Free internet for more than 5,000 rural schools sounds like the kind of deal a government should sign before the Minister reaches the parking lot. Starlink’s pitch promises exactly that: a R500 million injection to beam connectivity into classrooms that fibre and mobile towers have never reliably reached, touching roughly 2.4 million learners a year. For a headmaster in the Eastern Cape trying to download a matric exam paper on a 2G signal, or a Limpopo science teacher who cannot stream a five-minute demonstration video, the promise feels like oxygen. But the paperwork is not moving, and the bottleneck might be doing exactly what it was designed to do.
The hold-up is legal, not technical. To offer internet on South African soil, Starlink needs an Individual Electronic Communications Network Service licence from ICASA. That licence carries a hard requirement: 30% of local equity must sit with historically disadvantaged groups. Starlink has shown no appetite for relinquishing that slice of its operations. Instead, it wants the R500 million school spend to count toward its empowerment obligations, treating infrastructure gifts and education subsidies as a substitute for ownership transfer. Regulators have refused. The offer sits in limbo while both sides stare each other down.
The school offer on paper
Starlink’s proposal targets the exact places terrestrial networks have written off as uneconomical. Low-earth orbit satellites do not need trenching, wayleaves or tower permits; they simply need a clear view of the sky and a dish on a roof. For remote schools where a teacher’s WhatsApp message takes three minutes to leave the phone, that speed transforms daily operations. The company wants to supply it free, removing the cost barrier entirely. On the surface, it is the most generous consumer offer to hit the telecoms space in years. A generous consumer offer can still be a raw sovereignty deal.
The ownership test
To legally sell, or even give away, internet access here, Starlink must satisfy the Electronic Communications Act and the sector’s B-BBEE codes. The threshold is specific: three in every ten shares of the licensed local entity must belong to historically disadvantaged South Africans. This is a statutory gate, not a guideline or a preference. Starlink has proposed an alternative. It would rather spend R500 million on rural education and infrastructure than transfer a single share. ICASA’s position is that the law does not allow barter. Ownership is ownership; a school router is not a proxy for local control, and a donation cannot vote in a boardroom.
Who controls the traffic
Scepticism runs deeper than standard licence wrangling. SpaceX, Starlink’s parent, counts Palantir among its financial backers. Palantir’s primary business is data analytics for intelligence agencies and defence departments. Federal contracts supply the bulk of SpaceX’s income, paying for launches and military work that dwarf any consumer subscription revenue. For critics, the customer list is a warning that harvested user information may be worth far more than monthly fees. A classroom full of learners browsing Wikipedia today is a dataset tomorrow, and the operator’s view is unambiguous: data autonomy is not a luxury, it is a right.
Starlink’s privacy disclosures acknowledge collection of usage patterns, device identifiers, location traces and communication metadata. Under POPIA, that information belongs to the people who generate it. But if processing and storage happen in foreign jurisdictions, local law becomes a paper shield. Without a domestic ownership bloc that can demand audits, sue for breaches and vote on data governance, there is no one in the room to enforce those rights. The 30% rule was written to spread economic benefit. In this case, it doubles as the only structural barrier between 2.4 million school users and an unmonitored pipeline offshore.
The stalemate continues
Letting a multinational substitute capex for equity because it built a compelling school programme would set an expensive precedent. The next firm could arrive with an identical playbook: promise connectivity, ignore ownership rules, and treat local data protection as a polite suggestion. This risk means the current gridlock is doing more than annoying officials and frustrating rural principals. It is the only mechanism currently holding the line against an operator that wants to unilaterally define who controls national data access.
For now, neither side is shifting. ICASA maintains that an I-ECNS licence requires actual equity, not creative accounting. Starlink continues to offer classrooms instead of shares. The R500 million remains a line item in a proposal, not a router on a wall. For a nation that does most of its living, learning and transacting through phones, the question of who owns the pipe has never been academic. Until someone blinks, the delay is the only firewall on offer.
