2.2 Billion People Are Still Offline in 2026

Mobile network coverage is close to finished as a global project. ITU reports that 96% of the world's population lived within range of a 3G-or-better network in 2025 — 3.6% with only 3G, 37.6% reaching 4G and 55.1% reaching 5G. The last stretch is the hard one. Coverage first passed 90% in 2018 and has gained just six percentage points in the seven years since, because the roughly 312 million people still outside any footprint live where towers are expensive to build and lightly used once built. Almost half of them are in Africa.

That is what makes the subtraction worth doing. Take ITU's 2.2 billion offline, remove the 312 million with no network, and about 1.9 billion people are left who could connect to a mobile broadband network today and do not. That is roughly 86% of everyone offline. It is arithmetic on two ITU figures rather than an ITU finding, and both inputs are rounded — but the direction is not in question, and it reframes the whole problem.

What keeps those 1.9 billion off is a stack of ordinary obstacles rather than a missing tower: the price of a handset, the price of a gigabyte, electricity that is unreliable or absent, literacy and digital skills, content in the wrong language, and in many households an unwritten rule about who gets to use the one phone. ITU's own framing is that 96% of the people who are offline live in low- and middle-income countries — a different 96% from the coverage figure, measuring a different population, and worth keeping firmly separate from it.

A reader who checks a second source will quickly find a bigger number. GSMA's State of Mobile Internet Connectivity 2026 reports 4.8 billion mobile Internet users at the end of 2025 — 59% of the world — and 3.4 billion people not using mobile Internet, of whom 38% of the global population live under coverage without using it and 3% have no coverage at all. Set against ITU's 6 billion online and 2.2 billion offline, that looks like a flat contradiction.

It is not. The two organisations count different populations. ITU counts anyone who used the Internet in the previous three months on any device over any connection, including fixed broadband at home, a shared household phone, and public access at a school, library or café. GSMA counts people using mobile Internet on a device of their own. The distance between 6.0 billion and 4.8 billion is mostly made up of people who get online through something other than a handset they personally own.

Both figures are correct on their own definitions, and neither belongs in the other's sentence. The chart at the top of this piece is ITU's, on ITU's definition, for 2025 — which is also why the offline slice is simply 100 minus ITU's published 74, rather than a share the agency prints.

Slice ITU's 2025 data any way you like and income wins. High-income countries sit at 94% Internet use, upper-middle-income at 89%, lower-middle-income at 63% and low-income at 23%. The 71-point distance between the top and bottom groups is the widest gap anywhere in the dataset. The next widest cut, urban against rural, is 27 points. Age is 10 points and gender is 6. Every much-discussed divide turns out to be smaller than the one between rich and poor countries.

The divides also compound rather than sit side by side. A city dweller in a high-income country is at 96%. A rural resident of a low-income country is at 14% — ITU's own phrasing is "only one in seven rural dwellers". That 82-point spread is wider than any single-dimension cut in the report, and it is the gap that actually describes who the 2.2 billion are: rural, poor, and usually both at once.

The UN's special country groupings tell the same story from another angle. The least developed countries reached 34% in 2025 and the landlocked developing countries 38%, against a world figure of 74%. Small island developing states sit at 64%, and are the one group in the entire dataset where women and men are online at exactly the same rate — 64% each, which ITU's own text confirms as genuine parity rather than a rounding artefact.

ITU's six regions run from the CIS at 93% and Europe at 92% down to Africa at 36%, a spread of 57 percentage points. The Americas reach 88%, Asia-Pacific 77% and the Arab States 70%. Only Africa and the Arab States sit below the world's 74%, and Africa sits so far below it, with so large a population, that it pulls the global figure down on its own.

Africa is the outlier more than once over. It has the lowest use, the lowest 5G coverage at 12% of population, the highest prices at 5.3% of average income for a 5GB mobile plan, the widest urban-rural ratio at 2.6 to 1, and almost half of everyone on earth with no mobile network at all. It is also where most of the remaining growth has to come from, because the regions already above 88% have very little headroom left to gain.

One caveat matters before anyone reads the table below as a map. ITU's six regions are administrative groupings, not continents. The Arab States cut across Africa and Asia — Egypt, Algeria, Morocco, Sudan, Libya and Tunisia are counted in both Africa and the Arab States — and the CIS cuts across Europe and Asia, with Russia and Belarus in both. There is no Oceania row, because Oceania sits inside Asia-Pacific. The six overlap, they do not partition humanity, and they do not sum to the world.

