Kenya's Home Internet War: Safaricom Pulls Away as Starlink Hits a Capacity CeilingKenya's Home Internet War: Safaricom Pulls Away as Starlink Hits a Capacity Ceiling

Rivals are cutting prices and doubling speeds to keep up, while Starlink, the satellite service that was meant to upend the market, has hit the limits of its own network.

For households and businesses this is the best time yet to buy home internet. Fixed connections grew 32.4% in a year to 2,838,656, and most major providers have improved what they offer since April. But the competition is also tilting the market towards one dominant player, and the headline speeds now advertised hide changes in data caps and real-world performance that matter as much as the monthly bill.

A million connections and a widening gap

The regulator's figures to 30 June 2026 show how far ahead Safaricom has pulled. It holds 36.1% of fixed connections. Jamii Telecommunications, which sells as Faiba, is second with 541,003 (19.1%), and the Wananchi Group, owner of Zuku, is third with 294,375 (10.4%).

The gap is widening. Between December 2025 and June 2026 Safaricom added 166,556 connections, about 44.2% of all new subscriptions in the market, according to a tech-ish analysis of the regulator's data. Faiba grew 9.5% over the same six months, but its share slipped from 20.1% to 19.1% because the market grew faster than it did.

Safaricom's advantage is reach on two fronts. Its fibre now passes 807,016 homes, and its 5G home service, which needs no cable to the door, had 132,060 customers in March, up 53%, according to its full-year results for the year to March 2026.

Fixed wireless delivered 52.3% of its new home and business connections last year. Safaricom earned KSh 20.2B ($155.62M) from fixed internet in that year, up 12.2%: small next to its KSh 83.4B ($642.53M) mobile data revenue, but growing faster.

It also sells to a captive audience. Safaricom holds 64.4% of mobile broadband subscriptions, which gives it a customer list, a billing channel in M-Pesa and a brand that no fixed-only rival can match.

Price cuts dressed as speed upgrades

The competition this year has mostly been fought on speed rather than headline price. In April Safaricom raised its Bronze plan from 15 Mbps to 40 Mbps and Gold from 80 Mbps to 150 Mbps, keeping the list prices of KSh 2,999 ($23.10) and KSh 6,299 ($48.53) a month, the technology site tech-ish reported.

New connections were offered about 25% off, bringing Bronze to KSh 2,250 ($17.33). It later extended faster speeds to cheaper Fibre Lite and Wi-Fi Bamba tiers, including a 10 Mbps plan at KSh 800 ($6.16).

Zuku answered within weeks. It doubled speeds on its residential plans in May, then in August introduced discounted prices so that its 30 Mbps Shujaa plan costs KSh 2,099 ($16.17) and its 100 Mbps Stream plan KSh 3,099 ($23.88), according to TechTrendsKE The effect showed up quickly: 17,768 of the 21,573 connections Zuku added in the first half of the year came between April and June.

Airtel Kenya joined the fight in late April with Xstream Fibre, starting at KSh 1,999 ($15.40) for 15 Mbps and rising to KSh 4,999 ($38.51) for 100 Mbps.

It has not yet appeared among the ten largest providers in the regulator's table, and in September tech-ish reported that Airtel is closing its separate Kenyan wholesale fibre company which never earned revenue.

The consumer fibre business sits in a different company and is not affected.

Faiba has held its line, with Home Plus plans from KSh 3,000 ($23.11) for 35 Mbps up to KSh 30,000 ($231.12) for 1 Gbps.

Zuku and the smaller players win on price per megabit

Dividing the monthly price by the advertised speed gives a rough guide to value, using prices listed by TechTrendsKE and the providers' published plans:

• Savanna Fibre, a smaller entrant, charges KSh 2,000 ($15.41) for 100 Mbps: about KSh 20 per megabit.
• Zuku Stream, on its August price, works out at about KSh 31 per megabit.
• Safaricom Gold costs about KSh 42 per megabit, and Bronze about KSh 75 at its list price.
• Airtel's 100 Mbps plan costs about KSh 50 per megabit; its entry plan about KSh 133.
• Faiba's 90 Mbps Family plan costs about KSh 56 per megabit; its entry plan about KSh 86.

On this measure Safaricom is not the cheapest, and at the entry level it sits mid-pack. It is winning because it is available in more places and easier to buy, not because it costs least.

