NASA spent $5.9 billion on moon hardware it is now throwing away

Space 4 min 80 sources

An inspector general's audit found four canceled Artemis projects had more than doubled in cost and slipped years behind. Plus a six-galaxy pileup in the early universe, a new Starship breathes fire, and a Japanese probe lines up an asteroid flyby.

Key takeaways

A NASA audit found four canceled moon-program projects had more than doubled in cost to $5.9 billion and slipped years behind — and would have cost even more if finished.
A $131 million rocket adapter had grown toward a projected $497 million, nine years late: a small contract that shows how big programs quietly drift past sense.
The James Webb telescope caught six galaxies merging in the early universe, SpaceX test-fired its next Starship, and Japan's Hayabusa2 lined up a close asteroid flyby on July 5.

NASA’s own watchdog put a number this week on a hard decision: the agency spent close to $5.9 billion building pieces of its moon program that it has now canceled and may never fly [24][57]. The audit is less a scandal than a window into how the most expensive engineering projects on Earth quietly drift past the point of sense — and what it costs to stop them.

The hardware NASA walked away from

In March, NASA reshaped its Artemis moon program at an event it called “Ignition Day” [43]. It dropped plans for a space station orbiting the moon, the Gateway, in favor of a base on the surface. It abandoned an upgraded version of its giant Space Launch System rocket in favor of a single design. The first crewed landing slipped from the Artemis 3 mission to Artemis 4 [24].

That left a trail of nearly finished hardware. On June 24, NASA’s Office of Inspector General — the agency’s internal auditor — released a memo tallying four canceled projects [57][24]. Three were tied to the bigger rocket: a more powerful upper stage, an adapter to connect it to the Orion crew capsule, and a second, larger launch tower. The fourth was a crew habitat module for the now-canceled Gateway [57].

The combined contracts had “ballooned from nearly $2.8 billion to $5.9 billion,” the memo found, with delivery dates pushed back by up to seven years [43][57]. The auditor’s blunt conclusion: if NASA had let the work finish, it would have cost even more and taken even longer than the contracts already promised [43].

A $500 million adapter

The smallest of the four contracts tells the story best. In 2017, NASA hired Dynetics to build the Universal Stage Adapter — a 33-foot composite ring that links the crew capsule to the rocket’s upper stage [43][57]. The first one was supposed to arrive in 2021 for $131 million [57].

By the time NASA pulled the plug this year, the contract had nearly tripled to $353 million, with delivery slipped to 2028 [57]. The auditor projected the real numbers would have landed at $497 million and 2030 — roughly four times the price, nine years late, for one adapter ring [57]. The larger projects ran the same way: the upgraded upper stage grew from $962 million to a projected $3.7 billion; the second launch tower from $383 million to as much as $2 billion [57].

NASA’s leadership used the audit to defend the cancellations. “For too long we tried to satisfy every stakeholder,” administrator Jared Isaacman said in March. “Billions of dollars wasted. Years lost. Hardware that never launched” [43]. The agency now says it is trying to “modernize acquisition practices” — change how it buys things [43].

Why this matters beyond NASA: the same week, the House Appropriations Committee approved about $55.5 billion for the U.S. Space Force, and singled out the service’s effort to reorganize how it buys satellites as a thing worth praising [29]. Two arms of the U.S. space effort, in the same week, naming the purchasing process — not the rockets — as the thing they most need to fix.

A six-galaxy pileup, 12 billion years back

While NASA counted costs, the James Webb Space Telescope — NASA’s big infrared observatory — caught something rarer: at least six galaxies smashing into one another in the early universe [2]. Radio astronomers had spotted strange emissions that looked like an active black hole. When Webb looked closer, it found not one galaxy but “an entire complex of at least six,” said Aayush Saxena of the University of Oxford [2].

The pileup sits at a distance astronomers measure as redshift 4 — light that left when the universe was about 1.8 billion years old, roughly 12 billion years ago [2]. Redshift is the stretching of light as it crosses an expanding universe; the more stretched, the older and farther the source. This one is expected to feed a supermassive black hole and eventually fuse into one of the largest galaxies in the cosmos [2]. We are watching, in slow motion, a giant being assembled.

A new Starship breathes fire; a probe lines up a flyby

Closer to home, the launch business kept moving. SpaceX fired the engines of its next Starship — the fully reusable rocket it is betting its future on — for the first time on June 26 [1]. The test, called a static fire, holds the rocket down and runs its engines to confirm they work before flight; this one lasted about 15 seconds [1]. It follows a mostly successful test flight of the upgraded “Version 3” Starship on May 22 [1].

