Skip to content
Velloreth
Nepal & Innovation 10 min read

Internet Infrastructure in Nepal: Current State

Mobile broadband dominates Nepal, FTTH is growing fast, and the real gaps are rural, backhaul, and last-mile reliability. A grounded look at the country's internet infrastructure in 2026.

By

Nepal’s internet infrastructure in 2026 is best understood through one tension: the country has more broadband subscriptions than people, and still has households with no meaningful access at all. Both facts come from the same data. The Nepal Telecommunications Authority’s MIS reporting, based on end-of-2025 data, counted over 30 million broadband subscriptions against a population of roughly 29 million. The same period’s household survey work puts real household broadband access closer to one-third, with a stark urban-rural split underneath.

Subscription counts and lived access are different measurements. This post walks through both.

The numbers, as of late 2025

The NTA’s MIS data (subscriptions as of end November 2025) gives the cleanest picture of the market structure:

Service typeSubscriptionsMarket share
Mobile broadband (3G/4G)~26.9 million~89%
Fixed wired (FTTH, ADSL, cable)~3.3 million~11%
Fixed wireless (radio/WiFi)~36 thousand~0.1%
Total broadband~30.3 million100%

A few things jump out.

Mobile dominates, completely. Nearly nine in ten broadband subscriptions are mobile. Nepal Telecom holds the larger share of mobile broadband (~15.1 million subscriptions) with Ncell close behind (~11.8 million), a split that has been stable as the market moved from 3G to 4G. CDMA and EVDO-era services have been phased out by Nepal Telecom, consolidating the market on modern mobile technology.

FTTH is the fixed-line story. Of the ~3.3 million fixed wired connections, virtually all of the growth is fiber to the home: roughly 3.34 million FTTH connections, of which private ISPs account for about 3 million and Nepal Telecom around 340,000. ADSL and cable are legacy residue. The ISP segment — WorldLink, Subisu, Broadlink, and a long tail of local operators — built this market, and it is one of the more competitive broadband markets in South Asia by operator count in major cities.

Fixed wireless is negligible. Radio-based WiFi broadband — the technology that theoretically solves rural last-mile cheaply — sits at around 36,000 subscriptions, a rounding error. The economics of point-to-multipoint wireless never scaled here the way anyone hoped.

The headline “over 100% penetration” figure that appears in NTA reporting is a subscription-to-population ratio, inflated by multi-SIM behavior and multiple subscriptions per household. It says more about how people buy connectivity than about how many are connected.

The divide that the averages hide

The Nepal National Social Audit Council’s analysis of Nepal Living Standards Survey microdata (2022/23) is the corrective to the subscription numbers: approximately one-third of households reported having broadband internet access, with connectivity strongly stratified by geography, education, and income.

The headline gaps:

  • Urban roughly 65% versus rural roughly 22% in household access at the time of the survey.
  • Education gradient: around 90% connectivity among households with master’s-level education versus around 23% for those at secondary school completion or below.
  • National policy targets — the National Broadband Policy aims at 95% broadband coverage by 2030 — implicitly acknowledge that the current trajectory alone will not close the gap.

Household access and individual subscription counts measure different things, and both matter. A farmer with a 4G SIM has internet; a household with shared phone-based access and no computer, no fixed line, and metered data does not have internet in the way that digital services design assumes. The gap between “has a data plan” and “has broadband” is where most digital inclusion policy lives.

Backhaul: the landlocked constraint

Nepal is landlocked. It has no coastline, no submarine cable landing, and therefore no direct international bandwidth infrastructure of its own. Nearly all international internet bandwidth arrives overland from India — through cross-border fiber interconnects, with traffic predominantly landing in Mumbai, the region’s dominant connectivity hub.

This shapes everything downstream:

Latency. A user in Kathmandu accessing a service hosted in Mumbai sees the low tens of milliseconds — fine for most applications. Accessing anything in Singapore, Europe, or the US East Coast adds the Mumbai hop plus the intercontinental path, which is why cloud services feel slower here than the raw line speed suggests. Applications designed with that latency profile in mind (edge caching, regional CDNs, Mumbai-region cloud deployment) perform far better than naive single-region deployments pointed at Virginia.

Cost and resilience. International bandwidth is priced through the upstream chain, and a single-geography dependency means maintenance events or cuts on the Indian side degrade Nepali connectivity. Domestic redundancy exists — the Nepal Internet Exchange (NPIX) keeps in-country traffic local where networks peer, and multiple operators maintain separate cross-border links — but the fundamental dependence on overland routes to India remains the structural fact of Nepali internet geography.

Domestic peering. NPIX and operator-level peering keep domestic traffic domestic where participants cooperate — WhatsApp calls between two Kathmandu users, local streaming caches, domestic banking traffic. Traffic that does not peer locally takes the longer international path. The peering ecosystem is functional but smaller than what a country of India’s size maintains; growth in local content and local cloud instances would improve it further.

Mobile networks: the 4G decade, and what comes after

Mobile is where most Nepalis actually get internet, so the state of the two national networks deserves its own look. Both Nepal Telecom and Ncell have spent the last several years densifying 4G coverage across the Terai and valley corridors and extending it into hill districts where terrain makes every additional site expensive. The practical experience in 2026: solid 4G in cities and along highway corridors, usable but variable 4G in smaller district towns, and patchy or edge-of-coverage service in remote mountain areas — where a single hillside between you and the nearest site is the difference between video calls and nothing.

5G, where trials and early deployments have been discussed and tested by operators, does not yet change the national picture: the economics of 5G investment favor dense urban demand, and Nepal’s data demand is still primarily a 4G-and-FTTH story. For most users, the next several years of improvement will come from 4G densification, spectrum refarming, and fiber backhaul reaching more cell sites — the unglamorous work that raises average experience — rather than from a flagship 5G launch.