The clearest reason a signal overhead does not become a connection is price, and ITU measures it in the only way that travels across borders: as a share of average income. An entry-level mobile broadband plan with at least 5GB a month costs 0.4% of GNI per capita in Europe and 9.4% in low-income countries. Africa as a region sits at 5.3%, against a world median of 1.4%. ITU's own summary is that subscribers in low-income countries spend about 22 times more, relative to what they earn, than subscribers in high-income countries.

The affordability benchmark ITU measures economies against is 2% of GNI per capita. In 2025, 130 of 205 economies met it on the mobile basket, six more than the year before, and 88 of 195 met it on the fixed basket. But only about four in ten low- and middle-income economies met the target on either basket — which is to say it is largely met where it was never really binding and missed where it matters most.

One thing not to do with these figures is read them as a trend. ITU moved to a data-only 5GB basket in 2024, and the 2024 numbers printed in its 2025 report are experimental back-casts fitted to that new methodology rather than the figures the agency published at the time. Year-on-year movements on this measure are not comparable in the ordinary way, so this post quotes 2025 levels and leaves the direction of travel alone.

The rural-urban split is the second widest cut in ITU's data: 85% of city dwellers were online in 2025 against 58% of rural residents, a ratio of about 1.5 to 1 that has narrowed only slightly from 1.6 four years earlier. That ratio is 1.1 in Europe and 2.6 in Africa — so the penalty is not really about countryside as such. It is about countryside in places without much money.

The gender gap is real and narrower than its reputation. Women are at 71% globally and men at 77%, a six-point difference that works out to roughly 280 million more men than women using the Internet. ITU's parity score has sat at 0.92 since 2019, unmoved in six years. Europe, the Americas and the CIS have reached parity; the Arab States sit at 0.86 and Africa at 0.78, though Africa's figure has climbed sharply from 0.70 in 2019.

Age is the narrowest cut of all at world level: 82% of 15-to-24-year-olds are online against 72% of everyone else. But the average hides the interesting part. In high-income countries young people are only about 5% more likely to be online than the rest of the population, because the rest of the population is already at 94%. In low-income countries they are 1.9 times more likely — 38% against 20%. Youth connectivity arrives first, and everybody else follows slowly.

Growth is real. World Internet use rose from 54% in 2019 to 74% in 2025 — twenty percentage points in six years — and the 2025 year-on-year increase of 3.3% was slightly faster than 2024's 2.9%. But the gains have not landed evenly. Lower-middle-income countries moved from 33% to 63% over the same six years, a thirty-point leap. Low-income countries moved from 16% to 23%. Seven points.

The infrastructure curve is flattening for the same underlying reason. Mobile coverage passed 90% of the world in 2018 and has added six points since, and ITU describes extending it further as "slow and complex". The places the remaining growth must come from — rural Africa, the least developed countries, the landlocked group whose mobile data prices actually moved up rather than down between the two most recent readings — are exactly the places where each additional percentage point is most expensive to buy.

A measurement note belongs here too, because the headline figures themselves move. ITU revised its 2024 estimate upward, from 5.5 billion Internet users to 5.8 billion, which pulled the 2024 offline count down from 2.6 billion to 2.3 billion. The older 2.6 billion is still widely quoted, and setting it against 2025's 2.2 billion would invent a one-year fall of 400 million people that never happened. On a consistent basis the move was 2.3 billion to 2.2 billion.

Three quarters of humanity was online in 2025 and about a quarter was not. Read as a coverage story, that looks like 2.2 billion people waiting for a network to arrive. Read against ITU's own coverage figures, it looks like something else entirely: roughly 1.9 billion of them already live under a mobile broadband signal, and what stands between them and a connection is a handset they cannot afford, a data plan priced at nearly a tenth of a month's average income, a power supply that does not hold, or skills nobody taught them.

That changes what closing the digital divide has to mean in practice. Building networks out to the remaining 312 million uncovered people is still worth doing and still genuinely hard — six percentage points in seven years is the measure of how hard. But the larger job is on the demand side, and it is counted in handset prices, data tariffs, electrification and basic digital education rather than in base stations. The 71-point gap between high- and low-income countries will not close because a tower went up.

A closing note on freshness: every figure in this piece comes from ITU's Measuring digital development: Facts and Figures 2025, published in Geneva in November 2025 and still the current edition. ITU publishes the series annually in November, so a 2026 edition is due shortly and these numbers should be re-read against it once it lands.