In 2026, Chinese researchers successfully demonstrated AI inference directly on satellites, enabling ground-based robots to receive processed commands from orbit instead of cloud or data centers. This marks a structural shift toward decentralized, space-based computing infrastructure.

What happened: AI calculations were moved into orbit

A joint project between GuoXing Aerospace Technology and Shanghai Jiao Tong University demonstrated a fundamentally new architecture: voice commands sent from Earth were processed not in the cloud, but directly on satellites.
The calculations used the Qwen3 model , which parsed the query and generated a sequence of actions. The instructions were then returned to Earth, where the OpenClaw AI agent translated them into actual robotic movements.
The key point: the satellite acted not as a communication channel, but as a full-fledged computing node.

The classic model of autonomous systems is built around cloud computing. Robots, drones, and unmanned aerial vehicles depend on a stable internet connection and data centers.
Under normal conditions—in cities in the US, EU, or Asia—this isn't a problem. But in disaster zones, remote regions, or with damaged infrastructure, communications becomes a bottleneck.
Moving computing into orbit solves this problem. Satellites become an external computing circuit, accessible from virtually anywhere on the planet.
This means that autonomous systems can remain functional even in the complete absence of a ground network.

How does orbital AI architecture work?

The fundamental difference of the new scheme is that inference is performed directly on the satellite.

In the traditional model, the signal travels via a satellite and returns to Earth for processing. Here,
the request is sent into orbit;
an AI model analyzes it on the satellite;
a ready command is generated;
and the result is returned to the device.

This reduces dependence on data centers and redistributes the computing load.
It's especially important that the model isn't simply stored onboard, but actually processes tokens—that is, it operates in full mode.
This architecture opens up specific use cases. In the face of disrupted infrastructure, drones can receive navigation commands, robots can perform rescue missions, and autonomous vehicles can continue to navigate without network access.
In effect, a new category of infrastructure is being formed: computing as a service accessible from space.

This is especially relevant for:
remote regions without communication;
disaster zones;
military and strategic operations;
long-distance logistics.

GuoXing Aerospace Technology has already launched 12 satellites with the integrated Qwen3 model into orbit.
Development plans demonstrate the scale of ambition. By 2030, it is planned to deploy approximately 1,000 satellites, and by 2035, up to 2,800 devices.

The architecture is divided into two types:
the main part of the satellites will perform inference;
a separate segment will be responsible for training models.

This means moving from an experiment to a full-fledged computing ecosystem.
Technological barrier: the problem of heat in space
The main limitation is not computing power, but cooling.
On Earth, data centers use complex heat-dissipation systems. In space, such capabilities are unavailable: the vacuum prevents convection, and heat can only escape through radiation.

AI computing, especially on accelerators, generates significantly more heat than standard satellite functions.
This makes thermal management a key engineering challenge. Without effective cooling, scaling orbital computing is impossible.

From a global markets perspective, the emergence of orbital computing is shaping a new technology sector.
For investors, this means:
growing interest in space technologies;
the development of edge computing beyond Earth;
and increased competition between the US and China in the field of AI infrastructure.
In the long term, such decisions could change not only telecommunications, but also the principles of building a digital economy.
Although technology is not directly related to the foreign exchange market, its influence is felt through macroeconomics.
The development of space computing strengthens countries' technological leadership, which influences investment flows and currency dynamics. In 2026, the technology sector will remain one of the drivers of capital flows between the US, China, and the EU.

This is indirectly reflected in the Forex market through:
changes in investment flows;
growth in technology indices;
and the increasing role of innovation in the economy.

What's Next: The Next Step After Satellite Internet

The current trend goes beyond the idea of ​​an "internet from space." It's about transferring data processing logic to orbit.
If satellite internet provided access to the network, the new model provides access to computing.
This is a fundamental change in the architecture of digital infrastructure.

Orbital AI computing represents a shift from satellites as a communications channel to satellites as a computing platform. An experiment by Chinese developers demonstrates that autonomous systems can receive intelligent support independent of ground-based infrastructure. In the coming years, such solutions could form the foundation of a new global digital architecture, where computing extends beyond Earth.

Written by Ethan Blake
Independent researcher, fintech consultant, and market analyst
March 24, 2026

Join us. Our Telegram: @forexturnkey
All to the point, no ads. A channel that doesn't tire you out, but pumps you up.