The Satellite Internet War: Starlink versus Project Kuiper versus China’s Thousand Sails

Three Constellations, Three Visions

For most of the internet’s history, getting online meant connecting to a wire in the ground. Fibre, cable, DSL, even dial-up — the assumption was that data traveled through physical infrastructure. Satellite internet existed, but it was slow, expensive, and suitable only as a last resort. The physics of geostationary orbit — at 35,786 kilometres above Earth — imposed a minimum latency of roughly 600 milliseconds, making video calls painful and online gaming impossible.

Low Earth orbit (LEO) changed everything. By flying satellites at 340 to 1,200 kilometres, operators can deliver latency comparable to ground-based broadband — 20 to 40 milliseconds in ideal conditions. The catch is that LEO satellites don’t hover over a fixed point. They zip across the sky at 27,000 kilometres per hour, completing an orbit every 90 minutes. Covering the planet requires thousands of them, working in concert, handing off connections between satellites like a cellular network in the sky.

Three major players are now racing to build these constellations, and the outcome will shape who gets internet access, on what terms, for the next generation.

Starlink: The Incumbent Nobody Can Catch

SpaceX’s Starlink is the undisputed leader, and the numbers are staggering. As of early 2025, Starlink operates more than 6,000 active satellites — roughly 60% of all operational satellites in orbit. The constellation serves over 2.7 million subscribers across more than 70 countries, generating an estimated $6.6 billion in revenue in 2024, according to industry analysts at Quilty Space.

Starlink’s advantage isn’t just scale; it’s vertical integration. SpaceX manufactures Starlink satellites at its facility in Redmond, Washington, and launches them on its own Falcon 9 rockets — often 20 to 23 at a time — at an internal cost far below what competitors pay for launch services. A Falcon 9 launch costs SpaceX an estimated $15-20 million internally. A competitor paying for a ride on someone else’s rocket might spend $60-70 million for similar capacity.

The economics are starting to work. Starlink’s standard residential service costs $120 per month in the US for unlimited data, with a one-time hardware fee of $599 for the user terminal. The company claims to be cash-flow positive as of late 2023, though detailed financials aren’t public. Speed tests compiled by Ookla show median US download speeds of 90-100 Mbps, with latency averaging 40-50 ms — not gigabit fibre territory, but more than adequate for streaming, browsing, and video calls.

Starlink’s user terminal — the “Dishy” — has evolved through several generations. The current v4 terminal is a flat rectangular panel roughly the size of a pizza box, simpler to manufacture and more capable than earlier versions. SpaceX claims the production cost per terminal has fallen below $600, which makes the hardware economics sustainable for the first time.

Project Kuiper: Amazon’s Slow-Building Heavyweight

Amazon’s Project Kuiper is Starlink’s most credible competitor, at least on paper. The company has FCC approval for 3,236 satellites and has committed over $10 billion to the project. Kuiper’s stated goal is to deliver speeds of 100 Mbps to 1 Gbps, competitive with terrestrial broadband.

But Kuiper has a timing problem. As of early 2025, it has launched exactly two prototype satellites — both on United Launch Alliance’s (ULA) Atlas V rocket in October 2023. Production satellites will fly on a mix of ULA’s Vulcan Centaur, Arianespace’s Ariane 6, Blue Origin’s New Glenn, and eventually SpaceX’s Falcon 9. The irony of Amazon — Jeff Bezos’s company — paying Elon Musk’s SpaceX to launch Kuiper satellites hasn’t been lost on anyone.

The FCC requires Kuiper to have half its constellation (1,618 satellites) in orbit by July 2026. At current launch rates, that deadline looks challenging. Amazon has ordered more than 80 launches across its launch providers, but the launch industry’s capacity constraints are real. Vulcan has flown only a handful of times. Ariane 6’s maiden flight was delayed until mid-2024. New Glenn hasn’t flown at all.

Amazon’s user terminal strategy is ambitious. The company has shown off a terminal design smaller and cheaper than Starlink’s, with projected manufacturing costs below $400. Amazon’s consumer hardware expertise (Kindle, Echo, Fire) and its global logistics network could become significant advantages if Kuiper achieves operational scale. But “if” is doing a lot of work here.

China’s Thousand Sails: The State-Backed Contender

The third competitor is fundamentally different from the first two. China’s Thousand Sails constellation — also known as the G60 Starlink — is a state-backed initiative to deploy more than 12,000 satellites, managed by Shanghai Spacecom Satellite Technology. The first 18 satellites launched in August 2024 on a Long March 6A rocket, and China has since accelerated the cadence.

China’s motivations are strategic as much as commercial. Starlink has proven militarily valuable in Ukraine, where it has provided resilient communications that Russian forces have struggled to disrupt. Beijing views reliance on an American-controlled satellite internet constellation as an unacceptable vulnerability. Thousand Sails, along with the smaller Guo Wang (“National Network”) constellation, is China’s answer — a sovereign internet infrastructure that operates on its own terms.