The measure also misses one thing that matters. When Safaricom raised its speeds in April, it cut its fair usage limits, which had been set at 15 TB a month for every plan only in November 2025.

Bronze customers now drop to 4 Mbps after using 1.5 TB, a cut of about 90% in the cap, the technology site Techweez reported, and the change appeared only on the company's terms and FAQ pages. A faster connection reaches a cap sooner. For a heavy-streaming household, the cap can matter more than the speed.

Zuku, Airtel and Faiba do not publicly detail their fair usage limits, according to TechTrendsKE, which makes a like-for-like comparison harder.

Starlink's full sky

Starlink, whose owner SpaceX reported KSh 1.90T ($14.6B) in quarterly revenue this month, arrived in Kenya in 2023 promising to bring fast internet wherever a dish can see the sky. Three years on, it has 27,616 connections, 1.0% of the fixed market but 99.7% of all satellite subscriptions. It is growing fast, up 23.9% in six months, but from a small base.

It is also the most expensive service to get started on. Its Residential Lite plan costs KSh 4,000 ($30.82) a month for speeds of up to 200 Mbps, and the full Residential plan KSh 6,500 ($50.08), according to plan listings reported by The Kenyan Wall Street.

The standard kit costs about KSh 49,900 ($384.44) and the Mini about KSh 27,000 ($208.01), though Starlink has softened the upfront cost with a rental option at KSh 1,950 ($15.02) a month and, since January, an instalment plan

The bigger problem is capacity. In July Starlink stopped completing new residential orders in seven counties, including Nairobi, Kiambu and Mombasa, and asked buyers to pay a deposit to join a waiting list. By October the list covered eight counties after Kilifi was added, TechTrendsKE reported A similar freeze ran for about seven months from late 2024.

Measured speeds have slipped as users have joined. Ookla data cited by TechTrendsKE put Starlink's median download speed in Kenya at 34.55 Mbps in March 2026, down about 26.5% from roughly 47 Mbps a year earlier.

On that measured figure, the Lite plan works out at about KSh 116 per megabit, against KSh 20 on its advertised maximum. Latency is a bright spot: Ookla figures reported by Tech Africa put it at 39 milliseconds in early 2026, served from a Nairobi gateway, good enough for video calls.

Starlink's real market is the one fibre will not reach for years: farms, lodges, rural schools, clinics and construction sites. In Nairobi's estates it is a premium backup, not a threat to Safaricom or Zuku.

The estate wireless challengers

The fastest-growing part of the market is not fibre at all. Terrestrial wireless connections, beamed from masts to rooftops, rose 42.9% in a year to 1,030,770, against 29.7% growth for fibre.

Three smaller providers show what that means. Vilcom grew 39.7% in six months to 186,240 connections, the fastest of any provider above 100,000. Ahadi Wireless, which sells as Konnect in estates such as Zimmerman and Githurai 44, grew 21.9% to 270,586 and passed Poa Internet for fourth place. Mawingu Networks crossed 100,000 connections for the first time.

Poa, an early name in low-cost estate internet, is the exception. It lost 14,704 connections in six months and has shrunk for four straight quarters, the only one of the ten largest providers to get smaller. The regulator does not say why.

Better service is the hardest claim to test

The question readers ask most is which service is best, and it is the one the public data answers least well. The regulator's quarterly report counts connections by provider but does not rank home internet providers by measured speed or reliability, so claims about quality rest on company marketing, crowd-sourced speed tests and customer complaints.

What the numbers do show is clear enough. Safaricom has the biggest customer base and the strongest balance sheet, and is pulling away. Zuku and smaller fibre players offer more speed per shilling where their cables reach. The estate wireless providers are winning customers that fibre has not yet reached. Starlink serves places nobody else does, at a premium, and cannot currently add customers in Kenya's biggest towns.

For buyers, the practical advice is to compare the fair usage limit and post-cap speed alongside the price, check coverage street by street, and ask neighbours about evening performance. For the market, the risk is concentration.

If one provider keeps winning almost half of all new customers, the price cuts of 2026 may not last. The Communications Authority could help by publishing measured speeds for each provider, as it already does for connection numbers, so households can judge quality on evidence rather than advertising.