You buy a 1 TB drive. Windows says 931 GB. It feels like being short-changed by seven percent, and there have been class-action lawsuits about it. The drive is exactly the size advertised.
931 GB to TB
1.00 TB
The 931 your computer shows is really 931 GiB, so it is the full terabyte you paid for — 0.909 TiB, or 999,653,638,144 bytes. Read as a literal decimal figure instead, 931 GB is 0.931 TB.
| On the box | Windows shows | Bytes |
|---|---|---|
| 1 TB | 931 GB | 1,000,000,000,000 |
| 2 TB | 1863 GB | 2,000,000,000,000 |
| 4 TB | 3725 GB | 4,000,000,000,000 |
| 8 TB | 7451 GB | 8,000,000,000,000 |
| 16 TB | 14901 GB | 16,000,000,000,000 |
Two ways to count
Every other measurement prefix means a power of a thousand. A kilometre is 1,000 metres. A kilogram is 1,000 grams. This is the SI system and it is unambiguous everywhere except computing.
Computers address memory in powers of two, and 210 is 1,024 — close enough to a thousand that early engineers borrowed the prefix. A “kilobyte” became 1,024 bytes. It was a convenient approximation, and at 2.4% off nobody minded.
The trouble is that the error compounds at every step:
At kilobytes it was a rounding error. At terabytes it is a tenth of your drive, and it keeps growing: at petabytes the gap is 12.6%.
What actually happened to your drive
The manufacturer sold you exactly 1,000,000,000,000 bytes. That is one terabyte under the SI definition, the same definition that governs every other use of the prefix “tera” in science and engineering.
Windows divides that number by 1,024 three times, arriving at 931.32. Then it writes GB beside it.
That is the entire problem. The number is correct. The unit label is not — what Windows has calculated is 931.32 gibibytes, and it is calling them gigabytes.
The names nobody uses
The International Electrotechnical Commission settled this in 1998 by giving the binary units their own names: kibibyte, mebibyte, gibibyte, tebibyte — KiB, MiB, GiB, TiB. Each is “kilo binary”, “mega binary” and so on, contracted.
The standard is unambiguous and correct, and it has been almost entirely ignored outside technical documentation. Linux tools use it. Some file managers use it. Windows, after more than twenty-five years, does not.
The names are admittedly awkward to say. That is a poor reason to keep a genuine ambiguity in the measurement of something people buy by the terabyte.
Who uses which
Decimal — powers of 1,000: drive and USB stick manufacturers, network speeds, macOS since 10.6, most Linux file managers, and anything quoting a transfer rate.
Binary — powers of 1,024: Windows file sizes and drive capacities, RAM of every kind, and most programming languages' internal reporting.
This is why the same drive reads 931 GB on Windows and 1 TB on a Mac. Neither is malfunctioning; they made different choices, and only one of them labels the result accurately.
RAM is the interesting exception. Memory really is manufactured in powers of two, because of how address lines work, so a “16 GB” module genuinely contains 17.18 billion bytes. For RAM the binary interpretation is not just a convention. It is the physical truth.
The other factor of eight
While we are here: your 100 Mbps internet connection does not download at 100 MB/s, and this is a different confusion entirely.
Mbps is megabits per second. MB/s is megabytes per second. There are eight bits in a byte, so 100 Mbps is 12.5 MB/s at the theoretical maximum. A 1 GB file takes at least 80 seconds, not 10.
The distinction lives entirely in the capitalisation of one letter — lowercase b for bits, uppercase B for bytes — and it is a factor of eight. Network engineers measure in bits because that is how data moves on a wire. Everyone else thinks in bytes because that is how files are stored.
Combine both confusions and you can be out by a factor of nearly nine while every number involved is technically correct.
What to do about it
Nothing, mostly. Knowing the reason removes the feeling of being cheated, which is the main cost of not knowing it.
When buying storage, assume you will see about 7% less than the label for a drive measured in gigabytes and about 10% less in terabytes. A “2 TB” drive shows as 1.81 TB. That is the whole adjustment.
When a specification matters — a backup that must fit, a quota, a file size limit — convert explicitly rather than assuming. The converter lists decimal and binary units in separate groups precisely so the two cannot be confused.
The short version
Your drive holds exactly what it says. Windows divides by 1,024 and labels the answer GB when it means GiB. macOS divides by 1,000 and labels it correctly. The gap is 7% at gigabytes, 10% at terabytes, and it is a naming failure rather than a missing-storage problem.