BEIJING, May 2, 2026. China has 168 satellites in low Earth orbit for a constellation called Guowang, with plans to deploy 12,992 by the early 2030s. Western coverage routinely calls it China’s Starlink. The framing is wrong, and the gap between what Beijing is building and what Western analysts are describing is the story.
Starlink is a commercial broadband product the US Department of Defense bought into through a customer modification called Starshield. Guowang is the inverse. China filed it with the International Telecommunication Union in September 2020 as a state-owned project, operated by China Satellite Network Group, with a payload mix broadband alone does not require. Chinese state sources have stated on the record that Guowang carries broadband, laser communication, synthetic aperture radar, and optical remote sensing payloads. That sentence ends the Starlink comparison.
The structural read is not that China is disguising a military satellite network as a commercial constellation. It is that Beijing is building one platform for both missions. Guowang collapses communications, sensing, relay, and positioning support into a single state architecture, deployed under one launch cadence and tied to one ground segment. Broadband is the cover story. The military value sits in the rest of the stack.
What Is Visible
As of mid-April 2026, Guowang has approximately 168 operational satellites in orbit after 21 launch missions, per tracking data compiled by KeepTrack. All have gone to the higher GW-2 sub-shell at roughly 1,145 km altitude, in two orbital planes inclined at 86.5° and 50°. The lower GW-A59 sub-shell at 500 to 600 km, which would deliver the low-latency broadband most consumer comparisons assume, has not seen a single launch. The analysis that follows applies to GW-2. GW-A59 may carry a different mission profile and is worth its own piece when deployment begins.
The deployment plan calls for 310 satellites in 2026, 900 in 2027, and 3,600 per year starting in 2028. ITU rules require China to launch 10 percent of the constellation by 2029 and 50 percent, roughly 6,500 satellites, by 2032 to protect spectrum rights. Three institutions are building the satellites: the China Academy of Space Technology, the Innovation Academy for Microsatellites at the Chinese Academy of Sciences, and the commercial firm Galaxy Space. CAST has confirmed two distinct platforms exist on the same constellation, designated large and small, and has not explained the functional difference.
Launch vehicle correlation does the explaining. The Long March 8A and Long March 12 carry the smaller platform at roughly 695 kg per satellite. The Long March 5B and Long March 6A carry the larger platform at up to 1,000 kg, with some launch-mass calculations pushing higher. SpaceX’s Starlink V2 Mini operates at roughly 800 kg as a pure broadband product. OneWeb, at a similar altitude band to Guowang, weighs approximately 150 kg. Guowang’s mass class brackets and exceeds the leading commercial baseline. Independent analysis from LeoLabs Senior Technical Fellow Dr. Darren McKnight, based on radar cross-section measurements of the first deployed batch, indicates a body roughly 3 meters by 1 meter with a 10-meter wingspan when arrays are deployed, and a dry mass near 800 kg.
What Sits Underneath
The mass-budget evidence is the first piece of the proof chain. The payload admission is the second. A pure broadband constellation does not need synthetic aperture radar. It does not need optical remote sensing. The presence of those payloads in CASIC’s published capability list places Guowang in a different functional category from any operational LEO broadband network in service today.
The orbit choice is the third piece. GW-2 at 1,145 km is not where commercial broadband operators put their primary shells. Starlink operates at roughly 550 km. The 1,145 km altitude is approximately the same band as China’s Yaogan-31 RF detection and geolocation triplets, which orbit at 1,100 km and 63.4° inclination and are operated by the PLA Aerospace Force to geolocate naval forces through electromagnetic emission interception. Putting SAR and optical sensing payloads in that orbital regime gives them persistent revisit characteristics that align with strategic ISR, not low-latency broadband.
The fourth piece is institutional. The PLA dissolved the Strategic Support Force in April 2024 and split it into three forces directly under the Central Military Commission. The Aerospace Force inherited launch infrastructure, telemetry tracking and control, and military space ISR. The Information Support Force inherited space-information transmission and battlefield environment support. As of February 2026, most space information support and assurance functional groups have transitioned to the ISF. A new institution, Aerospace Information University in Jinan, exists to train ISF personnel and the commercial space sector together. There is no civil-military firewall to cross because there was never one to begin with.
The cross-link technology that turns a comms constellation into a sensor-relay-compute mesh has already been demonstrated in Chinese orbit. Laser Starcom achieved a 400 gigabit-per-second optical inter-satellite link between Guangchuan 01 and 02 in March 2024, at 640 km separation, with tracking errors under 5 microradians. The Three-Body Computing Constellation completed nine months of orbital testing in February 2026, demonstrating onboard AI processing and laser inter-satellite networking up to 100 gigabits per second. Combined with Guowang’s confirmed laser comms payload, the architectural pieces exist to operate the constellation as an integrated processing and relay network rather than a series of isolated comms pipes.
The Industrial Base Behind It
China hit 92 orbital launches in 2025, surpassing its 2024 record of 68. By the end of March 2026, China had logged 34 successful orbital launches against 29 for the United States. Guowang, Qianfan, and adjacent Chinese constellations together consumed roughly 45 launches in 2025 and are projected to consume 70 or more in 2026. Two LEO constellations are now claiming the majority of the national launch manifest, and that manifest is being expanded around them.
Reusable launch is the supply-side lever. The Long March 12A flew December 22, 2025 Eastern (December 23 UTC), reached orbit, and crashed approximately two kilometers from its landing pad during the first-stage recovery attempt. A second flight attempt is expected in 2026. The Long March 12B completed a static fire test in January 2026, with first flight planned in the first half of the year. LandSpace’s Zhuque-3 has flown twice. None of these vehicles has reached SpaceX’s roughly 600 booster landings by April 2026, but the gap is no longer categorical. It is a matter of cadence. The launch industry is being scaled around the deployment requirement, and the deployment requirement is driven by ITU spectrum deadlines that are themselves a function of state policy, not commercial demand.
Why It Matters
The United States is not absent from low Earth orbit. SpaceX operates Starlink and Starshield. The Space Development Agency is deploying the Proliferated Warfighter Space Architecture for tracking and transport layers. The National Reconnaissance Office operates its own proliferated layer for overhead intelligence collection. None of these programs was designed as a single integrated state platform fusing communications, sensing, relay, and positioning. Each was designed for a discrete mission, procured through a discrete contract, operated by a discrete agency. A patchwork of commercial products with military variants and parallel agency programs cannot be expanded, hardened, retasked, and protected as a coherent national asset. A single integrated platform under unified state ownership can. The seams in the US architecture are where Chinese counterspace planning will look for leverage.
The implication for the next decade is not that China has won the LEO race. It is that the United States and China are running different races. SDA, NRO, and Starshield are real programs at real scale. None of them is being built as the comprehensive state infrastructure layer Beijing began filing with the ITU in September 2020.
Conclusion
The LEO race is not about who launches more satellites. It is about whether the constellation was designed as integrated state infrastructure from the architecture filing forward. The United States is launching more satellites and operating multiple proliferated programs. None of them was designed to do what Guowang is being designed to do.
Western analysts have spent the last sixteen months waiting to see whether China can match Starlink’s deployment cadence. The deployment cadence is the wrong metric. The metric is what fits inside the satellite bus, what payloads share the platform, who operates the ground segment, and what the architecture looks like when it reaches scale. By all four of those measures, China is not building a Starlink competitor.
The hardest question the constellation raises is not whether the United States can launch faster. It is whether the United States is organized to build the kind of system Beijing began filing six years ago.