| Kibibits (Kibit) | Exaoctets (Eo) |
|---|---|
| 1 Kibibit | 1.28 × 10-16 Eo |
| 2 Kibibits | 2.56 × 10-16 Eo |
| 3 Kibibits | 3.84 × 10-16 Eo |
| 4 Kibibits | 5.12 × 10-16 Eo |
| 5 Kibibits | 6.4 × 10-16 Eo |
| 10 Kibibits | 1.28 × 10-15 Eo |
| 20 Kibibits | 2.56 × 10-15 Eo |
| 25 Kibibits | 3.2 × 10-15 Eo |
| 50 Kibibits | 6.4 × 10-15 Eo |
| 100 Kibibits | 1.28 × 10-14 Eo |
| Reference | Kibibits (Kibit) | Exaoctets (Eo) |
|---|---|---|
| A plain text message (160 characters) | 1.25 Kibit | 1.6 × 10-16 Eo |
| A three-minute MP3 | 23437.5 Kibit | 3 × 10-12 Eo |
| A smartphone photo | 31250 Kibit | 4 × 10-12 Eo |
| A high-definition film | 31250000 Kibit | 0.000000004 Eo |
| A dual-layer Blu-ray disc | 390625000 Kibit | 0.00000005 Eo |
The kibibit is a unit of digital information equal to 1,024 bits. Its symbol is Kibit. It is the first of the binary prefixes, a set of units defined by the International Electrotechnical Commission in 1998 to end a confusion that had run through computing since the 1960s.
The problem was straightforward. Computers address memory in powers of two, so memory came in chunks of 1,024 rather than 1,000. Engineers borrowed the metric prefix kilo for that quantity because 1,024 is close to 1,000, and for small numbers the approximation was harmless. But storage and transmission counted in true thousands, so the same prefix meant two different things depending on which part of the machine was being described.
The IEC's solution was to coin new names. Kibi is a contraction of kilo binary, and the pattern continues with mebi, gibi, tebi, pebi, exbi, zebi and yobi. Each is 1,024 times the one below, and each symbol takes the form of a capital letter followed by a lowercase i: Ki, Mi, Gi, Ti and so on. The kilobit then means one thousand bits and nothing else.
A kibibit is 128 octets, and the gap from a kilobit is 2.4 per cent. That small difference is why the two were confused for so long: at this scale nobody notices. The error compounds by 2.4 per cent at every step, reaching 5 per cent at the mebibit, 7 per cent at the gibibit and 21 per cent by the yobibit, which is where the ambiguity became genuinely expensive.
Adoption has been partial and uneven. Standards bodies, the Linux kernel and most technical documentation use the IEC prefixes correctly. Consumer software largely does not, and many programs still write KB while dividing by 1,024. The result is that a reader must often infer from context which convention a number follows, which is exactly what the standard was written to prevent.
In practice the kibibit itself appears mainly in the specifications of small memory chips, in serial memory used by embedded systems, and in protocol documents where an exact power of two matters. Anywhere the number 1,024 is meant rather than 1,000, this is the correct unit.
One kibibit equals 1,024 bits, 128 octets, or 1.024 kilobits.
The exaoctet is a unit of digital information equal to a thousand petaoctets, or a billion gigaoctets. Its symbol is Eo. It is the unit in which the world's data flows and stockpiles are counted, and it belongs to civilisations rather than to companies or machines.
The most quoted figure involving it is global internet traffic, which passed one exaoctet per month around 2004 and now runs to several hundred exaoctets monthly. Written as an annual total that is several thousand exaoctets, which is why the zettaoctet was pressed into service for yearly figures a decade ago.
A frequently repeated claim holds that all the words ever spoken by human beings would amount to about five exaoctets in text. The estimate comes from a 2003 study and is very rough, but it makes the point: five exaoctets is now less than a fortnight of internet traffic. Whatever the exact number, the flow of data through the network long ago overtook the total of everything humanity has ever said.
Storage at exaoctet scale exists only at the largest cloud providers, and even there it is spread across dozens of buildings on several continents. The annual production of all the world's hard drives and flash memory together is a few hundred exaoctets, so the exaoctet is also the natural unit for describing the output of an entire industry.
What is stored at this scale is mostly not what people imagine. Backups, replicas, logs, telemetry, machine learning training corpora and video kept for regulatory reasons dominate. Deliberately created human work — documents, photographs, recordings that someone chose to make and keep — is a small and shrinking fraction of the whole.
The prefix exa has been part of the metric system since 1975, adopted at the same conference as peta. At the time these were prefixes without applications, defined for completeness. The fact that both became working units of a commercial industry within thirty years is the clearest single measure of how quickly digital information grew.
One exaoctet equals 1,000 petaoctets, 1,000,000,000 gigaoctets, 8 exabits, or about 0.8674 exbioctets.