The internet map of the world is not drawn by engineers. It is drawn by lawyers, by regulators, and by whoever is willing to sign the local paperwork.

A rocket can reach any patch of ground on Earth. A license cannot. That gap between what a technology can do and what a government will let it do is where the satellite internet war is actually being fought, and it rarely goes to whoever owns the most satellites.

For most of the past decade, that war looked settled. SpaceX (SPCX) turned Starlink into the default choice for anyone living past the last mile of cable, stacked thousands of satellites overhead, and rode that lead into the largest public offering in market history.

Rivals talked. Starlink shipped. Anywhere with open skies and open rules, Musk’s network got there first.

Anywhere with open rules. That qualifier is doing quiet, heavy work, and it is exactly where Amazon (AMZN) just found its opening.

Amazon Leo, the retail giant’s low-orbit satellite network, has signed a deal to bring internet to South Africa through Herotel, the country’s largest fixed internet service provider, under a new consumer brand called evry. It is Amazon Leo’s first agreement of this kind in Africa, and it drops the company into a market Starlink still cannot legally enter.

Amazon Leo will launch satellite internet in South Africa through Herotel in 2027.

ZINYANGE AUNTONY / Getty Images

How Amazon slipped past the licensing wall

The mechanics are the whole story. Amazon is not trying to sell satellite internet to South Africans directly. It is selling capacity to Herotel, a licensed local operator that already holds the paperwork and already runs the ground game.

Herotel serves more than 350,000 active customers across over 550 towns, with 120 local offices handling installs and support, according to Amazon. Those offices will run evry from day one.

Amazon supplies the sky. Herotel supplies the license, the trucks and the customer service. Its chief executive said the service will reach people that even fiber and fixed wireless cannot, and that with it, “it no longer matters where you live,” according to Amazon.

More Telecommunications:

That structure sidesteps the exact problem that has frozen Starlink for years. Amazon Leo’s satellites orbit roughly 590 kilometers up, close enough to carry video calls, streaming and remote work, Amazon said. Customers connect through a compact antenna, with no fiber needed at the home.

The company is not stopping at home internet, either. Amazon Leo is also working with the network firm Vanu to push cellular coverage into rural South Africa, according to Amazon. evry itself is expected to launch commercially in 2027.

What struck me when I mapped the deal against Starlink’s playbook is how differently the two companies chose to enter the same continent. Starlink wants to be the operator.

Amazon is content to be the wholesaler standing behind a local one. In a market built to protect local ownership, the wholesaler walks in the front door.

Related: AT&T may be left out of the Starlink deal everyone wants

Why Starlink still can’t crack South Africa

South Africa’s block is not technical, and it is not really a ban. It is an ownership rule.

The country’s Electronic Communications Act requires telecom license holders to be at least 30% owned by historically disadvantaged groups, a core plank of Black Economic Empowerment, the post-apartheid policy known as BEE. SpaceX has refused to hand over that equity in any market, and it has never actually filed a South African license application, according to Snopes.

Musk, who was born in South Africa, has made the standoff personal. He said Starlink cannot get a license there “because I’m not Black,” according to Snopes, and has called the ownership rules openly racist.

South African officials counter that the service was never banned and would be welcome if it followed local law. The standoff sits in what one South African telecom analysis called “a genuine stalemate,” according to FastestFibre.

There was supposed to be a way out. In December 2025 the communications minister proposed letting foreign firms invest in local projects instead of ceding equity, but the country’s regulator reaffirmed in May 2026 that it cannot bypass the ownership rule without Parliament amending the law itself, according to FastestFibre. That amendment has not been tabled.

Here is the part that reframes the whole race. While the fight played out on X, Starlink quietly switched on service all around the border. The connectivity gap Amazon just stepped into is wide, and the numbers are blunt:

  • Starlink operates in more than 20 African countries, including the neighbors Lesotho, Botswana and Zimbabwe, but not South Africa, according to Connecting Africa.
  • South Africa’s regulator reaffirmed in May 2026 that it cannot waive the 30% ownership rule without Parliament changing the law, according to FastestFibre.
  • Amazon Leo has more than 390 satellites in orbit, enough to begin initial service this year, according to Amazon.
  • Nearly a quarter of southern Africa’s population sits outside any network coverage, a gap that could unlock up to $16.9 billion in yearly economic value, according to an Access Partnership report cited by Amazon.

SpaceX has since pledged about $145.6 million in local investment to try to satisfy the rules. The paperwork still has not moved.

Amazon did not wait for the rules to change. It found a partner the rules already approved.

What the South Africa deal means for Amazon investors

For Amazon shareholders, evry will not move the needle on next quarter’s revenue. A single satellite deal with one African internet provider is a rounding error against a company Amazon’s size. The signal underneath it is the part worth pricing in.

Amazon Leo just proved it can enter a market Starlink is locked out of, and it did so with a model that travels. Ownership-restricted telecom regimes are common across Africa, Asia and the Middle East.

Every one of them is a place where a wholesale deal with a licensed local carrier beats a foreign operator demanding to run the whole show itself. If Amazon can copy the Herotel move a dozen times, the idea that satellite internet is a one-horse race stops holding up.

The prize is not small, either. More than 18 million South Africans still lack high-speed internet, and evry is aimed straight at the farms and small towns fiber never reached. That is a first-mover position in a wealthy, wired-but-unequal economy, handed to Amazon while its rival argues about principle.

My analysis of the two rollout timelines is where the honesty comes in. evry goes commercial in 2027, and Starlink’s own earliest legal path into South Africa lands around the same window if Parliament acts.

Amazon has not beaten Starlink onto South African rooftops yet. What it has done is beat Starlink to a signed deal and a clear, legal route in, while its rival has neither.

That is the quieter lesson for anyone holding either stock. The satellite race has been priced as a contest of rockets, launch cadence and hardware in orbit. This deal is a reminder that the next round turns on something far less glamorous, which is who is willing to share the upside with the countries they want to serve.

Musk has staked a principle on refusing. Amazon just showed what the other answer buys you, and it bought Africa’s most industrialized economy first.

Related: Starlink just notched a win U.S. investors should watch