goJumboGPT

Devices Looking after devices and the data on them

How long does an SSD keep data without power?

At least 1 year at 30°C for a worn out client SSD, under JEDEC's JESD218 standard. The full retention table in weeks, why the 7 days claim is wrong, and what to do.

8 min read How we write

The short answer

  • A client SSD, the kind in laptops and desktops, must keep its data for at least 1 year powered off at 30°C (86°F) under JEDEC's JESD218 standard, measured after the drive has used up its rated writes.
  • An enterprise SSD only has to keep data for 3 months at 40°C (104°F), mostly because its rating is measured at a hotter storage temperature.
  • In JEDEC's own retention table, every 5°C (9°F) rise in storage temperature roughly halves the time: 105 weeks at 25°C, 52 at 30°C, 26 at 35°C and 14 at 40°C for a client drive.
  • The 7 days claim comes from the extreme corner of that table, a completely worn out drive stored at temperatures like a hot car, not a drive in a drawer.
  • A lightly used drive keeps data much longer, and the popular per NAND type figures of 1 to 10 years have no published source.
  • Store spare drives cool, power them up and read everything at least once a year, and keep long term copies on a hard drive or in the cloud.

About a year, for an ordinary SSD that has been worn out and is kept at room temperature. That is the floor set by JEDEC, the semiconductor standards body: its JESD218 standard says a client SSD, the kind in laptops and desktop PCs, must still hold its data after 1 year powered off at 30°C (86°F), measured once the drive has used up its rated write endurance. An enterprise SSD only has to manage 3 months at 40°C (104°F). A lightly used drive kept somewhere cool does much better than either figure, and heat cuts the time sharply: in JEDEC's own retention table, every 5°C (9°F) rise in storage temperature roughly halves it. The viral claim that SSDs lose data in 7 days describes none of this.

Where the 7 days headline came from

In May 2015 Guru3D ran a story headlined "SSDs lose data if left without power for just 7 days". It said a drive could start to lose data after as little as a week, citing a presentation by Alvin Cox of Seagate, chairman of the JEDEC committee that wrote the SSD standards. The story has been recycled ever since.

The presentation is real, and so are short retention figures. Cox's 2012 Flash Memory Summit deck reproduces the retention table from JESD218A, and its shortest entry is two weeks: an enterprise drive, written at 55°C (131°F) until it had used every program and erase cycle it was rated for, then stored at 55°C. Curtiss-Wright, a defense electronics supplier, quotes even shorter numbers from JESD218 for a client drive stored hot: 500 hours at 52°C (126°F) and 96 hours, four days, at 66°C (151°F).

So the week is not invented. It is the far corner of a table, describing a drive that is completely worn out and stored at temperatures you meet in a car parked in summer sun, not in a desk drawer. Headlines stripped those conditions off, which is why the scary version persists.

JEDEC's retention table, in weeks

JESD218A defines two drive classes. Transcend and Curtiss-Wright both publish the client figures, and Cox's slides reproduce the table.

ClassUse while powered onStorage with power offMust keep data for
Client (laptops, desktops)40°C (104°F), 8 hours a day30°C (86°F)1 year
Enterprise (servers)55°C (131°F), 24 hours a day40°C (104°F)3 months

Behind those two lines sits Table C.1, expected retention in weeks at different storage temperatures. The columns below use each class's rated use temperature, so these are the numbers behind the headline figures.

Storage temperature, power offClient drive, used at 40°CEnterprise drive, used at 55°C
25°C (77°F)105 weeks, about 2 years101 weeks
30°C (86°F)52 weeks, the 1 year rating50 weeks
35°C (95°F)26 weeks25 weeks
40°C (104°F)14 weeks13 weeks, the 3 month rating
45°C (113°F)not listed7 weeks
50°C (122°F)not listed4 weeks
55°C (131°F)not listed2 weeks

Read down either column and the halving is plain: 105, 52, 26, 14. Guru3D attributed the same rule to Cox, that retention halves for every 5°C rise in storage temperature.

Read across and something less obvious appears: the two columns are almost identical. Its 3 month rating is short mostly because it is measured at 40°C storage rather than 30°C, and because, as Cox's slides put it, enterprise retention is set shorter to allow a longer working life in a machine that is maintained on a schedule. A laptop SSD left in a hot loft would land on the same 14 weeks.

Why a lightly used drive does much better

Every figure above is for a drive at the end of its rated life. JESD218 created the endurance rating printed on spec sheets as TBW, the terabytes a computer can write to the drive, and Cox's slides list end of life data retention among the requirements that define what that rating means. The kind of spec sheet number worth checking when you buy is covered in the laptop buying checklist.

Flash stores each bit as an electrical charge trapped in a memory cell. Curtiss-Wright describes the charge degrading over time, much faster at high temperatures, following the Arrhenius equation, which means exponentially as the temperature rises. Wear matters too. Cox's slide under the retention table says it directly: the numbers reflect data retention for NAND after 100% of its program and erase cycles, and less than 100% has much longer data retention.

