Two Ways to Build the Internet in Space -- China, Inc. vs Starlink et al

SpaceX has over 10,000 working Starlink Internet satellites in orbit today, and is far ahead of its Western and Chinese competitors, some of whom have not yet begun launching satellites. In my last post, I estimated the number of satellites each constellation would have in orbit at the end of the current Chinese five-year plan in 2030, and said one could think of the three Chinese constellations as divisions of a single enterprise, China, Inc. 

The Western approach of self-contained competitors has clear advantages. It allows rapid deployment, tight control over performance, and independence from other systems. The deployment of Starlink in Ukraine illustrates this point. On February 26, 2022, Ukraine requested Starlink service, and by the end of the day, service was activated, and terminals were en route. By March 19th, there were over 5,000 terminals in the country, and there were 15,000 by June 9th.

That’s the good news, but the bad news is that each provider must build and operate its own full-stack constellation, leading to duplication and a lack of global awareness. The Chinese approach to the Internet in space reflects their practice of making decisions within the context of five-year plans; it is slower but allows for a degree of optimization.

Rather than building multiple competing full-stack systems, Chinese constellations implement a division of labor: Guowang (backbone and sensitive government applications), Qianfan (general Internet service), and Honghu-3 (direct-to-device, Earth observation, IoT, etc.). For example, in providing global airline or maritime connectivity, Qianfan or Honghu-3 would both use Guowang for long-distance traffic. In this emerging architecture, the Internet stack is distributed across constellations. Instead of duplication, there is specialization and coordination.

The architectural difference becomes most visible in routing. In the Western stand-alone systems, routing is internal in each. In the Chinese layered systems, routing may span constellations, enabling system-level coordination and optimization. This does not imply fragmentation of the Internet, but it changes how traffic is managed and controlled.

It also alters our interpretation of my estimates of how many satellites each constellation will have in orbit in 2030. My estimate was that Starlink would have 16,083 satellites in orbit in 2030, far more than Western and Chinese runners-up Amazon LEO with 2,729 and Qianfan with 5,217. However, if we consider the Chinese constellations part of a unified, optimized collective with 11,692 satellites, the gap is smaller.

Note that, given today’s international political differences, the potential markets for Western and Chinese connectivity differ, as shown in this Venn diagram of the number and GDP of nations with Starlink connectivity, Chinese Belt and Road Initiative (BRI) projects, and both. The BRI projects tend to be in poorer nations, but there are more nations with BRI projects than with Starlink availability. The current success of Starlink suggests that the market is large enough to support it and the Chinese constellations. 

Finally, in a study of the Internet in India and China in the late 1990s, my colleagues and I observed that “China has been able to execute plans effectively by allocating resources to competing, government-owned enterprises.” That pattern is evident here. China’s earlier, now-discontinued, LEO broadband initiatives—Hongyun, sponsored by the state-owned China Aerospace Science and Industry Corporation, and Hongyan, sponsored by the state-owned China Aerospace Science and Technology Corporation—were initially conceived as competing systems. 

The creation of the China Satellite Network Group and the subsequent prioritization of the Guowang constellation following its initial announcement and ITU filing in 2020 marked a turning point. Experience gained from Hongyun and Hongyan must have informed this transition, with technology, personnel, and spectrum filings absorbed or redirected. Subsequently, Qianfan and Honghu-3 have emerged in more differentiated roles. The result is not simply the replacement of earlier projects, but a progression from competing proposals to a more coordinated, multi-layered architecture—one that reflects both accumulated experience and the broader organization of China’s space and telecommunications sectors.