In the space digest of September 13, a news item flashed by that everyone swallowed as an anniversary milestone: Falcon 9 launched the final trio of O3b mPOWER satellites for operator SES, and it was the 700th launch of the Falcon family rocket in history. Seven hundred, 687 nines plus 13 heavies, the 661st first-stage landing — beautiful numbers, everyone discussing SpaceX's record.
But the name itself — O3b — no one decoded. And behind it hides one of the most bitterly ironic stories in the satellite industry. O3b stands for "Other 3 billion" — the "other three billion" inhabitants of Earth who in 2007 had no decent internet. The project was built as a social revolution: Google invested money, the founder dreamed of "leveling the playing field of the global economy," the satellites were supposed to gift broadband to the poor of Africa and Pacific islands.
Today the constellation is fully assembled — and its main client is not a poor farmer from Rwanda but the U.S. Space Force and the Pentagon, buying its channels for "the most demanding military missions." How an idealistic "internet for all" project turned into protected military infrastructure in fifteen years — that's the route of our investigation.
Greg Wyler is not your typical satellite magnate, but rather an obsessed telecom engineer with a mission. In 2003 he came to Africa and over several years connected more than 200 schools to the internet, bringing 3G and fiber to homes on the continent for the first time. In 2006 his company Terracom bought out 99% of Rwanda's national operator Rwandatel — and tried to connect a country that had just digested genocide, where copper pairs barely existed anywhere.
The project ended in a rollback: in 2007 the Rwandan government bought the operator back, and Wyler took away the main lesson from the failure. Pulling fiber and building towers in poor countries is infinitely long and expensive — you need a way to leap over all this infrastructure entirely. Thus the idea of a satellite network was born: not geostationary like everyone else, but special, "intermediate." In 2007 Wyler founded O3b Networks.
The name itself was a manifesto. "Other three billion" — three billion people living beyond the reach of normal broadband access. By Wyler's calculations, what they lacked wasn't websites or devices, but what telecom people call backhaul — the channel by which the signal from a tower reaches the backbone network. "When you type google.com on your phone, the signal goes from the phone to the tower. From the tower it has to get to Google's servers. And most of the world doesn't have access to that channel," he explained to BBC before the first launch.
What technically distinguished O3b from everyone before it? Altitude. Regular telecom satellites hang on geostationary orbit at 36,000 km — motionless above a point on the equator, but at the cost of monstrous delay: the signal round trip takes 500–600 ms, and a phone conversation through GEO is heard with that characteristic "oceanic" lag. O3b chose a circular orbit around 8,063 km — four times closer, with a delay around 120–130 ms. That's already a level where video calls and even text-based games feel "like fiber": Wyler himself swore that his network "competes bit for bit with fiber."
The price for proximity is dynamics. At this altitude satellites don't hang in place but circle the planet, and antennas on the ground have to mechanically "track" them across the sky, handing off the stream from a departing craft to an incoming one. But the whole construction worked within the ±45° latitude band — "primarily the developing world," as SpaceNews delicately phrased it in 2013.
They raised serious money for the project: total budget estimated at $650 million, investors including Google, Liberty Global, and HSBC. The internet giant was seriously trying to "connect the planet" then and invested €43 million in this attempt. They ordered the craft from Thales Alenia Space, the first twelve at 700 kg each, with 10 beams at 1.2 Gbps each across seven regions. Amusing historical detail: the Ka-band frequencies the O3b was built for had to be seized in a long regulatory battle — they belonged to the dead Teledesic project, which in the 90s promised "internet in the sky" on 840 satellites and died without launching a single one. The dead project's spectrum went to the new dream.
The first launch took place June 25, 2013: European Soyuz ST-B lifted the first quartet of craft from Kourou to equatorial orbit. The new network's first subscriber — symbolically — the Cook Islands in the Pacific, a state that submarine fiber never reached at all. By 2013 O3b had attracted $1.3 billion and signed contracts with nearly two dozen operators and providers, mostly from developing countries.
But look closely at who exactly signed these contracts. Not village internet cafes, but national telecom operators buying satellite channels into the backbone of their networks — to distribute it to people via cheap 3G mobile service. This is an important architectural compromise: O3b from the very beginning sold not "internet to the poor person's home" but wholesale channels to operators. The poor person got mobile internet, but their operator paid for it — at a commercial rate.
And then the business showcase turned toward where there's paying demand. Cruise liners Royal Caribbean in 2014 presented "boatloads of bandwidth" on the flagships Oasis and Allure via O3b: a tourist on the upper deck suddenly got video streaming in the middle of the ocean. Then oil and gas platforms, aviation, seaports. The 130 ms latency and "fiber quality" were bought not by poor countries but by rich industries for whom the geostationary 600 ms was no longer enough.
Wyler, however, was no longer with this project. Back in 2012 he founded OneWeb — with almost the same mission of "internet access for all," but in low orbit, where craft fly 20 times closer than MEO, and latency drops to tens of milliseconds. SES, the Luxembourg operator giant that had been an O3b investor since 2009, simply exercised its option in 2016 and bought out the company entirely. The founder left his own brainchild — and went to build its LEO version from scratch.
Then everything went according to a scenario that now looks like an industry textbook. OneWeb assembled the contours of the first big LEO constellation after Starlink — and in March 2020 collapsed into Chapter 11, unable to find money to finish the build. The asset was bought out by a pair consisting of the British government and the Bharti conglomerate, Wyler was ejected from the company without ceremony. The winner of the race was an old acquaintance: Elon Musk's Starlink, launching the same "real" low-orbit idea with a conveyor of its own rockets.
