China Just Added 27 Satellites to Its Starlink Rival. The Arctic Test Was the Real Story.

At 3:27 a.m. Beijing time on October 10, a Long March 12 rocket lifted off from the Hainan Commercial Space Launch Site and placed 27 internet satellites into low-Earth orbit. It was the 27th batch of spacecraft deployed for China's national satellite internet project, and it brought the total number of operational satellites in the network to 222.

The launch barely made headlines outside space trade media. But the number that matters may not be 27, and it may not be 222. During China's 16th Arctic scientific expedition this year, the same constellation completed in-orbit cross-polar communication trials, holding real-time video calls across the polar regions with end-to-end latency under one second and peak transmission speeds reaching 30 megabits per second. A sovereign satellite network that can carry video across the poles is a network that is no longer just a launch schedule. It is a working system.

The launch, by the numbers

The October 10 mission was the ninth flight of the Long March 12, a 62-meter rocket that is China's largest single-core launch vehicle. Developed by the Shanghai Academy of Spaceflight Technology, it can deliver at least 12 metric tons to low-Earth orbit and six tons to a 700-kilometer sun-synchronous orbit. A streamlined horizontal assembly, testing, and transport workflow is designed to cut launch pad preparation times, which is exactly the kind of operational detail that matters when your job is to fill a constellation measured in thousands.

The mission also marked the 671st flight of the Long March rocket family and the 24th launch from the Hainan Commercial Space Launch Site, the facility that has become the primary hub for China's high-frequency constellation deployments. The site itself was constructed in 878 days, and Saturday's launch was the tenth from its Launch Pad 2.

And the cadence is accelerating. The 26th satellite batch flew on a Long March 8A on September 23, the 25th flew on a Long March 12 on September 17, and the 24th flew on August 16. Three numbered groups in less than two months. Independent launch tracking puts China's 2026 deployment goal at around 310 satellites for this constellation, with roughly 900 planned for 2027 and as many as 3,600 per year thereafter. If that cadence holds, the network reaches its full planned size of about 13,000 spacecraft sometime in the early 2030s.

A sovereign satellite network that can carry video across the poles is no longer just a launch schedule. It is a working system.

What the GW constellation actually is

The network is formally China's national GW megaconstellation, operated by China Satellite Network Group, a state-owned enterprise backed by the Chinese government. It is commonly called Guowang or StarNet. The published plan calls for 12,992 satellites, which would make it the second-largest megaconstellation design on the planet, after Starlink.

The architecture is layered. The GW-A59 sub-constellation, stationed below 590 kilometers, provides high-capacity, low-latency broadband services. The GW-A2 sub-constellation, deployed at 1,145 kilometers, handles wide-area coverage and system redundancy. Both orbital tiers use Ka-band payloads and inter-satellite laser links, the same two technical pillars that define the modern generation of broadband constellations.

The spacecraft themselves are built by the Innovation Academy for Microsatellites of the Chinese Academy of Sciences. The small satellite platform weighs roughly 695 kilograms and uses electric propulsion to maneuver in orbit, powered by two solar panels. A larger platform, about 30 percent bigger and weighing up to 1,000 kilograms, flies on the Long March 5B and Long March 6A. For now, the operator's stated priority is services focused on China, with worldwide internet services as the longer-term goal.

Why the Arctic test matters more than the launch

Launching satellites is logistics. Getting them to carry a coherent broadband service across the planet's poles is engineering, and it is the part that is hard.

Polar regions are notoriously difficult for satellite coverage. Geostationary satellites, parked over the equator, cannot reach them at all, which is why polar communications have historically relied on expensive, low-bandwidth links. A low-Earth orbit constellation with polar coverage fills that gap, but only if the inter-satellite links, handoffs, and ground infrastructure all work together at full operational tempo. A video call with under one second of end-to-end latency across the Arctic is evidence that the constellation's laser-linked mesh is doing what its architecture promises.

There is also a geopolitical dimension that is impossible to ignore. The Arctic is becoming one of the most contested regions for communications infrastructure: new shipping routes, scientific stations, and military presence all need connectivity that does not depend on undersea cables or someone else's satellites. A Chinese constellation demonstrating polar broadband in 2026 is a statement of capability aimed at exactly that audience.

The bigger picture

The global satellite internet race now has two clearly defined lanes. In one, SpaceX's Starlink continues to launch at an unmatched pace: independent tallies credit SpaceX with 114 successful launches in 2026 alone, roughly tied with the entire rest of the world combined, with China the clear second at 61. In the other, China is building not one but multiple state and commercial constellations (GW alongside the Shanghai-backed Qianfan/Spacesail project) with launch cadences that are finally starting to look serious.

The important question for the rest of the world is not who launches more rockets. It is whose network becomes the default for countries that want broadband without dependence on an American company. Sovereign constellations are, in this sense, a form of infrastructure diplomacy: the hardware is the message. Every successful batch, and especially every demonstrated capability like the Arctic video calls, makes the pitch easier for countries deciding which system to buy into.

222 satellites out of 12,992 is 1.7 percent. The network is a rounding error on its own plan. But it is a working rounding error, getting bigger every few weeks, and it has now proven it can connect the hardest-to-reach places on Earth. That is worth paying attention to, even at 3:27 in the morning.