Amazon Leo, the retail giant's telecommunications division, has filed a request with the U.S. Federal Communications Commission (FCC) to place up to 5,105 new satellites into space. The purpose of this initiative is to establish a low Earth orbit internet network designed to provide services directly to mobile devices.
This project aims to serve users located in isolated regions or outside the range of conventional phone towers. Implementation of this constellation is scheduled to begin only in 2028, and to operate, the company plans to utilize the frequency spectrum belonging to Globalstar.
An important step occurred in April of this year when Amazon completed a merger agreement that fully transferred all infrastructure, licenses, and satellite mobile service operations from its partner to Amazon.
The new ecosystem will be integrated with the broadband constellations the company already maintains in orbit. The main advantage of Amazon Leo's Direct-to-Device (D2D) system lies in its simplicity of use for the end consumer. The network will connect to any compatible mobile device, utilizing the satellite communication chips present in current smartphones, without the need to install additional antennas.
In practice, this service will function as a complement to terrestrial operators. If a user enters a shadow zone and loses terrestrial signal, their device can access the space network to maintain active communication, which includes everything from sending messages to receiving emergency alerts and disaster response.
To ensure continuous coverage, the company's space infrastructure will be segmented into multiple orbital layers. Some of these layers will focus on mid-latitudes to serve more densely populated areas, while others will have high inclination to ensure the service reaches the poles and high latitudes of the planet.
Communication between satellites and cell phones will occur through radio frequencies in the L and S bands. To reconnect this space network to the traditional internet, the equipment in orbit will use high-capacity links in the Ka and V bands, establishing contact with Amazon stations located on Earth.
This technology already has guaranteed demand, as Amazon signed an agreement with Apple to provide connectivity for iPhones and Apple Watches. This agreement ensures support for features such as satellite Emergency SOS, message sending, updates to the Find My feature, and roadside assistance activation.
The expansion of Amazon Leo places the company in direct competition with SpaceX, which filed a similar request with the same FCC earlier this year to launch a massive constellation of one million satellites. Although the focus on mobile phones is planned for 2028, Amazon's broadband infrastructure continues to advance.
Currently, the company operates with over 390 first-generation satellites. Furthermore, the fixed service intended for residences and businesses is expected to begin commercial operations by the end of the current year. The Brazilian market is part of this global strategy, where the service should compete with Starlink in terms of high speed for remote areas of the country, although this fixed connection requires specific receiving antennas, such as the Leo Ultra, unlike smartphone access.
Elon Musk's satellite broadband service Starlink, which has over 10,000 satellites and millions of customers in dozens of countries, is actively operating almost across all of Southern Africa, including Mozambique, Malawi, Madagascar, Botswana, Zimbabwe, Zambia, Eswatini, and Lesotho. However, this week, a competing group, Amazon Leo, received the route to South Africa, despite having only 390 satellites and no commercial customers anywhere in the world yet.
The internet provider Herotel, owned by Maziv, an operator of fiber optic and wireless networks, will distribute Leo satellite broadband under the new brand Evry, with a commercial launch expected in 2027. Thus, Amazon has overtaken Starlink. This success was achieved not through technological superiority, large financial reserves, or regulatory privileges, but because Amazon viewed the South African licensing regime as an engineering problem to be solved, while Starlink waited long and loudly for rule changes.
The rules applicable to both companies are identical: to obtain a network license in South Africa, an operator must own at least 30% of historically disadvantaged groups. SpaceX refused to give up a stake in its local operations, and Musk publicly disputed this requirement for years, calling it racially discriminatory. He argued that he could not launch Starlink in the country because 'I am not Black,' turning the licensing issue into a diplomatic and political spectacle. As a result, Starlink remains unavailable in Africa's most advanced broadband market, despite thriving in almost all neighboring countries.
Amazon studied the same rules but reached a different conclusion. Icasa reported in a notification last month that satellite operators currently cannot obtain the necessary network licenses, and that the realistic path to the market is through an existing license holder. Amazon did exactly that: Herotel will hold the licenses, and Amazon will not. This allowed them to avoid conflict over equity, political theater, and posts on X. It turned out that the boring strategy is effective and aligns with Amazon's overall approach to South Africa, including the launch of the regional AWS in Cape Town and years of quiet preparation. Amazon views regulatory hurdles as part of the cost of doing business, whereas Musk sees them as an ideological reason.
Musk's stance led to years of headlines, a central role in tensions between South Africa and the US, but brought no revenue from a market with enormous product demand. This demand is so high that Icasa was forced to combat the smuggling of Starlink terminals through neighboring countries. Consequently, Amazon's compliance secured it a first-mover advantage: an approximate launch date (2027), a distribution partner with wide coverage in small and rural settlements where satellite broadband works best, and an active registration page. Nevertheless, the first announcement does not mean first service, as Leo has not launched commercially anywhere yet, and its debut in South Africa in 2027 leaves room for delays.
Despite this, it cannot be considered a victory for transformation policy. The licensing regime did not extract 30% equity from Amazon because Amazon (apparently) does not need a license at all. If the measure of success is broad ownership of the digital economy, then channeling foreign operators through distribution structures achieves this only formally. Therefore, it would be unwise if the government did not finalize an equivalent to equity capital. The December 2025 policy of Communications Minister Solly Malatsi, allowing multinational corporations to fulfill empowerment obligations through equivalent equity investment programs (EEIPs) instead of ownership stakes, is causing political debate, and experts warn that implementation could take years.
Starlink has already submitted a proposal worth about 500 million rand to connect 5,000 schools, plus 2 billion rand for local infrastructure. However, not every multinational corporation will be able or willing to follow the partnership model like Amazon's; some prefer to operate directly, have their own licenses, and invest accordingly. South Africa, with its stagnant economy and high unemployment, is not in a good position to deter companies actively trying to invest by insisting on one narrow form of compliance. The goal of the rules is transformation, not checkpoint control. The main lesson of this week is the temperament of corporations: in regulated markets, the scarce resource is not the satellite or the capital, but the willingness to find a way through local rules that may not be liked. Amazon, apparently, found such a way, while Musk, with a six-year head start and a product that South Africans are literally smuggling across borders, could not.
