Space-based computing moves into focus as costs begin to fall

Space-based computing has become a new focus of global space infrastructure in 2026. Governments and technology companies are exploring ways to move data processing beyond Earth.

In recent months, large-scale satellite constellation plans have accelerated. On February 2, SpaceX applied to deploy up to one million satellites. The plan aims to build a network of orbital data centers. In January, China also submitted proposals involving about 200,000 satellites.

Space is no longer used only for communication and observation – it is increasingly being used to process data.

That shift was reflected at a conference held in Beijing on January 26. The China Academy of Information and Communications Technology and more than a dozen research institutions and companies released a joint initiative, promoting the development of a "computing satellite network."

Why space-based computing is gaining attention now

Interest in space-based computing is rising as communications and artificial intelligence become more tightly linked. At the same time, computing capacity is becoming a key constraint.

Under current models, raw data is sent back to Earth. This takes time and reduces the value of time-sensitive information. About 90 percent of data generated in space is not effectively processed, Li Chao, director of Zhejiang Lab's space-based computing center, said at the conference.

With data processed in space or through integrated space-ground systems, space-based computing looks forward to high-performance computing, data storage, and AI inference and training in space.

Energy is another driver. In low Earth orbit and sun-synchronous orbits, satellites can receive sunlight for most of the day. Day-night cycles are short or minimal, which allows a stable and continuous supply of solar power.

Additionally, the extreme cold of space could help reduce cooling demands for data centers. If radiative cooling technology improves, high-performance processors could shed heat without large cooling systems, while ground data centers rely heavily on power-hungry air and liquid cooling.

What steps has China taken in space-based computing?

China has begun to place space-based computing within its long-term space plans.

China Aerospace Science and Technology Corporation recently said it would promote space-based digital infrastructure during the 15th Five-Year Plan period. The plan includes building gigawatt-scale space computing facilities.

Zhejiang Lab is leading the "Three-Body Computing Constellation." The project aims to place artificial intelligence models directly in space, which would process orbital data on site.

The program currently has 39 satellites under development and is expected to reach 100 satellites by 2027. In May 2025, twelve computing satellites developed by Zhejiang Lab were launched into orbit.

Each satellite can perform up to 744 trillion operations per second. Laser links between satellites can reach speeds of 100 gigabits per second. When connected, the satellites provide about five quadrillion operations per second of computing powerand 30 terabytes of storagecapability.

The satellites are also part of the "Star Computing" program led by The GuoXing Aerospace Technology, a Chinese commercial satellite service firm. This program aims to build a network of 2,800 computing satellites by 2030.

In January, the company deployed a large language model, Qwen3, in orbit for the first time globally.

What may determine the entry points for space-based computing

Despite growing interest, major challenges remain. Thermal management is one of the biggest.

Space is cold, but satellites cannot release heat easily. In a vacuum, heat escapes only through radiation. This puts pressure on chip-level heat collection. It also affects internal heat transfer and the design of large radiators.

Thermal stability also depends on precise orbit and attitude control. Satellites must be positioned carefully to maintain stable temperatures. Other challenges include radiation-hardened chips, high-speed space-to-space and space-to-ground links, and reliable power systems.

Cost will be another deciding factor. Large-scale satellite manufacturing and reusable launch vehicles are critical. Both are needed to reduce deployment costs.

On one side, low-cost construction of space-based data centers could be within reach. Consulting firm Starcloud estimates that space-based computing could approach cost parity with ground data centers in 2030. Models from Deutsche Bank suggest a breakeven point around 2032.

On the other side, progress in launch technology continues. On February 7, China launched a reusable experimental spacecraft aboard a Long March-2F rocket. The mission aims to verify reusable spacecraft technologies and support the peaceful use of space.

Energy remains essential since space-based data centers rely on solar power. Space photovoltaics are therefore a key component and China has a fully integrated solar power supply chain.

In early February, a SpaceX team came to China to visit photovoltaic companies, including solar giant JinkoSolar.

The team also looked at companies developing perovskite-based solar technologies, including GCL and Jolywood. Perovskite materials are lighter and more flexible than conventional options and are seen as better suited to the extreme conditions of space.