NASA’s Artemis cost overruns make case for greater adoption of AM in the future

The Artemis II mission to the Moon had the great benefit of reawakening many people’s passion for human space exploration and making it seem more within reach after half a century of very successful satellites and robotic probes. Yet NASA’s Artemis program has also become one of the most expensive and delayed space efforts in the agency’s history, with costs estimated at over $93 billion through 2025 according to NASA’s own Office of Inspector General. The reasons are multiple and well-documented, and astronaut safety—particularly in the critical re-entry operation—is of the utmost importance. Less examined is a contributing factor that the AM industry has watched with mounting frustration: the program’s limited, almost cautious embrace of additive manufacturing at a moment when the technology had matured enough to deliver meaningful savings.

To be fair, when the Artemis program began, AM was probably not yet sufficiently developed for large-scale implementation. In addition, NASA’s regulatory hurdles and safety concerns have to be prioritized, as is always the case for “first-ever” missions (let’s be fair, this program has almost nothing to do with the Apollo missions and all technology had to be reinvented almost from scratch). Can you imagine what would have happened if the mission failed or—even worse—the astronauts did not safely re-enter Earth? The entire manned space program would likely have had to be scrapped. Better not to take any risk.

But the future belongs to AM. Commercial companies are developing rockets at extremely low cost, and NASA will progressively open up to using more AM in its own processes and supply chain. Even L3Harris (the owner of Aerojet Rocketdyne, the maker of NASA’s rocket engines since the Apollo missions) recently highlighted the importance of AM in its manufacturing.

The use of AM in the Artemis program to date has been real but modest. We probably don’t know the full extent across the program, and if Artemis used AM without publicizing it, that would be a missed opportunity as well. What is on the record so far falls well short of a manufacturing strategy. But that is likely going to change.

AM on board

The premises were good. Back in 2018, Lockheed Martin, working with Stratasys and Phoenix Analysis & Design Technologies (PADT), began integrating more than 150 3D printed thermoplastic parts into the Orion spacecraft, using ULTEM 9085 and Antero 800NA—a PEKK-based material with electrostatic dissipative properties—for components requiring chemical resistance and structural performance in deep space conditions. These include brackets, covers, ducting components, and a standout piece: the external docking hatch cover and ring, a six-piece assembly roughly one meter in diameter, printed in PEKK-based ESD material specifically to eliminate the need for secondary coatings or nickel plating. When Artemis II flew in 2026, observers also spotted what appeared to be a printed ULTEM component visible through the crew module window.

On the metal AM side, 12 RCS thruster nozzle extensions on Orion’s crew module were produced on a single AM machine in approximately three weeks, roughly 40% faster than conventional manufacturing methods. These are attitude control components that directly influence re-entry orientation and crew safety. The qualification process behind them, encompassing build parameters, post-processing, coupon testing, non-destructive evaluation, and flight feedback, represents the genuine complexity of human-rated AM. These contributions are real but minimal and not a program-wide directive.

L3Harris also makes the RL10 engines, which serve as the crucial upper-stage propulsion for NASA’s Artemis missions, powering the Interim Cryogenic Propulsion Stage (ICPS) on SLS Block 1. Using liquid hydrogen/liquid oxygen, it provides ~25,000 lbs of thrust to propel the Orion spacecraft from Earth orbit to the Moon. Artemis II & III (both crewed) use the upgraded RL10C-2 variant, which features improved ignition and propellant control.

Unreal(ized) savings

The contrast with what was possible is instructive. Aerojet Rocketdyne (now owned by L3Harris), a key propulsion supplier for Artemis, demonstrated years ago that metal AM could transform the economics of rocket hardware. Using Velo3D’s laser powder bed fusion platform and topology optimization from nTopology, the company redesigned its R-4D reaction control system thruster in titanium, achieving a component that is one-fifth the mass, half the size, and one-third the cost of its conventionally manufactured predecessor, with far fewer assembly parts and a lower failure risk. That work was conducted as an internal R&D effort—not as a program-wide directive.