The scale of China’s ambition is enormous. The country’s space industry launched 67 rockets in 2023 and exceeded that in 2024, largely driven by commercial launch providers like Galactic Energy, LandSpace, and iSpace alongside the state-owned CASC. China’s new commercial spaceports — including facilities in Hainan and near the Gobi Desert — are designed to support rapid launch cadences.

But challenges abound. Frequency coordination in an increasingly crowded LEO environment is contentious. Starlink and Kuiper have secured priority spectrum rights through the International Telecommunication Union (ITU). China will need to navigate a regulatory landscape shaped by the first movers. And the user terminal — the consumer-facing piece — remains a significant manufacturing challenge that few companies have solved at scale.

The Capacity Question: How Many Satellites Are Too Many?

Space is big, but low Earth orbit is getting crowded. Starlink satellites performed more than 50,000 collision-avoidance maneuvers in the first half of 2024 alone — double the rate from the previous year. Each maneuver consumes propellant and incrementally shortens the satellite’s operational life. And while collisions are rare, they’re not hypothetical: a near-miss between a Starlink and a OneWeb satellite in 2021 came within 60 metres, and a Chinese complaint to the United Nations about Starlink “close approaches” to its Tiangong space station in 2021 highlighted the diplomatic dimension of orbital congestion.

Astronomers are increasingly frustrated. Starlink’s “VisorSat” design and brightness mitigation efforts have reduced but not eliminated the streaks that satellite constellations leave across telescope images. The International Astronomical Union has warned that large constellations threaten ground-based astronomy, particularly wide-field surveys like the Vera C. Rubin Observatory’s Legacy Survey of Space and Time, which begins operations in 2025.

Yet despite these concerns, the satellite internet race is accelerating, not slowing. The addressable market is enormous: roughly 2.7 billion people lack internet access, according to the ITU, and hundreds of millions more have slow or unreliable connections. Satellite internet won’t serve everyone — dense urban areas will always be better served by fibre and 5G — but for rural communities, ships, planes, emergency responders, and anyone beyond the reach of terrestrial infrastructure, LEO constellations are the most viable solution since the internet was invented.

Who Wins?

Starlink’s first-mover advantage is enormous, and SpaceX’s launch dominance creates a moat that competitors will struggle to cross. But monopoly isn’t destiny. Kuiper has Amazon’s balance sheet and consumer hardware expertise. China’s state-backed constellations have national security imperatives that don’t depend on commercial viability. OneWeb (now part of Eutelsat) and Telesat’s Lightspeed add additional competition.

The most likely outcome is a oligopoly — Starlink as the dominant global player, Kuiper as a strong second in Western markets, and China’s constellations serving China and its strategic partners. The real winners will be the hundreds of millions of people who get internet access for the first time, or who upgrade from connections measured in kilobits to connections measured in megabits. The satellite internet war may be fought with rockets and regulatory filings, but its dividend is connectivity — and that’s worth fighting for.

The Regulatory Battles That Will Shape the Race

Orbital real estate is allocated through the International Telecommunication Union, and the first movers have locked in the best spectrum. Starlink’s filings cover tens of thousands of satellites across multiple frequency bands, giving SpaceX a formidable regulatory moat. Competitors argue that the ITU process was never designed for mega-constellations and that existing filings effectively allow first movers to hoard orbital slots and spectrum. SpaceX counters that it’s actually launching satellites while competitors are filing paperwork.

The Federal Communications Commission in the United States has been a crucial gatekeeper. The FCC rejected Starlink’s application for $886 million in rural broadband subsidies in 2022, finding that the service’s speeds didn’t meet the program’s requirements — a decision that angered SpaceX but demonstrated that regulators are willing to scrutinize even the most prominent players. Meanwhile, the FCC has approved Kuiper’s orbital debris mitigation plan, clearing a key hurdle for Amazon.

In Europe, the European Union’s IRIS² constellation — a €6 billion public-private initiative to build sovereign European satellite internet — reflects the same anxiety that drives China’s Thousand Sails: the recognition that internet infrastructure is strategic infrastructure, and that ceding it entirely to American companies carries geopolitical risk. The result is that the satellite internet race isn’t just commercial competition; it’s a proxy for great-power competition, with the United States, China, and Europe each backing their own sovereign constellations.

The Debris Problem Nobody Has Solved

There are now roughly 10,000 active satellites in orbit, up from about 2,000 in 2018. At the end of their operational life — typically 5 to 7 years for LEO satellites — these spacecraft must be deorbited, burning up in the atmosphere. Starlink satellites are designed to fully demise on re-entry, but the sheer volume of deorbits raises questions about atmospheric pollution. A 2024 study in Geophysical Research Letters found that satellite re-entries are depositing significant quantities of aluminum and other metals into the upper atmosphere, with unknown long-term consequences for atmospheric chemistry and climate.

Kessler Syndrome — the nightmare scenario where collisions create cascading debris that makes orbital space unusable — remains a theoretical risk, but it’s a risk that grows with every launch. The space sustainability community argues for stricter deorbit timelines, collision avoidance standards, and active debris removal. The commercial industry, focused on deployment deadlines, has been slower to engage. This tension will define the second decade of the LEO constellation era.