On May 14, China launched into orbit the first batch of satellites for its space computing constellation aboard a Long March 2D rocket from the Jiuquan Satellite Launch Centre.
Unlike traditional sensing or communication satellites, those 12 satellites are essentially super computers designed for space-based data processing and AI applications. They form part of the Three-Body Computing Constellation that will compose of 2,800 such satellites/supercomputers upon its completion by 2028.
The project is developed by Zhejiang Lab, a joint venture between Zhejiang Government, Zhejiang University, and Alibaba, with the mission of carrying out on-orbit data processing and cross-orbit laser communication in space.
The Three-Body Constellation is the world’s first such AI computing infrastructure. The 2,800 satellites in the constellation will serve as vehicles for computational expansion, creating a AI cloud computing network in space.
First unveiled in November last year at the World Internet Conference in Zhejiang, the Three-Body project aims to establish infrastructure in space to enhance computing efficiency compared to Earth-based data processing. The goal is to achieve a total computing power of 1,000 petaflops – equivalent to a quintillion calculations per second or the combined computing power of 200 million high-end mobile phones.
These satellites feature advanced AI capabilities, up to 100 Gbps laser inter-satellite links and remote sensing payloads. Once fully deployed, the Three-Body constellation will rival the most powerful terrestrial supercomputer data centers.
China is building such a massive computing constellation in space because space-based computing solves many existing issues with traditional terrestrial computing.
First, one major bottleneck for ground-based AI infrastructure is the massive energy required to power data centers. Data centres around the globe could consume more than 1,000 terawatt hours of electricity annually by 2026 – roughly equivalent to the entire electricity use of Japan – according to estimates by the International Energy Agency. Data centers are quickly becoming a main source of carbon emission.
Demand for more energy is expected to grow at double digit rates in the foreseeable future. As a result, many technology companies such as Microsoft, Google, and Oracle, are planning to build their own dedicated nuclear power plants to meet the energy needs.
While China generates more than twice as much electricity as the US already, new energy supply is needed to meet the increasing demand for AI computing. China is building massive-scale solar farms, wind mills, nuclear power plants, and hydro dams.
Such energy buildout projects include game-changing new projects such as thorium-based nuclear plants (https://huabinoliver.substack.com/p/what-is-really-going-to-change-the) and the Yarlung Zangbo Dam in Tibet, which is planned to be three times the size of the Three Gorges Dam – the world’s current largest dam.
In comparison, the Three-Body AI computing constellation will be powered by solar energy, which exists in infinite quantity in outer space. China is also building a mile-long giant solar panel in space which will generate enough energy for many smaller countries. These supercomputers will be self-powered by the solar panels installed on them.
The second bottleneck for ground-based data centers is the issue of heat management. Data centers generate an enormous amount of heat which requires vast quantities of water for cooling, exasperating existing water shortages in places like California. Cooling these facilities also requires vast amounts of water. In 2022 alone, Google used 19.7 billion litres to cool its data centres.
On the other hand, the space-based Three-Body Constellation will simply radiate heat into open space, leaving no carbon footprint.
In addition, unlike ground-based data centers, the Three-Body Constellation doesn’t have space constraints. Unlimited number of such satellites/supercomputers can be placed in orbit.
Third, the Three-Body Constellation can collect and process data directly without having to transmit them back to earth for processing.
Existing sensing or earth observation satellites collect vast amount of data from space 24/7. However, the data must be sent back to ground for processing and decision making, for example the coordinates of enemy navy fleets.
The traditional method is constrained by limited ground station availability and bandwidth. As a result, over 90% of the data collected by existing satellites don’t make it back to earth, often with significant delays. The sheer data volume and bottlenecks with real-time transmission have severely reduced the value of space-based data collection, especially in critical time-sensitive military situations.
On-orbit computing solves this problem through localized data processing and AI-based decision making. These supercomputers only need to transmit the outcome of analysis back to ground stations, rather than raw data. Such real time processing capabilities eliminate data leakage, reduce cost, and improve decision-making speed and quality.
According to Zhejiang Lab, each of the 12 satellites can process up to 744 trillion operations per second. Connected by high-speed laser links with data transfer rates of up to 100 gigabits per second, the initial network offers a combined computing power of 5 POPS and 30 terabytes of on-board storage.
The satellites also carry a space-based AI model with 8 billion parameters, capable of processing raw satellite data directly in orbit. They will be used to test capabilities such as cross-orbit laser communication and astronomical observations.
Fourth, the Three-Body Constellation is also safer from enemy attacks. Terrestrial AI infrastructure is prone to enemy attack in times of conflict. Space-based assets are harder to target and destroy.
The 2,800 satellites network, when fully deployed, forms a computing web through inter-satellite communication links so the loss of individual satellites will degrade overall computing capability.
Lastly, much of human’s future technological development will come from space explorations. The Three-Body Constellation provides a handy computing platform to support such outbound explorations.
The Three-Body Constellation also compliments the Thousand Sail ULEO (ultra low earth orbit) Satellite Constellation, launched by China last year, to enable 6G communication.
The new space-based Three-Body Computing Constellation is another milestone as China expands its technological leadership in sustainable AI computing and space development.
Liu Cixin must be pleased with the name, being a computer engineer and an author who says he is motivated by a love for science more than love for literature.
I can't find any confirmation in a quick search it's named after the first book of his popular Remembrance of Earth's Past trilogy, but given the lack of apparent relation to the physics problem and his popularity as a writer in China, I can only assume it's named in his honour.
China: the country so advanced, even popular writers are STEM scientists. Supposedly when Obama briefly met with master Liu while visiting China (his endorsement is on American covers of The Dark Forest, itself illustrating a contrast between the US and China), he was shocked to hear his daughters are engineers. American intelligentsia don't have engineer children, after all. Or often children at all these days.
China burning the pants off the West and especially America in that they no longer have to bend the knee anymore due to gap in technological prowess, and Jew World Order knows its hegemony was achieved only through meddling (ie. subversive propaganda to foment consent/dissent/revolutions), and at the barrel of a gun of course. Played the Player like a yo-yo until the Player found out too late due to hubris and greed.