NASA's Canceled Artemis Hardware Cost $5.9 Billion, Double the Plan

NASA's Office of Inspector General⧉ -- the agency's own independent watchdog -- put a combined price tag on four pieces of Artemis moon-program hardware NASA decided in February 2026 to cancel or repurpose: $5.9 billion, more than double the roughly $2.8 billion those contracts were worth when NASA first signed them.▸ Each system had also slipped its delivery date, in one case by seven years, before NASA halted the work. That money doesn't disappear when a stop-work order goes out; it becomes hardware -- some of it flight-ready, some of it corroding in a factory -- that, as of the memo's issuance, NASA had not decided whether it could use for anything else.

In February 2026, NASA announced it was reformulating the Artemis campaign⧉ under the White House's National Space Policy, opting to fly its existing Space Launch System rocket more often rather than build a more powerful version of it, and to build a permanent Moon base rather than stage missions through the Gateway lunar space station it had spent years developing.▸ That decision stranded four systems mid-construction: the Exploration Upper Stage (EUS), a more powerful SLS rocket stage Boeing was building; the Universal Stage Adapter (USA), a connector piece Dynetics was building to attach the EUS to the crew capsule; Mobile Launcher 2 (ML-2), the Bechtel-built tower needed to assemble and launch the upgraded rocket; and the Habitation and Logistics Outpost (HALO), the Northrop Grumman-built module meant to be astronauts' living quarters aboard Gateway. NASA issued stop-work orders on all four between February and April 2026.

The OIG's interim memorandum -- a preliminary status report issued partway through a broader, still-unfinished audit -- tracked what each contract was worth when signed against what it was worth the day the work stopped.▸ Every one of the four had grown. The EUS contract, added to Boeing's existing SLS work in 2017 for $962 million, had more than doubled to almost $2 billion⧉ by the time NASA stopped it, with no delivery date left on the books at all. The USA contract grew nearly 170 percent, from $131 million to $353 million, while its delivery date slipped a full seven years, from August 2021 to September 2028. ML-2's contract value grew 314 percent, from $383 million to almost $1.6 billion. And HALO, whose manufacturing subcontractor found corrosion spreading through the module's structure in 2025, grew from a combined $1.3 billion to $1.9 billion.

Because the audit's premise is a counterfactual -- would these systems have been worth building out, if NASA hadn't reformulated Artemis -- the OIG didn't stop at what each contract already cost⧉.▸ Using each contractor's own historical cost and schedule trends, its auditors projected what finishing the work would have cost and when it would have arrived. For EUS, that projection is $3.7 billion by November 2028 -- about 7.5 years behind Boeing's original schedule, and $750 million above Boeing's own March 2026 estimate. For USA, it's $497 million by May 2030, roughly nine years late, and the contract itself already built in 3 more years of vehicle integration, testing, and launch support after that before the adapter could fly. For ML-2, it's $2 billion by December 2026, with NASA needing one to two more years of verification testing after that -- meaning the launch tower likely would not have been ready to support a mission until at least 2028.

HALO is the one system the OIG couldn't fully project, because the contractor's financial reporting didn't go back far enough to build a reliable model.▸ What it could project is worse than the dollar figures for the other three: continued at its current pace, HALO would not have reached NASA until July 2031⧉, and Gateway -- the lunar space station HALO was built for -- would not have been operational until at least 2032, seven years past its original May 2025 target. Gateway's own independent review board had already found a zero percent likelihood the module would hit its most recent launch-readiness date, warning that the program's 'lack of schedule realism may be driving suboptimal engineering decisions during development.'

This is an interim memorandum, not a final report, and it carries no recommendations by design: the OIG says explicitly⧉ that what it documents here will feed into a still-in-progress audit that will assess NASA's management of the terminated hardware and the impact of the cancellations on the workforce, contractors, and international partners involved -- meaning the fuller reckoning, including any recommendations, is still being written.▸ NASA has also not finished deciding what to do with what it already paid for: as of the memo's release, the agency was still evaluating whether any of the EUS, USA, ML-2, or HALO hardware could be repurposed, without a finalized answer either way.

In its written response, reproduced in the memo, NASA didn't dispute the OIG's numbers.▸ The agency said the cost growth, schedule slips, and contractor performance problems the memo documents 'reinforce the rationale' behind reformulating Artemis toward what it calls a more disciplined, affordable architecture -- but pushed back on the OIG's forward-looking cost projections, arguing they 'rely on past performance under outdated architectural assumptions' that predate the February 2026 restructuring. That leaves two things true at once, both stated in the same document: the systems NASA canceled had already cost far more than planned before anyone pulled the plug, and NASA's own defense of the decision to cancel them is that continuing would likely have cost even more.