Google Is Taking AI Data Centers to Space: Project Suncatcher Gets Its First Orbital Test

Why Google Wants AI Data Centers in Space

One of Google's biggest arguments is energy availability.

Satellites in low Earth orbit can receive near-constant sunlight. Google says they could generate up to eight times more solar power than on Earth, potentially providing a continuous energy source for future AI computing infrastructure.

The concept could eventually involve constellations of satellites carrying multiple TPUs and working together on AI workloads. But Google says this remains a long-term research project rather than an immediate replacement for terrestrial data centers.

The First Test Has Three Major Challenges

The satellite will test how Google's TPUs respond to launch forces, radiation and extreme temperatures.

During launch, spacecraft can experience acceleration loads of up to 10 times Earth's gravity, while individual TPU components can face forces of 50–100 g. Google has already conducted ground-based vibration testing.

Radiation is another concern. Google says its Trillium TPUs survived testing at UC Davis that exposed them to radiation levels greater than what they would receive during a five-year space mission. The orbital test will provide real-world data.

Cooling presents a separate engineering problem. Conventional airflow cooling doesn't work in the vacuum of space, so Google is testing heat pipes and radiators to remove heat from the chips.

Projects like Google’s Project Suncatcher show how AI infrastructure is becoming a strategic technology decision, not just an engineering challenge. Develop the leadership and AI strategy skills needed to evaluate emerging technologies and lead AI-driven transformation with upGrad’s Chief Technology Officer & AI Leadership Programme. 

Google Plans Another Space Test in 2027

Google's longer-term vision requires satellites to communicate at extremely high bandwidth.

The company plans to put two satellites into orbit in 2027 to test laser-based links between spacecraft. Future designs could use clusters of satellites, each carrying dozens of TPUs, to handle larger AI workloads.

Google acknowledges that significant engineering challenges remain. Reuters also reports that experts consider commercial viability to be years away because of launch costs, engineering constraints and satellite-production bottlenecks.

For now, Project Suncatcher is about proving one basic question: can AI computing hardware reliably work in space?