And Japan’s Hayabusa2 — the probe that returned asteroid samples to Earth in 2020 — is lining up a bonus encounter [7]. On July 5 it will pass within 1 to 10 kilometers of a small asteroid called Torifune, roughly 450 meters wide, as the rock whips by at 5.3 kilometers per second [7]. “This is one of the closest asteroid encounters ever attempted by a mission of this class,” said JAXA’s Satoshi Tanaka [7]. A spacecraft that finished its job five years ago is still earning its keep.

02 · Lesson · why it matters

Why it's so much harder to stop something than to start it

The money already spent should never decide what you do next — yet it almost always does, and the reasons run deeper than any one person's mistake.

A ring that quadrupled in price

NASA agreed in 2017 to buy one piece of moon hardware — an adapter ring — for $131 million. By this year it had grown toward a projected half a billion, nine years behind schedule. Then NASA canceled it, along with three bigger projects, after spending $5.9 billion across all four.

The easy read is “NASA wasted money.” The harder, truer read is in the auditor’s own line: even at $5.9 billion and years late, finishing would have cost more. The work kept going long after the math said stop. That is not a NASA disease. It is one of the most reliable mistakes the human mind makes, and you make a version of it too.

The trap with a name

Economists call it the sunk-cost fallacy. The principle is simple: money or time already spent is gone, no matter what you do next. The only question that should matter is whether the next dollar is worth it. The adapter doesn’t get cheaper to finish because you’ve already paid $353 million. That’s spent. The real choice is always: from here, forward, is this worth it?

But the mind doesn’t work that way. We feel the spent money as a debt the project owes us — and walking away feels like admitting it’s lost. So we keep paying, to avoid the pain of facing what we already paid. You finish the bad book because you’re halfway in. You stay in the line, the lease, the plan, because leaving means the wait was for nothing. The bigger the spend, the harder the trap holds.

Why a giant program is the worst case

If this were only a quirk of individual minds, a big organization would correct it — that’s what oversight is for. But a program like Artemis sits on a structure that makes finishing easy and stopping expensive, by design.

A contract that size isn’t a line in a budget. It’s jobs in specific towns, in specific congressional districts, held by specific companies who employed people for years. The rocket’s contractors and their workers had every reason to keep the work alive, and the lawmakers whose districts hold those jobs had every reason to fund it. None of that is corruption. It’s the ordinary gravity of a system where real livelihoods are bolted to a project. The spent billions weren’t just sunk cost — they were a web of people whose Tuesday depended on the next phase getting approved.

So the cost grew, and the schedule slipped, and almost no one inside the web was positioned to say stop — because for each of them, individually, continuing was rational. The trap isn’t that people were foolish. It’s that the structure made the foolish choice the sensible one for everyone in it.

Cancelling is not a clean win either

Here is the part the headline misses. Pulling the plug looks like the brave, clear-eyed call — and sometimes it is. But it also throws away nearly finished hardware that real engineers spent years building, and it lands its own cost on the people who built it. NASA framed the cancellation as discipline. The contractors framed it as walking away from work that was almost done. Both were telling the truth from their seat.

That is the quiet shape under the story: there was no choice here without a casualty. Finishing wasted money. Stopping wasted work and jobs. The arrangement that made the program so hard to kill — districts, contracts, livelihoods — is the same arrangement that made killing it hurt. You cannot praise the cancellation without also counting who it cost.

What you’re inside, not above

It’s tempting to read this and feel clever — I’d have stopped it sooner. You wouldn’t have, not from inside. Nobody who depended on the next phase could see the whole, and neither can you, in your own version. The project you keep funding with your evenings, the relationship you stay in because of the years already in it, the career you can’t leave because of what it cost to get here — you are running the same loop, with a smaller budget and no inspector general to write you a memo.

Seeing the trap doesn’t make you immune to it. It just lets you ask the one question that cuts through, the next time you feel the pull of what’s already spent: not “how much have I put in?” but “from right here, forward, is this worth it?” The honest answer is often uncomfortable — and you’ll still find it hard to act on, because you, too, are part of a web that would rather you kept paying.

03 · Lab · your turn

The Sunk-Cost Desk

Rehearse the year-by-year choice to keep funding a project, and feel how the money already spent pulls you past the point where stopping is the right call.

04 · Hope · carry this

It is genuinely hard to stop something you have poured years into — which is exactly why a system that finally finds the nerve to do it is showing a kind of maturity, not failure. We are slowly getting better at telling the spent past from the open future.