On the Space Launch System itself, the RS-25 engines used for the first Artemis missions were originally built for the Space Shuttle. The most compelling AM data for the SLS relates to new-production engines being prepared for Artemis V and beyond: a 3D printed pogo accumulator assembly manufactured via selective laser melting eliminated over 100 welds, reduced production time by more than 80%, and cut costs by approximately 35% for that component alone. A metal AM Orion Main Engine injector was also hot-fire tested at White Sands across 21 firing runs in 2023–2024. These results are significant. They also arrived late, and they remain exceptions rather than the norm.

Isolated pockets of AM excellence within Artemis’s supply chain coexist with a procurement and qualification ecosystem that defaulted to traditional subtractive manufacturing for the vast majority of flight hardware. The SLS core stage, its engines, and its structural elements were built largely through processes familiar to Apollo engineers, at costs Apollo engineers never had to absorb at current labor and material rates.

This is also changing. L3Harris significantly enhanced its AM capabilities with its 2019 acquisition of 3D Materials Technology in Daytona Beach, Florida. The facility now handles the production of key thruster components, including nozzles, manifolds and combustion chambers, that were previously machined at other L3Harris facilities. These components are often made from niobium and other exotic, high-strength, heat-resistant metals that can withstand the rigors of spaceflight. Machining large blocks, or billets, of these expensive materials into complex engine components through subtractive manufacturing can be wasteful and inefficient.

With AM, by contrast, L3Harris can buy these metals in powdered form, which is less expensive and easier to store. The company uses LPBF at its Daytona Beach facility and has now overcome one of AM’s biggest challenges: variability in quality between identical machines. Through extensive testing and fine-tuning, the company minimized variability to enable production at scale and take advantage of AM’s key benefits, including reduced part count, increased design freedom and shortened build-to-test timeframes for rapid iteration and design refinement.

“The idea is to have multiple machines running simultaneously to achieve scale,” Houston said. “L3Harris-built thrusters with additively manufactured components are now flight proven on national security satellites, both experimental and operational.”

L3Harris’s embrace of AM began with satellite thrusters, but it is now also being used on the company’s RL10 engine, which powers the upper stage of United Launch Alliance’s Vulcan family of rockets and NASA’s SLS.

“We’ve been perfecting this for over a decade,” Houston added. “We’ve got the infrastructure and people with the proper training in place. We’re already demonstrating high production rates.”

Built for another era

NASA’s flight certification processes were designed around conventional manufacturing, and updating them to accommodate metal AM—with its build-parameter sensitivity, post-processing requirements, and microstructural variability—requires sustained investment in standards development that the agency has pursued incrementally rather than urgently. The result is that suppliers capable of delivering AM components often face qualification timelines that erode the cost and schedule advantages the technology would otherwise provide.

NASA cannot take risks the way a commercial operator can. Something similar (on a smaller scale) happened in Formula 1: the sport operates under extreme performance and safety pressure, and most teams were reluctant to implement AM. Yet, those who opened up to AM early gained measurable competitive advantages. The Aerojet Rocketdyne RCS work, the Lockheed Martin polymer parts, and the RS-25 pogo accumulator data are great examples. The technology is available. The decision not to deploy it systematically across Artemis was an inevitable policy choice, and the program’s cost and schedule record reflects it. We’re not saying that future rockets will be entirely 3D printed. Relativity, unfortunately, demonstrated that’s not yet a viable path. But more and more parts, larger parts in both metal and polymers (and ceramics) will and the heavy use of AM will greatly accelerate development times and reduce costs.

The positive dimension is that many price-conscious private efforts in the space industry benefited from the massive public investments required by Artemis, and their progress will carry forward into future missions. The Moon and, eventually, deep space will become more accessible as a result. But the lessons of Artemis—about what happens when proven manufacturing technology is treated as peripheral rather than foundational—should inform how the next program is built from the start.