Measured but not counted: How Starlink broke Potraz’s bandwidth framework

By Valentine Muhamba | 1h ago

Starlink kit

ZIMBABWE’S Q1 2025 Postal and Telecommunications Regulatory Authority of Zimbabwe (Potraz) report contains a number that should not be possible. Used incoming international internet bandwidth for the quarter stood at 2 693 127 Mbps. The total equipped international internet bandwidth capacity for the same period was 1 433 865 Mbps. A country’s internet usage cannot exceed its installed capacity. Except, in this case, it apparently did, by 1 259 262 Mbps. The explanation is Starlink, and the explanation reveals something about the limits of how Zimbabwe’s regulator was measuring its own market.

Starlink launched in Zimbabwe in September 2024 after years of regulatory obstruction, during which Potraz had warned citizens the service was unlicensed and threatened enforcement against users. By Q4 2024, VSAT subscriptions had jumped 513,79% to 23 410 as sign-ups flooded in. By Q1 2025, there were 30 907 active terminals. The Potraz Q1 report attributed 83% of the country’s entire used incoming international bandwidth to Starlink, directly stating it “commanded 83% of the usage in the quarter”. What the report did not do was include Starlink in the equipped capacity table. That table, which lists the licensed bandwidth infrastructure of Liquid, TelOne, Powertel, Dandemutande, Telecontract, and Africom, totalled 1 433 865 Mbps and contained no Starlink entry. The regulator was counting what Starlink consumed while having no framework to account for what it provided.

The arithmetic makes the problem concrete. Starlink’s 83% share of used bandwidth implies it consumed approximately 2 235 295 Mbps of incoming capacity in Q1. The entire equipped capacity of all licensed operators combined was 1 433 865 Mbps. Starlink alone was using 56% more bandwidth than the entirety of Zimbabwe’s formally registered internet infrastructure. The non-Starlink operators, by implication, used roughly 457 832 Mbps, which against their equipped capacity of 1 433 865 Mbps represents a utilisation rate of about 32%. That figure is coherent. The Starlink figure exists outside the framework entirely, because Starlink’s satellite network has no Zimbabwean ground infrastructure to measure.

From Q2 2025 onwards, Potraz changed its approach. A footnote appeared beneath the used bandwidth figures in the Q2 report, reading: “The figure above excludes Used International Internet Bandwidth Capacity for Starlink until a clear position is reached on how to accommodate its capacity.” The identical note appeared in Q3. To make its Q2 comparisons internally consistent, Potraz also retroactively restated the Q1 used bandwidth figure, stripping out Starlink to arrive at a revised baseline of 459 157 Mbps, against which Q2’s 504 778 Mbps represented a modest 9 94% increase. That restated figure appears only in the Q2 report. The Q1 report, read in isolation, still shows 2 693 127 Mbps as the used bandwidth figure. A reader moving sequentially through the four quarterly reports, comparing headline figures without reading footnotes, sees used incoming bandwidth collapse from 2 693 127 Mbps in Q1 to 504 778 Mbps in Q2, an apparent fall of 81%. The fall is entirely a methodological reclassification. No quarterly summary of Zimbabwe’s internet sector flagged the discontinuity.

Coverage of the Q1 figure focused almost entirely on the 83% statistic as a measure of Starlink’s dominance. Techzim, which reported the figure in July 2025, observed the anomaly directly: “We don’t even know how Starlink total available bandwidth and used bandwidth are measured, figures presumably coming from the company itself.” That scepticism was correct but was not pursued. The broader implication, that the Q1 and Q2 used bandwidth figures are measuring different things and therefore cannot be compared as a sequential series, was not drawn.

By Q4 2025, Starlink Zimbabwe held 35,05% of the country’s fixed internet and data traffic, having carried 168,21 petabytes in the quarter. It remained absent from the equipped international bandwidth capacity table. The Q4 table lists seven operators, runs to 1 688 770 Mbps total, and does not contain a Starlink entry. Starlink was by then the second largest fixed internet operator in Zimbabwe by traffic, accounting for more traffic growth in a single quarter than the rest of the market combined. The regulatory framework for measuring what it provides has still not been resolved.

The measurement gap matters for reasons beyond statistical tidiness. Equipped bandwidth capacity is the regulatory instrument through which Potraz assesses whether the sector has sufficient infrastructure to meet demand, allocates spectrum and gateway licences, and benchmarks operator investment. A framework that counts Starlink’s consumption in one quarter, excludes it in the next, and never accounts for its provision produces a picture of the market that is structurally incomplete. Dandemutande CEO Never Ncube, whose company is Starlink’s authorised reseller in Zimbabwe, described the adoption as operating across “primary and secondary internet links” for individuals, businesses, schools, and public institutions. An industry commentator, speaking to TechCabal, named the regulatory problem directly: “Starlink in Zimbabwe operates with fewer ground infrastructure obligations and lighter regulatory demands. It is hard for local players to compete when they are not playing by the same rules.” Potraz has signalled a willingness to revisit its licensing and spectrum frameworks to accommodate non-terrestrial networks. The four quarterly reports of 2025 show that the revisiting has not yet produced a coherent measurement standard. The operator that dominated Zimbabwe’s internet traffic in Q4 is still not in the infrastructure table.

Potraz measured Starlink’s consumption from the moment it arrived. It has not yet worked out how to count what Starlink provides. In the gap between those two things lies a year’s worth of sector data that cannot be read in sequence.