Voyager 1 carries six flight-computer units designed in the early 1970s, and NASA’s own accounting puts their combined memory at about 68 kilobytes. That is less data than many single photographs compressed by a modern phone.
The comparison is accurate only after one important qualification. Those 68 kilobytes are not a shared storage drive. They are divided among three specialised computer systems, with two units of each type, and the machines organise memory in 16-bit and 18-bit words rather than the eight-bit bytes used to describe phone storage.
Nearly 49 years after launch, Voyager 1 is travelling through interstellar space. It crossed the heliopause, the outer boundary of the Sun’s particle-and-magnetic-field bubble, in August 2012. Its original computer architecture still accepts instructions from Earth, although some redundant hardware has failed and most of the probe’s science instruments have been switched off.
Six units make three different computer systems
NASA’s Voyager frequently asked questions lists three computer types aboard each spacecraft, with two units of every type. The Computer Command System, or CCS, interprets commands, runs stored sequences and handles fault protection. The Flight Data System, or FDS, formats science and engineering information for transmission. The Attitude and Articulation Control System, or AACS, manages orientation and helps keep the high-gain antenna directed towards Earth.
That is the source of the six-computer description: two CCS units, two FDS units and two AACS units. They were paired partly for redundancy, but they were never six interchangeable versions of one machine.
An FDS could not assume the AACS’s antenna-pointing work as though it were a spare laptop. Each system had a defined role, its own design and its own memory. The SETI Institute’s Voyager spacecraft guide describes them more aptly as interconnected electronic brains, able to accept precoded sequences containing thousands of instructions for autonomous operation.
The word “six” also describes the hardware Voyager 1 carried, not six fully available computers today. The backup FDS failed in 1981, Voyager project manager Suzanne Dodd told Ars Technica in reporting republished by Wired. The surviving FDS therefore has no equivalent standby unit.
How six separate memories add up to 68 kilobytes
NASA describes each CCS processor as an 18-bit machine with 4,096 words of non-volatile plated-wire memory. Each AACS unit also has 4,096 words of 18-bit memory. The two FDS machines use 16-bit words and modular memories of approximately 8,200 words apiece.
Add all six capacities and the result is roughly 69,600 eight-bit bytes. Divide by 1,024, the convention behind a binary kibibyte, and it comes to about 68 KiB. NASA uses the more familiar label “68KB” for the rounded total.
The arithmetic is sound, but it does not create one addressable pool. Code running in the CCS could not claim unused AACS memory as extra room. Adding the figures is like totalling the cupboards in several locked workshops: every cupboard holds something, but each belongs to a different room.
“Storage” can cause a second misunderstanding. The 68-kilobyte figure refers to memory used directly by the flight computers for instructions and working data. Voyager also carries an eight-track digital tape recorder. A NASA backgrounder prepared for the 1981 Saturn encounter gave that recorder a reusable capacity of about 536 million bits, equivalent to roughly 67 megabytes. It buffered observations when they could not be transmitted immediately.
A phone photograph remains a fair illustration of scale. A one-megabyte compressed image contains about 15 times as many bytes as the six-computer total, and phone photographs are often larger than that. File sizes vary with resolution, image format and compression, so this is a comparison with a typical file, not a fixed rule about every phone image.
Why so little memory could run a planetary mission
Voyager’s computers were not general-purpose consumer machines expected to run a display, apps, wireless networking and a computational-photography pipeline. They performed a narrow set of spacecraft functions defined before launch.
The CCS received commands and stored sequences. The FDS collected, formatted and encoded telemetry. The AACS read sensors and issued instructions to thrusters and mechanisms. Navigation calculations and most scientific analysis could remain on Earth, where ground computers and mission teams had far greater resources.
That division of labour explains more than the memory number. A phone keeps software and data for thousands of possible tasks because its owner may ask it to do almost anything. Voyager had fewer jobs, no graphical interface and no need to carry a library of consumer software.
Its narrow purpose did not make it inflexible. The flight team could upload new sequences and alter parts of the software, allowing both Voyagers to adapt after their planetary encounters and respond to failures the original designers could not specify in advance.
A failed chip exposed the limits of the surviving memory
That ability was tested in November 2023, when Voyager 1 stopped returning readable science and engineering data. The probe continued to receive commands and transmit a radio carrier, but its FDS produced an unusable repeating pattern instead of properly packaged telemetry.
In April 2024, NASA’s Jet Propulsion Laboratory reported that one chip holding part of the FDS memory, including flight code, was no longer working. Engineers could not repair the chip. They divided the affected code into sections, moved the pieces into separate surviving locations and revised references elsewhere in the software so the fragments would still operate together.
No single free region was large enough to hold the complete block.
IEEE Spectrum’s reconstruction of the repair reported that the damaged region held 256 bits. Engineers searched for space occupied by processes the probe no longer required, including data modes created for the much faster transmissions possible during the planetary encounters.
The first rearranged section was transmitted on 18 April 2024. At the time, a radio command needed about 22.5 hours to reach Voyager 1, followed by another 22.5-hour wait for the response. JPL received readable engineering data on 20 April for the first time in five months.
The repair added no memory and restored no failed hardware. It fitted essential code around a dead section of the only remaining FDS by using small spaces that were still reachable.
Only two science instruments remain on
Computing capacity is no longer Voyager 1’s only tight constraint. Its radioisotope thermoelectric generators produce less electricity as their plutonium fuel decays, requiring controllers to switch off heaters and instruments in a planned sequence.
NASA’s instrument-status table, updated on 17 April 2026, lists only the magnetometer and plasma-wave subsystem as switched on aboard Voyager 1. The cosmic-ray subsystem was turned off in February 2025, and the low-energy charged-particle experiment followed in April 2026.
The surviving computers must still accept instructions, maintain attitude, manage fault protection and package the limited observations that continue to reach Earth. The hardware doing that work was built before the first mass-market personal computers appeared.
The 68-kilobyte comparison is best read as a description of deliberate engineering, not a contest between Voyager and a phone. The total is tiny by present consumer standards, but it was divided among machines built for a short list of specific jobs. Nearly five decades later, the open question is no longer whether that architecture was sufficient. It is how long declining electrical power and ageing components will let the remaining systems keep doing those jobs.
