Voyager 1 launched in 1977 with onboard computers holding less memory than a single photo on a modern phone — and that 1970s machine is still running, sending data back to Earth from interstellar space

Imagine building something today — anything, a car, a phone, a piece of software — and having someone tell you it needs to still be working perfectly in 2074.

Not just switched on. Working. Communicating. Producing useful output. Fifty years from now, on the far side of half a century of technological change, still doing the thing it was originally built to do — while running on its own power, having received no physical maintenance for the entire duration, in an environment more hostile than anywhere on Earth.

Almost every engineer alive today would tell you that this is impossible. Modern hardware simply is not designed to last that long. Modern software is not written to last that long. The idea of a smartphone made in 2026 still functioning normally in 2076 would be laughable, and everyone in the industry knows why. Planned obsolescence. Component miniaturisation past the limits of long-term reliability. Software dependencies that will be dead in a decade. The whole architecture of modern technology is built around a much shorter lifespan than fifty years.

And yet, right now, 25 billion kilometres from Earth, a machine built in 1977 is doing exactly that.

The specification

Voyager 1 launched on September 5, 1977. Its onboard computing hardware — three separate computer systems working together — totals a combined 69.63 kilobytes of memory. Not gigabytes. Not megabytes. Kilobytes.

The number is worth staring at. A single photograph taken on a current smartphone is several megabytes in size — thousands of times more data than Voyager 1’s entire computing capacity. A standard low-resolution JPEG file, the kind of image you might attach to an email without thinking, holds more information than the total memory of the spacecraft that has travelled further from Earth than any other human-made object in history.

Scientific data collected by the probe cannot be stored on standard memory because there isn’t enough of it. Instead, everything is recorded onto a physical 8-track magnetic tape and transmitted back to Earth as quickly as it can be processed. Old data is written over the moment it has been sent. There is no archive on the spacecraft. There is barely enough working memory to hold the current instruction set.

This is the machine that is, right now, sending measurements of the interstellar magnetic field back across 25 billion kilometres of space.

Why 1970s hardware outlasts modern hardware

The obvious framing — that Voyager 1 has survived despite its primitive computing — is exactly backwards. Voyager 1 has survived, in significant part, because of it.

Consider what 69 kilobytes of memory actually forces on an engineer. There is no room for bloat. Every byte in the operating system is there for a specific reason and has been reviewed line by line. There are no dependencies on external libraries, because there was no external anything. There are no automatic updates from a server that will one day stop existing. There is no operating system in the modern sense at all. The software is small enough that a single engineer could hold the entire program structure in their head at once, and the hardware is simple enough that individual components can be reasoned about, tested, and validated exhaustively.

The 1970s design philosophy at NASA also assumed that repair would be impossible. Everything critical had to be duplicated. Voyager 1 carried two of every essential system, so that if one failed, the other could take over. The Flight Data System had a redundant unit. The command computers had backups. The radios were paired. The spacecraft was, in Voyager project manager Suzanne Dodd’s phrase, “designed with nearly everything redundant.”

This kind of engineering is expensive. It doubles the weight of the spacecraft. It doubles the parts count. It doubles the risk of manufacturing defects. Modern hardware has largely stopped doing it, because modern hardware is usually within reach of a technician. If a modern satellite fails, engineers can send commands to reboot it, patch it, replace failing components with new ones. Voyager 1 will never be within reach of anyone. It has to work with what it was launched with — forever — and everything on board was designed with that assumption baked in.

The result is a machine that has now operated flawlessly for forty-nine years, in the environment of interstellar space, on hardware that a modern engineer would consider embarrassingly underpowered. It has done this by refusing every modern engineering shortcut.

What’s still running, and what isn’t

Voyager 1 launched with ten scientific instruments. As of 2026, only two remain switched on: the magnetometer and the plasma wave subsystem. The others have not failed. They have been deliberately powered down over the years to conserve the spacecraft’s declining electrical supply.

The power source is a set of radioisotope thermoelectric generators — small nuclear devices that generate electricity from the heat released by decaying plutonium-238. At launch, they produced about 470 watts. Today they produce roughly 230 watts and are losing about four watts per year as the plutonium slowly runs out.

That declining power budget, not any hardware failure, is what will eventually end the mission. NASA’s engineers have been working through a careful sequence of shutdowns — heaters, non-essential instruments, backup systems — trying to preserve the minimum power required to keep the primary science instruments and the radio transmitter running for as long as possible.

In 2024, the spacecraft went through a genuinely alarming period. A failed memory area in the flight data subsystem produced months of unusable data. NASA engineers, working across a round-trip communication delay of nearly two days, wrote and uploaded a workaround. It succeeded. Voyager 1 recovered. Old machines, run by careful people, tend to.

What actually matters about this

The story of Voyager 1 is often told as a piece of trivia — the little computer that could, primitive hardware punching above its weight, an amusing footnote in the history of computing.

The actual lesson is less amusing and more useful. Voyager 1 is a working demonstration of what engineering can produce when it is optimised for genuine long-term reliability, when redundancy is treated as a requirement rather than a luxury, when engineers assume there will never be a chance to fix what breaks. Almost nothing built today is engineered to that standard. Almost nothing has to be. But when something does have to be — when the object in question will be sent somewhere no repair crew can reach, and needs to keep working for decades — the principles that got Voyager 1 to interstellar space are the ones that still apply.

They are not principles about small memory, or slow processors, or 8-track tapes. Those are surface features. The underlying principles are about designing for the assumption that everything you make will be alone with itself, forever, with no help coming. About making sure the thing works before you launch it, and expecting no updates afterward.

Somewhere out beyond the heliosphere, at this moment, a 49-year-old machine is quietly doing what it was built to do. Not because it was clever. Because it was built to.