Data / Telecoms
A gigabyte of mobile data in Lesotho now costs M32.75. That is the Data Matrix Research September 2026 reading, taken from the cheapest widely sold 30-day bundle on operator tariff pages. Three years earlier the same yardstick stood at M42.00. The fall is real. The cause is contestable competition, and Starlink is the new constraint on the duopoly.
One gigabyte bought on a 30-day pack cost M42.00 in 2023, M38.00 in 2024, M35.00 in 2025 and M32.75 in September 2026. Data Matrix labels the series Observed. The shop window is public. The method is the same each year: read the cheapest widely sold 30-day bundle and divide by the gigabytes in the pack. The next snapshot is due on 1 October 2026.
That is a drop of M9.25, or 22 percent, in three years. Consumer inflation was 2.9 percent in July 2026. The real price of a gigabyte has therefore fallen faster than the nominal index. Data Matrix also records an Econet flagship pack at M655.00 (about $36.80) in the same September reading.
Figure 1. Lesotho mobile data: price of 1 GB, 2023 to 2026. Source: Data Matrix Research, September 2026 snapshot. Observed from operator tariff pages.
A gigabyte that cost M42.00 in 2023 costs M32.75 now. Data Matrix observed the tariffs. The market moved.
Why Lesotho data prices started falling before the dish
The Data Matrix series does not show a cliff in June 2025. It shows a slope. Prices were already easing while the market still had two terrestrial operators, Vodacom Lesotho and Econet Telecom Lesotho. Scale, 4G densification and larger bundles all push the unit price of a gigabyte down even in a quiet duopoly. That is the first fact. It should not be wished away to make a neater story about Starlink.
The second fact is the timing of the new constraint. The Lesotho Communications Authority granted Starlink a 10-year network service licence on 14 April 2025. T-Connect Lesotho launched the service in June 2025. Hardware sat near M7,400 (about $416) for the standard kit and M3,800 (about $213) for the Mini. The residential subscription sat near M950 (about $53) a month, with no data cap. In August 2026 Vodacom began reselling Starlink to enterprises on the same constellation. The sky became a third pipe.
Starlink is a poor substitute for a student buying 1 GB on a prepaid SIM. It is a close substitute for a household or a firm that was paying the operators for a large home or shop connection. That is the relevant margin. Once an unlimited satellite link exists at a known monthly fee, the old networks cannot hold the top of the tariff ladder. They cut large-pack prices and, because the packs are the source of the Data Matrix benchmark, the recorded price of a gigabyte falls with them.
How Lesotho data prices sit in the region
Converted at M17.80 to the dollar, the September Data Matrix reading is about $1.84 a gigabyte. That is not cheap by the standards of the poorest African markets, and it is not ruinous by the standards of the dearest. Cross-country tables mix methods, so they must be read as a map, not a court exhibit.
Figure 2. Southern Africa 1 GB price map. Darker red is dearer. The Lesotho cell is the Data Matrix September 2026 reading. Other cells are published 2025 to 2026 averages; methods differ.
Figure 3. What 1 GB costs across Southern Africa. Sources labelled on the chart.
MarketUS$ / 1 GBSource and yearZambia0.57StatRanker average, 2025Mozambique0.61StatRanker average, 2025Malawi0.69StatRanker average, 2025Eswatini0.91Cable.co.uk / Broadband Genie, 2026South Africa1.29StatRanker average, 2025Botswana1.74StatRanker average, 2025Lesotho1.84Data Matrix, September 2026Zimbabwe5.10Research ICT Africa RAMP, Q4 2025
Table 1. Cross-source comparison. Not a single survey. The Lesotho figure is converted at M17.80 per US dollar.
Lesotho sits with Botswana, above South Africa on this mixed map, and far above Zambia, Malawi and Mozambique. Cable.co.uk’s global average in June 2026 was $2.59 a gigabyte. On that global yardstick the Data Matrix reading is below the world mean. On a SADC yardstick it is still towards the dearer middle. The Alliance for Affordable Internet test, 1 GB at no more than 2 percent of monthly income, remains a stretch for many Basotho households even at M32.75.
The regional pattern has a simple industrial-organisation reading. Markets with more effective rivals, or with a credible threat of one, post lower unit prices. Markets that stay as a tight pair, or that load taxes and interconnection costs onto prepaid users, do not. Lesotho had the pair. It now has the threat.
Why competition cuts data prices for consumers
A two-firm mobile market tends toward Cournot logic. Each operator takes the other’s capacity as given and withholds some output to protect the margin. The result is a price above marginal cost and a deadweight loss: some users who value a gigabyte at more than it costs to supply do not buy it. Consumer surplus shrinks. The operators collect an oligopoly rent.
Entry changes the game even when the entrant’s volume is small. William Baumol’s contestable-market argument is the clean version. If a rival can arrive, serve the profitable slice of demand, and leave, the incumbents must pre-empt that raid by cutting price toward average cost. They do not wait to lose customers. They limit-price. Starlink did not need a mass prepaid base in Maseru to move the tariff page. It needed to be a believable alternative for the high-usage accounts that pay for the networks.
On the demand side the effect compounds. Data is price-elastic once the pack is large enough to be useful for work, school and video. A lower price raises quantity. Higher quantity spreads the sunk cost of spectrum, towers and backhaul across more gigabytes, which lowers average cost, which permits another cut. That is the virtuous loop competition is supposed to start. A protected duopoly interrupts it.
None of this requires a morality play. Operators are not villains for charging what the market will bear. They are firms with sunk metal in the ground. The public interest is the opposite of theirs at the point of sale. The public interest is a price close to cost, wide use of the network, and a surplus that sits with households rather than with a tariff committee. The cheapest way to get there is another pipe. Starlink is that pipe for the mountain districts the towers never quite reached, and a bargaining chip for everyone else.
Two caveats keep the claim honest. First, the Data Matrix fall from 2023 to 2025 happened before the licence. Technology and bundle redesign explain part of the 22 percent. Second, a M950 unlimited dish is still a luxury relative to a M30 (about $1.69) prepaid top-up. The poor buy small packs, and small packs remain the dearest way to purchase a gigabyte. Competition has done more, so far, at the top of the ladder than at the bottom. The September index of M32.75 is the proof that the ladder has moved. It is not proof that data is now cheap.
Data Matrix will read the tariff pages again on 1 October. If the operators are still answering the dish, the next print will be lower.
Related: [internal link: Starlink launches in Lesotho, June 2025]
We’d love your opinion
This week’s reader poll — takes 5 seconds
Economy
