. Multiple sites running at once, each needing its own site-specific licensing package, its own regulatory submissions, its own design engineering staffed by people qualified to sign nuclear safety documents.<\/p>\n Dominic Kieran, who chairs the Westinghouse board and runs Cameco UK, told the results call the company sees few bottlenecks around construction<\/a> and called its assessment “prudent”. He is probably right about construction. That is exactly why the constraint moved somewhere else.<\/p>\n The famous Vogtle number is the CA20 module. Georgia Power put it at more than 2.2 million pounds, taller than five stories, with a footprint of about 67 by 47 feet, craned into the Unit 3 nuclear island in March 2014 as the heaviest lift of the project to that point.<\/p>\n That is the photogenic part, and it is not the part anyone is short of now.<\/p>\nThe deal buys engineers, not welders<\/h2>\n
The AP300 has been in pre-application for 39 months<\/h2>\n
Amentum is already the hired engineer for two of Westinghouse’s rivals<\/h2>\n
Ninety-one reactors is a drawing problem before it is a concrete problem<\/h2>\n
Vogtle proved America could build one<\/h2>\n