Wyler arranged a third attempt: in 2021 he founded E-Space — a network of satellites "for governments and business." Full circle: the man who started with schools in Rwanda is now building space for clients with budgetary shells. Awards received along the way are worthy — from the Arthur C. Clarke Award to the French Legion of Honor — but the "other three billion" never connected to his third startup.
While Wyler was trading dreams across orbits, SES under the "O3b mPOWER" brand ordered from Boeing a second generation of its medium orbit: BSS-702X platform, fully electric propulsion system, phased arrays and software-controlled beams that can be reconfigured from the ground in real time. On paper — hundreds of gigabits per craft and a "multi-terabit" constellation; SES in 2017 promised ten terabits. They signed the launch contract with SpaceX back in September 2019, planning to start in 2021.
What followed is a chronicle of drama worthy of a separate novel. The service debuted not in 2021 but in 2023: the first two craft departed on Falcon 9 only December 16, 2022. And in October 2023 SES admitted what it feared to admit: the first four craft have an electrical problem, worse than thought, which significantly cuts both throughput capacity and the calculated lifespan of the satellites. They launched the fifth and sixth anyway in November 2023 — so as not to miss the service start date: a worse-performing satellite is better than an absent one.
The seventh through eleventh craft Boeing was already reworking with the defect in mind, and the constellation was expanded from eleven to thirteen machines — with two additional ones. The "reassembled" seven and eight shipped only in November 2024, then launches went in pairs in December 2024 and July 2025, and the finale — three pieces September 13, 2026. Electric engines lift the satellite from 8,000-kilometer orbit slowly, so SES promises full combat readiness of 13 craft only by mid-2027. The 1,900 kg satellite, however, already became a record-holder in another dimension: at its launch Falcon 9 flew its anniversary 700th mission of the family, and first stage B1080 after its 29th flight neatly landed on the barge — the 661st landing in company history.
And now the most interesting part — who's buying the constellation assembled with such difficulty. Even at the launch of the first mPOWER craft in December 2022, SES head Steve Collar directly told journalists that the new satellites are "aimed primarily at government and defense markets." And there's a technical reason for this, not just marketing: the network's open architecture allows countries to deploy their own gateways — "sovereign" channels that don't pass through someone else's commercial infrastructure. Governments, according to Collar, want "full control over the network and access to multiple layers — GEO, MEO, and LEO."
In April 2023 the U.S. Space Force budgeted $59 million for fiscal 2024 to purchase O3b mPOWER services — the first American military contract for this constellation, formalized through NATO's new agreement on joint procurement of satellite communications. In September 2024 the NSPA agency signed a multi-year MEO Global Services contract with SES: protected low-latency communications for governments and militaries of all alliance countries, with an option for national sovereign networks and coalition operations. The president of Boeing Satellite Systems reported on the constellation's completion in terms that no longer contain a trace of the "three billion": the system "has proven effectiveness on the most demanding missions of our commercial, joint, and allied military customers."
So the battle for the dead Teledesic's spectrum ended in a complete archaeological loop: frequencies reserved for "internet in the sky for all" today serve NATO tactical units.
How many of those very "uncovered three billion" remain today? According to ITU data for 2025, approximately 6 billion people are online — 75% of the planet's population; 2.2 billion remain offline, meaning the "uncovered" figure of three billion has almost played back down over seventeen years. But the structure of the offline is revealing: 96% of the disconnected live in low- and middle-income countries, 94% of residents of rich countries use the internet and only 23% of the poorest. Mobile broadband access is unaffordable in price in about 60% of low- and middle-income countries.
Also revealing is how the satellite internet industry distributed its attention. Starlink, the main "winner" of this race, by SpaceX's IPO in 2026 reported 10.3 million paid subscriptions — in a year they doubled — with average revenue of $66 per month per subscriber. The "connectivity" business brought SpaceX $11.3 billion in revenue for 2025 — 60% of the company's entire revenue. Satellite internet came to pass, became an industry the size of a small country — but its clients are suburban homes in developed countries, ships, planes, and armies, not the poorest layers of the planet. Even the drop in average price from $86 to $66 is explained not by care for the poor but by international expansion into countries with lower purchasing power.
The O3b approach of "selling channels to operators, not to people" was perhaps the only economically correct one — and precisely for that reason it imperceptibly reforged the social mission into a commercial one. An island nation operator pays the same for a channel as a cruise line — and chooses the same supplier for the same reasons.
I see three layers here. The first is purely engineering and almost unconditional: O3b's physics was right from the very beginning. Medium orbit really turned out to be the golden middle between the dead 600 ms of geostationary and the expensive density of low orbit, and mPOWER's software-controlled beams foreshadowed what the entire industry boasts about today. The founder Wyler also proved to be a visionary: fifteen years before mass satellite internet he understood that latency is not a "technical detail" but the main barrier.
The second layer is economic, and it's bitter. The "other three billion" couldn't become a market for anyone: not for idealist Wyler, not for Google, not for Starlink. Communications for the poorest remained a byproduct — cheap mobile telephony that's sold to them through national operators. Satellite broadband matured where people pay dollars per month for it: on liners, in military headquarters, and in suburban homes in California. A project born from the word "equality" grew into contracts with the Pentagon — not because of someone's evil will, but because a technology's trajectory is determined by who pays, not by who needs it.
The third layer is the most uncomfortable. We're used to measuring "connectivity progress" by the percentage connected, and the percentage, honestly, isn't bad. But ITU in 2025 introduces the term "meaningful connectivity" — because the gap has shifted from "have/have not" to "fast and cheap/slow and expensive." A user in a rich country generates eight times more mobile traffic than in a poor one, 5G there is available to 84% of the population, and here — 4%. Wyler's "other three billion" really did shrink to 2.2 billion — but those who remained offline now sit not on the other side of the digital divide, but at its other depth. And no orbit by itself fills in that depth.