Summary
Storage feels reliable, dependable, and permanent. It isn’t. You copy your files over, unplug the drive, and assume everything will be there when you come back. Sometimes it is. But every kind of storage loses data over time, and how fast depends on what the drive is made of, how worn it is, and where it’s been sitting. Flash storage holds data as trapped electrons But that means they can be a bit leaky SSDs and USB sticks both use NAND flash, which stores each bit as an electrical charge locked inside a microscopic cell. The cell is wrapped in insulation, but the insulation isn’t perfect, as, over time, the electrons basically leak out. That causes the cell voltages to drift, directly affecting the drive’s ability to hold information once unpowered. Cell design changes how quickly this process occurs, but it’s not always as simple as “more expensive hardware holds data longer.” There are four main types of flash memory, the type of which defines how long your drive will hold onto data:
- Single-Level Cell: SLCs deliver one bit per cell and hold data the longest, but you don’t find this in consumer drives
- Multi-Level Cell: MLCs increase to two bits per cell but are a step down from SLCs; still, they’re not generally found on the consumer market unless you happen upon some enterprise equipment or similar.
- Triple-Level Cell: TLC memory increases to three bits per cell and delivers higher write rates and endurance than QLC.
- Quad-Level Cell: QLC memory stores four bits per memory cell, but has the lowest endurance, write speeds, and capacity to store data long term. Most consumer SSDs use TLC and QLC because they’re cheaper and deliver higher capacities— the two main features everyone wants. Flash memory will eventually wear out It’s all down to the age I’ve amassed a huge number of storage drives over the years. I’ve even got some of those throwaway 512MB-2GB USB thumb drives given to me on a course 15 years ago. When I plugged one in recently, the original files were still sitting there, completely accessible. I can’t imagine this drive cost more than a buck, if that, back then. In that, some drives are surprisingly resilient, because drive and bit rot are affected by more than just the type of SSD you buy. The wear rate, environment, and storage make a big difference, too. Furthermore, the insulation used on most drives isn’t perfect, and wear reduces its efficiency further. Every write-and-erase cycle pushes electrons through that layer, and each pass damages it slightly. Damage gives electrons a path out, and the margin is thin: one research survey puts a cutting-edge cell at only about 100 electrons, so losing a few can change what the drive reads. Still, it all has an impact. Similar drives, consistent environment For example, in 2020, Zachary Vance bought ten Kingston DataTraveler SE9 32GB drives, filled them with random data, and set a schedule to test more of them every few years, rewriting the good ones afterward. His latest update, at the start of 2026, revealed that he’d tested five drives and found zero bit rot. Great stuff, right? Vance likely bought all the drives at the same time, and they’re stored in a dry, consistent climate. Similar drives, different environments Another Redditor, Carnildo, copied a bunch of data to three cheap USB drives, and stored them in different places to track climate sensitivity and data loss. One in the attic exposed to the elements, two in a climate-controlled room, and one of those climate-controlled drives periodically plugged into a device to read the data. After one year:
- On the attic drive, 138 images were corrupted
- On the indoor passive drive, 773 images were corrupted
- On the indoor active drive, 6 images were corrupted
In nearly all cases, corruption involved entire 4KB write blocks being completely or nearly-completely randomized. Visually, this results in the image being truncated somewhere within the corrupted block. In only one case did the corruption take the form of a single flipped bit and a stripe of distorted colors.
Similar drives, different write and wear levels
Of the three tests, YouTuber HTWingNUt’s unpowered SSD test is probably the most interesting, as it shows that the wear level directly impacts the ability of a flash cell to hold data long term.
He loaded four 128GB TLC SSDs with 100GB of random data, and verified it all with an MD5 hash to track any changes or corruption. Then, two SSDs were left with the original 100GB, while the other two were “torture-tested” by writing 280TB to the drive, massively exceeding their official 60TBW rating.
The drives were then stored in anti-static bags in a climate-controlled room, before being tested once per year.
After one year, all of the drives passed every data retention test and MD5 hash verification. After two years, one of the worn drives had four corrupt files, and verifying its data took more than 42 minutes, instead of 10 minutes from the previous year.
It’s a small sample size, but telling that the harder the drive has been pushed, the more likely it was to experience corruption.
There’s actually an official rule on how much data a drive is allowed to forget
Turns out drives do what they want
Among all of this testing, memory standards leader JEDEC has specific guidelines on drive data loss per year.
JESD218 [PDF - login required] states that consumer SSDs only need to retain data for a single year when stored at temperatures of 30°C/86°F and powered off (rising to 40°C/104°F in enterprise settings).
If that temperature drops, data retention time increases. If it increases, your data cooks and leaks.
Those guidelines were first published in September 2010, but have been maintained through each update of the JESD218 standard to the current day.
Hard drives also forget slowly, but have more ways to lose your data
Magnetic platters are great for long-term storage
Where flash-based drives slowly start to leak data, older hard drives with the classic magnetized spinning disc can lose their drive orientation, essentially forgetting where your data is.
Magnetized regions are sensitive to temperature changes and external magnetic interference, both of which can wipe a drive instantly. And don’t forget the cosmic rays, because they can also destroy data in the right circumstances.
Now, HDD error correction processes and defragmentation can flip a bit or two in the right direction and piece together broken data in parts. But once a big data patch is destroyed, it can be very difficult to recover.
Still, the biggest issue facing most HDDs isn’t actually data leakage. It’s years sitting in storage without power, causing the drive bearings to seize. If the drive can’t spin, the magnetic arm can’t read the data, and you’ll be heading to the nearest data recovery specialist.
Seagate BarraCuda 8TB - Storage Capacity
- 8TB
- Brand
- Seagate
- Transfer rate
- Sustained transfer rate OD: 190MB/s
- Warranty
- 2-year limited The Seagate BarraCuda 8TB (ST8000DM004) is an internal SATA 6 Gb/s hard drive for desktop PCs. It is aimed at everyday storage, gaming, and general computing, from music and video to photo editing, with a sustained transfer rate of 190MB/s, and a 2-year limited warranty. So, what drive should you trust with your data? The biggest problem is that there isn’t a single “best drive” for long-term data storage. In September 2026, we saw a 443,000-drive peer-reviewed study claim a winner between Seagate and Western Digital in an enterprise backup situation… and actually, HGST came out on top (Western Digital bought Hitachi’s drive manufacturing in 2012, then closed it down). Samsung wasn’t included because BackBlaze, the backup company, doesn’t use its drives in its arrays. Not because it makes poor quality drives, but because it doesn’t make drives in high-enough capacities. Other testing has found Samsung drives to last years, pushed way beyond original write capacities without failure. Most folks don’t really need to worry a great deal about this, either. Just keep your drives in a dry, relatively cool but consistent environment, and perhaps power them on every now and then to make sure nothing is going wrong. The biggest issue to avoid is keeping a single copy of important data on a single drive. The old 3-2-1 data backup rule is still very much accurate — but make sure you actively check that your backups work periodically; otherwise, it’s effectively pointless.