Society
Who is still offline
Three-quarters of the world is online, the remaining quarter is overwhelmingly poor and rural, and the headline count is a model output rather than a headcount.
Roughly 6 billion people — about 75% of the world's population — used the internet in 2025, an increase of 240 million on the previous year (International Telecommunication Union, 2025). The people who remain offline are not randomly distributed: they are concentrated in low-income countries, in rural areas, among older people and among women. Within rich countries the divide has changed shape rather than closed, splitting into a gap between people with a proper home connection and people who have only a phone. The most important caution is about the numbers themselves. Nobody counts the offline. The global figures are modelled estimates reconciling national surveys of very different ages and definitions, and they are revised substantially between editions. For a workplace-oriented example of technology measurement in practice, see remote workforce management software.
The remaining gap is almost entirely an income gap
The single cleanest statement about who is offline is a statement about national income. The pattern holds at every level of the data ITU publishes, and it is the reason the divide is not closing at the rate the global average suggests.
2.2 billion people remained offline in 2025, and 96% of them lived in low- and middle-income countries.
Direction: Decrease. Strength of evidence: Strong.
Caveat "Offline" means no internet use in the previous three months, which does not distinguish never-users from lapsed or intermittent users.
The spread between income groups is the largest in the dataset, larger than the gap between men and women, between town and country, or between age groups.
94% of people in high-income countries use the internet, against 23% in low-income countries.
Direction: Decrease. Strength of evidence: Strong.
Caveat Income-group aggregates hide enormous within-group variation, and low-income estimates lean heavily on modelling rather than direct measurement.
The other divides are real but smaller. Globally 77% of men are online against 71% of women, 85% of urban residents against 58% of rural residents, and 82% of 15–24-year-olds against 72% of everyone else (ITU, 2025). Each of those averages conceals its own structure: the gender gap is close to zero in high-income countries and much wider in South Asia and Sub-Saharan Africa, and "rural" is defined by each national statistics office, so the urban-rural boundary is not comparable between countries. The same modelling problem affects where mobile networks have actually been built: 5G reaches 55% of the world's population, but 84% of people in high-income countries have access against 4% in low-income countries, and coverage is reported by operators as signal availability rather than as usable speed on an affordable device (ITU, 2025).
The coverage-gap framing comes from the industry it exonerates
Policy writing about connectivity now runs almost entirely on one distinction: the coverage gap (people with no mobile broadband signal) and the usage gap (people who live under a signal but do not use it). It is a genuinely useful distinction, and it is worth knowing who produced it. It comes from GSMA, the mobile operators' trade association, using a proprietary model that has not been independently replicated.
The global coverage gap fell from 19% to 4% of the population between 2015 and 2024 while the usage gap fell only from 49% to 38%.
Direction: Decrease. Strength of evidence: Mixed.
Caveat GSMA is the mobile industry's trade association and the report is industry-funded; the framing places the remaining problem in consumer behaviour rather than in operator investment, and the model is not auditable.
That conclusion — network build is broadly finished, demand is the obstacle — is precisely the conclusion a trade body would want established before regulators write the next round of coverage obligations. It may still be correct. But it is not an independent finding, and the 4%/38% split has no published replication outside the organisation that produced it. The same source puts Sub-Saharan Africa's usage gap at 65% of the population in 2024, against 81% of people connected in North America and 77% in Europe and Central Asia (GSMA, 2025).
Affordability is benchmarked against the wrong income
Cost is where the coverage/usage framing does most of its work, because "does not use an available signal" quietly absorbs "cannot pay for it". The affordability measures in common use have a structural flaw.
Broadband remains unaffordable in around 60% of low- and middle-income countries.
Direction: Decrease. Strength of evidence: Mixed.
Caveat "Unaffordable" is measured against the Broadband Commission's target of 2% of GNI per capita, an average-income denominator that understates unaffordability for the poorest households.
A package that costs 2% of average national income can cost well over 10% of income for a household in the bottom fifth of the distribution. Almost no published series measures that, so the affordability picture is systematically flattering.
In rich countries the divide moved indoors
High-income countries have close to universal use and a persistent gap in connection quality. In the United States, near-universal internet use coexists with a substantial group whose only connection is a handset.
96% of US adults use the internet and 91% own a smartphone, but only 78% have home broadband and 16% are smartphone-dependent, with no home subscription.
Direction: Decrease. Strength of evidence: Strong.
Caveat Pew's demographic cross-tabs for income and geography sit behind interactive tabs, so the breakdowns are not quoted here.
Smartphone dependence matters because a phone is a poor instrument for the tasks that carry the most value — long-form applications, benefits claims, coursework, job applications. It is the practical difference between being counted as online and being able to do the things a connection is for, including buying and paying for things at the prices available to people who can shop around.
Europe's figures are the highest published anywhere, with an important asterisk on who is counted.
94% of individuals in the EU used the internet in the previous three months in 2025, with Ireland, Denmark and the Netherlands above 99% even in rural areas.
Direction: Increase. Strength of evidence: Strong.
Caveat The survey covers ages 16-74 only, excluding the over-75s, who are the most digitally excluded group, so it overstates population-wide connectivity.
The short version
- About 6 billion people, roughly 75% of the world, used the internet in 2025, and 2.2 billion did not (ITU, 2025).
- Internet use runs at 94% in high-income countries and 23% in low-income countries, the widest divide in the data.
- The coverage-gap/usage-gap framing that dominates policy writing was produced by the mobile operators' trade association using a proprietary, unaudited model.
- Near-universal use in rich countries hides a quality divide: 16% of US adults rely on a phone with no home broadband subscription (Pew Research Center, 2025).
- Every global figure here is modelled rather than counted, and the headline offline population has been revised by hundreds of millions between editions for methodological reasons.