Linux Flies into Space

Linux, open source software, and commercial off-the-shelf (COTS) hardware have reshaped space technology. They have replaced the one-off, proprietary programs and specialized hardware typical of older space-borne systems. The change suggests faster turnarounds and more innovation opportunities — a transition with long-term implications for anyone interested in space exploration. 

Linux’s first foray into space was in 1996, just five years after the operating system was introduced. That year, a Digital UNIX application was ported to a Debian-Linux–based IBM ThinkPad laptop, which was carried aboard a Space Shuttle flight.

The adoption of Linux in space has not always been smooth. For instance, NASA budget problems meant a 2000s experiment for carrier-grade Linux on Honeywell’s “Dependable Multiprocessor” did not get off the ground. The project intended to prove that it was possible to get supercomputer performance in orbit, using a cluster of COTS PowerPC boards and field-programmable gate arrays (FPGAs) without relying solely on traditional radiation‑hardened CPUs. 

Desktop Linux returned to space in 2013, when NASA moved the Windows laptops on the International Space Station (ISS) to Debian and Scientific Linux for astronauts and ground teams. That change made it easier for the operating system to find acceptance in space programs, and it’s been part of space architecture ever since. Now, with thousands of satellites in Starlink's constellation and NASA’s Ingenuity helicopter on Mars running Buildroot Linux on a Qualcomm Snapdragon processor, Linux has become the most popular operating system in the known universe.

It may be ready to take off — again. In a big way.

The Space Trajectory Toward COTS

Much of the hardware in space is legacy equipment. The ISS, for example, runs the station using six Command-and-Control Multiplexer/DeMultiplexer computers; their processors are 20 MHz Intel 80386SX CPUs from 1988. The New Horizons probe, which zipped by Pluto, is powered by a 12 MHz Mongoose-V CPU based on the MIPS R3000.

One reason that space programs use outdated hardware is that, per NASA’s rules, computer components must be radiation hardened, especially the CPUs. These customized processors undergo years of design and testing before they are certified for spaceflight. As a result, certified hardware is far from the most recent generation. The fastest radiation-hardened (rad-hard) CPU today is the 800 MHz PIC64‑HPSC RH, based on the RISC-V architecture. It’s been in development for eight years and even now is only available in sample quantities. So, these applications run relatively slowly.

Not everyone has bought into the idea that only rad-hard equipment is acceptable. In 2017, HPE successfully put a COTS server in low earth orbit (LEO) on the ISS, and it ran without trouble for more than a year. That project’s success got attention from engineers — and from those who control budgets.

Old Space vs. New Space

These two movements, toward Linux and toward COTS, are coming together. 

The aerospace industry’s attitudes are splitting between “Old Space” and “New Space,” says Rob Woolley, a senior principal technologist at Wind River who works on Space Grade Linux. The two attitudes will coexist, but the New Space viewpoint is making inroads.

In the Old Space model, engineers design systems years in advance using a waterfall development model, avoid open source, and run software on rad‑hard hardware. (Woolley would know. Wind River has decades of experience supporting space missions, in addition to its current projects.)

What doesn’t scale, argues Ramón Roche, general manager of the Dronecode Foundation, is the way many missions are still engineered as bespoke, one‑off software stacks. “It’s 2026, and we’re still in a phase like 1969,” where missions are one‑off and expensive. That model was fine back when launches cost millions of dollars, he notes, “but it no longer fits the economics of New Space.”

The New Space model embraces COTS processors and open source tech stacks, and it uses software redundancy to manage risk.​ In New Space, Woolley says, “People are embracing Linux and the Go language for development, as well as toying with the idea of using containers as both the delivery module and an isolation system.”

The Linux Foundation’s Space Grade Linux initiative aims to create common distribution and tooling that space missions can share, as opposed to everyone building bespoke kernels.​​

For example, SpaceX’s Falcon 9 rocket runs on Linux using a trio of dual-core 1.6 GHz x86 processors, as does the Dragon passenger-carrying spacecraft. 

Why three sets of processors? In place of expensive rad-hard processors, the Falcon 9 uses a triple‑redundant architecture to avoid errors. If one CPU’s answer disagrees with the other two, the majority wins. This enables inexpensive COTS hardware to be almost as error-resistant as customized radiation-proof processors.

Instead of relying only on rad‑hard chips such as the classic RAD750 PowerPC, New Space designs use current systems-on-chip. They compensate by adding error detection, voting schemes across redundant boards, and supervisory logic that can catch and correct faults in flight.​

A common software stack is important. Dronecode’s Roche explains that Space Grade Linux should help space companies stop “competing on the plumbing” and start sharing core software infrastructure.

The cost is “quickly approaching less than $100 per kilo to ship payload to space,” Roche says. That fuels a flywheel of more launches, more applications, and even data centers in space.

One-off and Liftoff

There’s an interesting discussion happening between the Old Space and the New Space folks. “New Space definitely wants to use an immutable image and be able to deploy their applications, possibly in a container,” Woolley says. “But the Old Space folks are very much focused on traditional embedded Linux,” which is not tuned for space applications.

The Old Space community does not use embedded Linux the way most Earth-bound developers would. Instead, Woolley explains, the developers build customized stacks with a minimized footprint. “They go so far as to replace [Linux components] SysVinit or Systemd with the main application that they’ve written to manage all of the services on the device.”

The New Space community wants to embrace different COTS architectures. And rather than using specialized hardware and software, Woolley says, they’re trying to use available open source software.

So, what does this mean for Linux and space? Woolley believes “the Old Space method will continue to be used for very risky space missions going forward,” such as missions to Mars and beyond. “Space missions will continue to keep using the Wind River VxWorks® real-time operating system in space because it’s extremely predictable, certifiable, and proven over decades of missions,” he says. “While Linux is powerful and growing in noncritical roles, its complexity, variability, and faster evolution make it harder to guarantee strict real-time behavior or certify for flight. When a spacecraft has to operate flawlessly for decades with no updates, the safest choice is an operating system with the longest track record.”

However, Woolley adds, “We’ll start to see Space Grade Linux being adopted for things like CubeSats or devices used on the space station and replacing Windows. All this will enable more people to participate in scientific research and try new experiments in LEO.”

In short, the new space era will finally give the penguins a permanent home in orbit. Instead of being experimental, Linux and COTS will become the standard operating environment for space software well into the 21st century, and — who knows — possibly beyond.

After all, we know that LCARS is Star Trek’s preferred graphical interface. Could L stand for Linux? Maybe!