The RS-25 is a great engine with a terrific ISP. It was used 42 years ago (on April 12th 1981) for the first Space Shuttle flight; therefore, the most of the patents on it must have expired long ago. But NASA still seems to be willing to pay a handsome price for it...

For example, the current cost of manufacturing a new RS-25 main engine—which will be used for the Artemis V mission and onward—is about \$ 100 million. NASA and Aerojet are trying to achieve a 30 percent cost savings by the end of this decade, bringing the cost down to $ 70.5 million.


Is there anything that prevents other companies (besides Aerojet Rocketdyne) from manufacturing it, or a very close knock-off of it, at this point?

  • 1
    $\begingroup$ The thought of a bootleg RS-25s sold from out the door of a white van on a street corner amuses me. My uninformed guess is that they are making too many to do them bespoke as one off items, but not enough to have a proper production line where every system can be tuned up. $\endgroup$ Jan 12 at 7:03
  • 2
    $\begingroup$ Even if someone else wanted to and was able to create new RS-25s I doubt they could do it cheaper than Aerojet. Setting up the tooling would be a huge investment for an engine that's only going to have a fairly limited market $\endgroup$ Jan 12 at 9:22
  • 1
    $\begingroup$ Elon has Twittered that "Raptor cost is tracking to well under $1M for V1.0." If he's right, I can't imagine much of a market for RS-25's costing a hundred times as much. $\endgroup$
    – Woody
    Jan 12 at 10:08
  • 2
    $\begingroup$ Reliability and safety don't come from the design alone; the manufacturing process and quality assurance controls are a huge part of it. Certainly they're the key to demonstrating that reliability to skeptical customers. And they're neither easy nor cheap. $\endgroup$
    – Cadence
    Jan 14 at 10:25
  • 1
    $\begingroup$ The RS-25 is a fantastically complex engine that was never designed for cost effective production. It's also a hydrolox engine that's not optimized for vacuum and has about half the T/W ratio of more modern booster engines, so who's going to want a cheap one? This is like trying to use the engine from a Formula One racecar...from 1980...as a modern low-cost truck engine. $\endgroup$ Jan 15 at 14:54

1 Answer 1


The first key issue here is market. The flight history of Falcon Heavy vs Falcon 9 suggests there are not that many people who want to send single heavy payloads to orbit. Even if a hypothetical RS-25 clone manages to snipe the entire SLS program off Aerojet, the sales might only be 20-30 engines, and, if part of a re-usable system, unit sales are potentially even worse.

With access to a couple of museum specimens and the public domain documentation it would certainly possible for any manufacturer with suitable machinery to make an RS-25 shaped object.

The issue is that this object would have no meaningful relationship with the flown RS-25s in terms of expected reliability or performance needing a complete engine test series destroying engines until the details are pinned down. A naive test program could easily destroy more engines in testing than market exists for currently.

A hydrolox engine needs to handle extreme temperature variation and exciting chemistry from the cold soak startup cycle through ignition and operation and then through shutdown without cracking or having critical parts intersect with each other.

Expected issues would be:

Metal stock used to make the original engine are not easily available (and if they are, in metric sizes). In many cases that is because better options exist, but these may not be suitable for direct substitution without a clear understanding of the environment of an existing operational engine.

Finding out what happens inside an operating engine involves measurements, and, while the data from original testing may exist, the sensors used will have introduced errors due physics. A rebuild either needs to find and calibrate the original sensors or find ways to match new sensors against old data.

Actually reading the sensors is also a challenge. RS-25s used an integrated controller that would either need to be tightly replicated using new old stock, or redesigned using new hardware, losing any savings and reliability history. Failure of the controller will probably destroy the engine due to the criticality of the RS-25 shutdown sequence executing properly to avoid chemistry becoming physics.

Most metals change properties with heat cycles, so a CNC-carved or 3D-printed component will behave very differently from one welded from sub-assemblies.

Welding changes the properties of the material through direct heat, the alloy from the welding rod, and the gasses used. This detailed information is probably not documented and would need to be reverse-engineered from examples.

Casting and Forging parts will also change properties in ways that mean you need to know the shape of the original casting and the dies used to forge it to get a final product that matches the original performance AND shape.

These engines are physically large. Moving, cutting, and joining the parts will need jigs and other equipment, the design of which will be critical to reliable fabrication but not obvious from looking at the finished product.

Welded joins need to fully penetrate the material to be joined without compromising it through adjustment of current, pulse, rate, and distance. Achieving this normally involves multiple full-size test runs that get cut apart for inspection.

Surface coatings can protect or change materials, with some of these coatings being an 'accidental' side effect of a prior process step where the cleaning process (or failure to clean) may impact the following steps or finished part in terms of thickness, finish or corrosion resistance.

Oxygen is very hostile to metals and tends to burn them, given half a chance, making details of the final product matter to avoid running engine-rich without warning.

Hydrogen both penetrates through materials and makes metals brittle meaning some design choice outcomes will not become apparent until well after the initial test firing of an engine.

The quality control/testing information of the various sub-assemblies to allow destructive testing of proposed parts probably still exists, but if the design has changed to make the engine easier to build these tests will not be directly applicable (e.g. if alloys have changed X-rays of known good vs new welds cannot be compared).

With enough information it is probably possible to replicate exactly a working RS-25, it will, however, be MORE expensive than the originals were to build, and less reliable due to the need to use new old stock for things like the engine computer/controller and training staff in old techniques.

It is also quite possible to make an RS-25-derived engine, but since, from the testing/safety perspective, it would be a new design it might as well be a fully clean sheet design built using RS-25 (and other engine) data.

What is actually possible if you have access to lots of original records and multiple working RS-25s is to design new units and fit them to working old engines and compare behavior. This allows you to iterate until you have a 'new' RS-25 with minimal destructive failures and some sort of reliability history linked to the original. This is probably only possible by Aerojet.

The unsuccessful efforts to make a clone of the Russian RD-180 may be relevant.

  • 3
    $\begingroup$ This part of the answer nails the situation perfectly certainly possible for any manufacturer with suitable machinery to make an RS-25 shaped object . There is a very large amount of "trade secret" type information that goes into all phases of the manufacturing processes. This is not "patent protected" info but rather it is closely held build process recipe type stuff. $\endgroup$
    – BradV
    Jan 15 at 14:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.