Google launches Project Suncatcher to test AI data centres in space

The project envisions a network of small satellites, each equipped with Google's custom-designed Tensor Processing Unit (TPU) chips

Google Chief Executive Officer Sundar Pichai has unveiled Project Suncatcher, an ambitious initiative to examine whether future artificial intelligence (AI) computing could be powered by solar energy harnessed directly from space.

 

Google’s vision for AI in space

 

As part of the project, Google plans to deploy specialised AI chips on solar-powered satellites in low-Earth orbit. The objective is to process AI workloads in space, reducing the environmental footprint of terrestrial data centres while benefiting from uninterrupted solar exposure.

 

Google described Project Suncatcher as a long-term research “moonshot” rather than a commercial product for now, but one that could lay the groundwork for the next era of computing.

 

 

“More testing and breakthroughs will be needed as we count down to launch two prototype satellites with @planet by early 2027, our next milestone of many. Excited for us to be a part of all the innovation happening in (this) space!” Pichai said in a post on X.

 

What is Project Suncatcher?

 

The initiative envisions a constellation of compact satellites, each carrying Google’s custom-built Tensor Processing Unit (TPU) chips—processors designed specifically for machine learning and AI applications.

 

How will the system work?

 

Each satellite will feature high-efficiency solar panels for power and will be connected through free-space optical communication beams, enabling them to relay data between satellites and back to Earth.

 

Operating beyond the planet’s atmosphere, Google said these TPUs will have access to abundant solar energy while avoiding the cooling, land and water requirements of Earth-based data centres.

 

Testing in 2027

 

In its initial phase, Google intends to launch two prototype satellites by early 2027 to test TPU operations in orbit. These trials will evaluate aspects such as energy generation, chip performance under extreme conditions, data transmission efficiency and long-term reliability.

 

A new frontier for AI and space technology

 

Details of the project were released alongside a preprint research paper titled 'Towards a future space-based, highly scalable AI infrastructure system design'. The announcement aligns with Amazon founder Jeff Bezos’ prediction that gigawatt-scale data centres could be established in space within the next 10 to 20 years.