The most distant object humans have ever built runs on computers with less memory than a typical phone photograph.
According to IEEE Spectrum, each of Voyager 1’s main computer systems has about 69.63 kilobytes of memory, roughly comparable to a small, heavily compressed thumbnail. Nearly five decades after launch, that 1970s hardware still receives and carries out commands from Earth. It is a figure that complicates the usual assumption that difficult problems always demand more computing power.
A note on framing: we are writers and journalists, not aerospace engineers or mission specialists. What follows is our reading of NASA’s reporting and technical coverage of the Voyager programme, not an independent engineering assessment.
What Voyager 1 is actually running
The spacecraft does not rely on a single computer. It has three principal computer systems, all designed in the 1970s, each responsible for a different part of the mission.
The Computer Command Subsystem interprets instructions from Earth and manages spacecraft sequencing and fault protection. The Flight Data Subsystem collects and packages science and engineering data before transmission. The Attitude and Articulation Control Subsystem controls the spacecraft’s orientation and helps keep its antenna pointed towards Earth.
Together, these systems run the mission with memory measured in kilobytes rather than gigabytes.
The architecture also included redundant hardware, allowing another unit or component to take over when parts of the spacecraft failed. That backup design was deliberate. Suzanne Dodd, the Voyager project manager, has described a spacecraft that “was designed with nearly everything redundant. Having two spacecraft — right there is a redundancy.” Major systems and many critical components were duplicated so that the mission could continue after a failure.
But that redundancy has limits. Voyager 1 originally carried two Flight Data System computers. Its backup FDS failed in 1981, leaving the spacecraft dependent on the remaining unit.
Anyone who has kept an ageing car running long past its expected life by working around failed parts knows the small version of this. Scale it up to nearly five decades and more than 24 billion kilometres, and you have the operating reality of the most remote machine ever made.
Five months to restore engineering data
On 14 November 2023, Voyager 1 began sending back a repeating pattern of meaningless ones and zeros rather than usable science and engineering information. The spacecraft was still receiving commands and appeared otherwise healthy, but its Flight Data System could no longer properly package the information it was meant to transmit.
A diagnostic command sent in March 2024 produced a memory readout that allowed engineers to identify the damaged area. NASA concluded that about 3 percent of the FDS memory had been corrupted, probably because a single memory chip had failed.
The agency said the chip may have been damaged by an energetic particle from space or may simply have deteriorated after more than 46 years of operation.
The repair ran into the same constraint that makes Voyager remarkable. Linda Spilker, the Voyager project scientist, put it plainly: “The size of the memory was the biggest challenge in this anomaly.”
The failed chip contained part of the software responsible for packaging data, but there was no single undamaged section of memory large enough to hold all of that code. Engineers therefore divided it into smaller pieces and searched for unused sections of memory where each piece could be stored.
They also freed space by identifying processes and data modes the spacecraft no longer needed. Some had been created for parts of the planetary mission that ended decades earlier.
The first relocated section of code was transmitted on 18 April 2024. It reached Voyager about 22.5 hours later. On 20 April, the spacecraft returned readable engineering data for the first time in more than five months.
Consider what that required. A team on Earth diagnosed a probable hardware failure in a computer more than 24 billion kilometres away, then rewrote its software to route around the damaged memory. Each meaningful test took almost two days to confirm, and there was no possibility of physically examining or touching the machine.
Built for the unknown, not the expected
It would be easy to turn Voyager into a simple story about how engineers built things better in the 1970s. The more careful reading is about a way of thinking rather than a single piece of hardware.
John Casani, the Voyager project manager through launch, has said that the team “didn’t design them to last 30 years or 40 years, we designed them not to fail.”
That distinction matters. Designing something not to fail means assuming that individual parts eventually will. It means creating alternative routes, duplicated systems and enough flexibility for engineers to respond to problems that cannot yet be predicted.
That flexibility is exactly what the 2024 repair used. The engineers who divided and relocated the damaged software were using small margins of memory left by people who could not have known what the eventual emergency would be. They knew only that failures were possible and that the spacecraft needed room to survive them.
Voyager is now so old that every intervention carries increasing risk. Dodd has said that “all the decisions we will have to make going forward are going to require a lot more analysis and caution than they once did.”
What the repair showed is not that Voyager’s hardware is invincible. Parts have failed before, and more will fail as the spacecraft continues to age.
What it showed is that a group of engineers in the 1970s built enough redundancy and flexibility into machines with only kilobytes of memory for another group of engineers, nearly half a century later, to work around a failure none of them could have predicted, aboard a spacecraft no human being will ever touch again.