A satellite map can make the whole argument look settled. Starlink’s constellation passes overhead, so South Africa might seem technically covered and commercially ready. This is an easy mistake. ICASA’s position, confirmed in June 2026, is that a provider selling directly to local users needs the right licences first, and those licences are the real gatekeeper.
The licensing stack slows everything down. Before a single dish can be sold or supported here, the operator needs an Individual Electronic Communications Service licence, an Individual Electronic Communications Network Service licence, and the relevant radio-frequency spectrum licences. Without them, physical reach is just physics; it does not make the service lawfully available to retail customers.
The licence stack
ICASA’s June notice drew a hard line between signal coverage and legal operation. A satellite constellation that wants to sell internet access directly to users needs three separate approvals, not one broad blessing.
The first is the Individual Electronic Communications Service licence, which covers the service sold to end users. The second is the Individual Electronic Communications Network Service licence, which covers the network itself. The third layer is spectrum, because satellite links still depend on assigned frequencies for uplink, downlink, and user equipment.
That middle licence is the awkward one. ICASA says it cannot simply take a fresh I-ECNS application whenever a new entrant turns up. A new licence of that kind needs the proper policy basis and an Invitation to Apply process. The regulator does not run an open door for this category.
A prospective entrant may try to buy or otherwise acquire an existing individual licence, yet that still needs regulatory approval before it becomes usable. The paperwork may move faster than a full new allocation, but the regulator still has to bless the transfer.
What Starlink has not done
As far as the government’s latest stated position goes, Starlink has not applied for the licences ICASA says are required. That includes the I-ECS licence, the I-ECNS licence, and the spectrum licences that would sit underneath both.
That leaves a familiar confusion in place. People see the satellites and assume the company is ready to sell. ICASA’s view is the opposite. Coverage over the country does not equal lawful retail access for South African customers. A signal footprint is not a trading licence.
The gap is especially obvious when the discussion drifts toward the Karoo farm example. The question is not whether a terminal can lock onto a satellite overhead. It is who is allowed to sell the service, support the customer, and do so under the country’s telecom rules. This is a regulatory question, not a map question.
What still has to happen
Even if the core licences were in hand, the job would not be finished. ICASA’s June notice points to several extra steps that follow the licence question.
Equipment needs type approval. User terminals, ground station gear, and associated network hardware have to meet local technical standards and avoid causing interference.
Spectrum assignments still need to be settled in practical terms. A licence does not remove the need for frequency coordination for actual deployments.
Lawful interception obligations also sit in the picture. A provider operating here has to be able to comply with interception and information-provision rules under RICA. This usually pushes operators toward local operational arrangements, as the state wants a service it can regulate and, where required, intercept in lawful circumstances.
Then there are the commercial mechanics that make a service real rather than theoretical. Billing, customer support, tax compliance, and local support arrangements all matter once a provider is trying to serve end users in the market.
Equity talk is not the licence requirement
One reason this argument keeps getting muddied is the separate debate over equity-equivalent transformation measures. Those proposals are political and policy-level discussions. They are not the same thing as the licences ICASA says a satellite provider must hold.
This distinction changes the problem from symbolism to procedure. A company can talk about transformation frameworks for months and still not have the authorisations needed to sell a connection. ICASA’s licensing demands sit in one lane. Equity-equivalent proposals sit in another. Mixing them up only helps people dodge the real issue.
The local market is already shrinking
The timing is not flattering for anyone hoping to sell satellite broadband as a clean growth story. Recorded satellite-broadband subscriptions in 2025 fell from 13,667 to 10,597, a drop of 3,070 connections.
That decline leaves a market that is still alive but clearly under pressure. Vox Telecom remains in the field, using platforms such as YahClick and Eutelsat Konnect. Morclick is also active, again using Eutelsat Konnect capacity. These are the local names already doing the slow, difficult work of serving remote customers where fibre and mobile coverage are weak or unreliable.
The broader picture is not mysterious. Satellite remains useful for farms, lodges, and businesses that sit outside dependable terrestrial networks. But the numbers suggest that the old model is not booming. People are voting with their subscriptions, and the direction is down.
What to watch next
The decisive question is whether Starlink moves from a coverage story to a compliance story. That would mean dealing with the licence sequence ICASA requires, then the technical approvals, then the operational pieces that make retail service lawful.
Until that happens, the legal barrier is doing more work than the orbit. The satellites may pass overhead. The harder part is getting permission to sell the connection underneath them.