The structural issue on the mobile side is site economics: every rural base station needs power, backhaul (itself often microwave or fiber along a road), physical security, and maintenance access on difficult terrain. Operators weigh those costs against subscriber density, and the arithmetic favors Terai and valley expansion over remote-hill coverage. That is why the rural connectivity gap is as much an infrastructure-finance question as a technology one — and why policy tools (universal service obligations, shared infrastructure, tower sharing) matter more for the last 20% than any radio technology choice.

Power, the quiet dependency

Internet infrastructure assumes electricity. Outside the Kathmandu valley, grid reliability has improved substantially from the severe load-shedding decade but has not reached the “never think about it” level that dense urban broadband markets assume. ISP equipment, mobile base stations, and customer premises gear all depend on power continuity, and operators carry the cost of battery and generator backup across thousands of sites.

For fixed-line users in cities, power is a minor inconvenience — an occasional outage that takes down the ONT until current returns. For mobile network expansion into rural areas, power cost is a first-order constraint on site economics, which is part of why rural 4G coverage lags the population map.

The improvement trajectory is real and worth acknowledging: the grid is dramatically better than it was in the mid-2010s, and that improvement quietly enabled the FTTH and 4G expansion of the last several years. Continued reliability gains will do more for rural connectivity than any single last-mile technology choice.

What the infrastructure means in practice

For urban users

Kathmandu and Pokhara residents on FTTH have access to genuinely good broadband: hundreds of Mbps downstream at prices competitive with regional peers, from a competitive ISP market where switching is common. Mobile 4G covers dense urban areas adequately, with congestion in specific neighborhoods at peak hours being the main complaint category. For remote work, streaming, video calls, and cloud development tools, the urban experience in 2026 is not the constraint it was five years ago. Power backup at the ISP and building level has made outages a nuisance rather than a work stoppage for most knowledge workers.

For rural users

The picture is mobile-first or nothing. 4G reaches a large share of the population through Nepal Telecom and Ncell coverage maps, and it carries nearly all rural internet use. Fixed broadband, where it exists outside cities, is typically point-to-point wireless or a fiber spur along a highway corridor — not the dense FTTH mesh cities enjoy. Data cost as a share of rural income remains meaningful, which shapes usage toward messaging and video rather than the always-on, multi-device pattern that urban broadband enables.

This is not a market failure so much as an arithmetic one: rural last-mile economics in a mountainous country with dispersed settlement are genuinely hard, and the policy question is which parts of the gap justify public subsidy (the 95%-by-2030 target implies some do) versus which will close as incomes and mobile coverage grow.

For businesses and developers

A few practical implications for anyone building or operating on Nepali infrastructure:

  • Deploy in Mumbai, cache globally. For services used in Nepal, a Mumbai-region cloud instance plus a CDN with regional PoPs gives the best latency profile available. Kathmandu-local hosting exists but adds operational burden without meaningfully beating Mumbai-plus-CDN latency for most workloads.
  • Design for intermittent connectivity at the edge of your user base. Mobile-first, offline-tolerant patterns matter for any product that touches users outside the major cities — the same patterns that serve emerging markets globally.
  • FTTH is reliable enough to depend on in cities; power is your local risk. For home offices and small infrastructure, UPS or inverter backup for the router and ONT solves the most common outage mode. This is basic advice anywhere; in Nepal it is closer to required equipment.
  • The payment and cloud-billing gap is harder than the bandwidth gap. A developer in Kathmandu can get fast internet; getting a predictable international card for cloud and SaaS billing is a separate, still-annoying problem. Budget for it operationally.

Policy and trajectory

The institutional framework is coherent even where execution lags: the Digital Nepal Framework sets digital economy targets, the National Broadband Policy aims for 95% coverage by 2030, the NTA regulates and reports, and private ISPs plus two mobile operators build most of the actual infrastructure. The market structure — competitive ISPs in cities, a mobile duopoly nationally, an exchange for domestic peering — is reasonable for a country of this size.

The open questions for the second half of the decade are mostly not technical. Whether rural coverage targets get funded and enforced; whether cross-border bandwidth diversity improves beyond the India-dependent topology; whether domestic cloud and content hosting grows enough to make peering economically compelling; whether power reliability gains continue at the recent pace. Any one of those has more influence on what the internet feels like in Nepal in 2030 than any new radio technology.

The infrastructure is past the point of being the binding constraint for most urban digital life and most technology-sector work. It remains the binding constraint for inclusion — for the roughly two-thirds of households that still experience the internet, if at all, as a shared phone on a metered plan rather than a utility in the wall. That gap is the honest state of the union.

Frequently asked questions

Is internet in Nepal fast enough for daily use?
For most urban users, yes. FTTH plans in Kathmandu and Pokhara offer hundreds of Mbps at prices competitive with regional peers, and 4G covers most populated areas. The experience degrades in rural areas, during outages, and on congested mobile networks in dense neighborhoods.
Why is fixed broadband penetration lower than mobile?
Cost and coverage. Mobile broadband reaches where cables do not, with a SIM and a phone being the only hardware required. Fixed broadband requires physical last-mile infrastructure, which private ISPs have built aggressively in cities and barely touched in most rural areas.
Does Nepal have its own internet exchange?
Yes — the Nepal Internet Exchange (NPIX) and related peering arrangements keep some domestic traffic local. But a large share of international traffic still routes through India (typically Mumbai), which determines most latency to global cloud services and platforms.
How does Nepal's internet compare to neighbors?
Urban fixed broadband is competitive with Indian tier-2 cities on speed and price. Rural coverage and reliability lag. The structural constraints — landlocked backhaul, smaller peering ecosystem, power reliability outside the main grid corridors — are geographic and economic, not primarily technical.