California: Google has launched a prototype satellite carrying its custom artificial intelligence chips into orbit, beginning a major experiment to determine whether AI computing infrastructure could eventually operate in space.

The mission is part of Project Suncatcher, Google’s long-term research effort exploring the possibility of building solar-powered computing systems in low Earth orbit. The prototype was launched aboard SpaceX’s Transporter-18 rideshare mission, and Google has confirmed that the satellite reached orbit and is operating as expected.

Four AI Chips Aboard the Satellite

The refrigerator-sized prototype satellite carries four Google Tensor Processing Units, or TPUs. These specialised processors are designed for machine-learning workloads.

The initial mission is not a commercial space data centre. Instead, Google is using the satellite to collect real-world information about how its AI hardware performs under the difficult conditions of space.

Testing AI Hardware in Orbit

The experiment will examine how the TPUs respond to radiation, temperature extremes and the physical stresses associated with spaceflight.

Radiation is a particular concern because electronics operating outside Earth’s protective atmosphere can experience disruptions caused by energetic particles. Google has already conducted ground-based radiation testing, but the company says some information can only be obtained through actual operation in orbit.

Cooling Is a Major Challenge

Managing heat is one of the biggest technical obstacles for orbital AI computing.

On Earth, data centres commonly rely on airflow and other cooling systems to remove heat from processors. In the vacuum of space, there is no air to carry heat away.

Google is therefore testing specialised thermal systems involving heat pipes and radiators. The prototype’s AI processors will operate in limited periods while engineers collect information about thermal performance.

Solar Power Could Drive Future AI Systems

Project Suncatcher is based on the idea that satellites in orbit can receive sunlight for long periods and potentially generate more solar energy than comparable systems on Earth’s surface.

Google envisions a future network of satellites equipped with AI processors that could work together to perform increasingly demanding computing tasks.

The company has stressed, however, that substantial engineering challenges remain before such a system could become commercially practical.

Future Satellites Could Use Laser Links

Google is also researching high-speed communication between satellites.

Future versions of the project could use laser links to connect satellites and allow them to operate as a distributed computing network. The company plans additional tests involving two satellites in 2027 to study this technology.

A New Direction for AI Infrastructure

The experiment comes as demand for AI computing continues to increase and technology companies search for additional sources of computing capacity and energy.

Moving AI infrastructure into orbit could theoretically provide access to abundant solar energy while reducing some of the physical constraints faced by large terrestrial data centres. However, launch costs, radiation protection, thermal management, communications and maintenance remain significant challenges.

For now, Google’s Project Suncatcher is primarily a technology demonstration. The data collected from the first satellite will help determine whether the concept can progress toward larger orbital AI computing systems.