A drive that has written a fraction of its TBW is, by definition, far from the end of life the table describes. Dell lists the same factors in its guidance for server drives: retention depends on how much rated endurance remains, the temperatures the drive ran at, and the temperature it is stored at.

What nobody publishes is a single figure for a half worn drive in a cool drawer. "Much longer" is the standard's own phrase, and any page that turns it into a precise number of years is guessing.

The QLC, TLC, MLC and SLC figures have no source

The other answer in circulation gives a figure for each type of flash: QLC, which stores four bits per cell, keeps data for about a year; TLC, three bits, for 3 years; MLC for 5; SLC, one bit, for 10. starline, a German storage company, gives the SLC, MLC and TLC figures with no source, and other pages repeat them.

No standard contains those numbers. JESD218 rates a drive by class and by the condition it must meet at the end of its endurance, not by cell type. A client QLC drive and a client TLC drive certified to JESD218 have to meet the same 1 year at 30°C after their rated writes. A per type table that ignores wear and temperature, the two things the standard says matter, cannot be relied on.

When these numbers do not apply

USB flash drives and memory cards. JESD218 covers SSDs. USB sticks and SD cards use the same kind of flash and lose charge for the same reasons, but they are not rated under it, and makers rarely publish a retention figure. There is no honest number to give, so treat them as a way to move files rather than a place to keep the only copy.

Drives that are not certified. The 1 year figure is the requirement a manufacturer claims to meet when it rates a drive to the standard. A cheap drive with no published endurance rating has made no such claim.

Hard drives. A spinning hard drive stores data magnetically, not as trapped charge, so this table says nothing about it.

Drives that are powered on. A connected SSD maintains itself. Dell explains that drive firmware runs background data retention tasks that refresh the charge in the cells, and that a powered off drive cannot do this. This article is about a spare drive, an old laptop in a cupboard, or a backup SSD that comes out once a year. And if that old laptop is on its way out, wipe it properly before you sell it rather than leaving it in a drawer with your data on it.

What to do with a drive in a drawer

Store it cool. Dell says powered off drives should be kept at no more than 40°C (104°F), and the table shows why cooler is better still. Heat is also what ages a phone battery fastest, so the same cool shelf suits both.

Power it up at least once a year, and while it is connected, read everything on it. Dell's advice for its server drives is stricter, power up every 2.5 months for at least 3 weeks, and it adds that a read of all used cells triggers the drive's retention tasks and is the better option for large drives. For a home drive, copying its entire contents to another disk, or running a backup tool's verify, reads every file and tells you at once if something has already gone.

Do not make an unpowered SSD the only long term copy of anything. Keep archive copies on a hard drive or in the cloud as well, following the 3-2-1 backup rule in real life,. An unplugged drive is also the copy that ransomware cannot reach, which is one more reason to keep it readable. For a family photo collection, the export step in a photo backup system that survives a lost phone is the part that should land on a disk you check every year. A cloud copy has its own rules: deleted files stay recoverable in the cloud for only 30 days on most services, so the cloud is a backup, not a safety net for things you delete.

If you are buying storage with this in mind, a laptop with modest storage plus an external drive and a cloud account is a sound arrangement, and the budget in whether 256 GB is enough for a laptop works through that trade off.

If an old drive has already sat for years, connect it, copy off what you need first, before running anything that writes to it, and check the files open.

Common questions

Do SSDs really lose data after 7 days without power?

No, not in normal use. The 7 days headline comes from a 2015 story about a JEDEC presentation. The shortest figure in the JEDEC table that presentation reproduced is two weeks, for an enterprise drive that is completely worn out and stored at 55°C (131°F). A client SSD is rated to keep data for at least a year at 30°C (86°F) after its rated writes are used up.

How long will an external SSD keep data in a drawer?

At least a year if it is a drive rated to the JEDEC client standard, even at the end of its write life, and longer if it is lightly used and kept cool. Retention roughly halves for each 5°C rise in storage temperature, so a drawer indoors is far safer than a car or an attic. Power it up and read its contents at least once a year.

Does plugging in an SSD for a minute refresh the data?

Not reliably. Drive firmware refreshes the charge in its cells with background tasks that need time, and Dell's guidance for its server drives calls for weeks powered on, or a read of all the stored data. For a home drive, the practical version is to copy everything off it, or run a full backup verify, which reads every file and shows you whether anything has already been lost.

Is a USB flash drive good for long term storage?

No. USB sticks and memory cards use the same charge based flash as SSDs, but they are not rated under the JEDEC SSD standard and makers rarely publish a retention figure, so there is no number you can rely on. Use them to move files, and keep long term copies on a hard drive or in a cloud backup, checked at least once a year.

Does QLC flash lose data faster than TLC?

Nobody has shown that it does in a way you can count on. The popular figures of 1 year for QLC and 3 for TLC come with no source. Under JESD218, a client drive of either type must keep data for at least a year at 30°C once its rated writes are used up, and wear and temperature matter more than cell type.