
Microsoft Unveils MAI-Image-1, Its First In-House AI Image Generator
November 5, 2025
editorial_staff
Google has revealed an ambitious project called Project Suncatcher, a research initiative exploring the possibility of operating AI data centers in space. The company aims to launch satellites equipped with its Tensor Processing Units (TPUs) that run on continuous solar power. This concept could help Google bypass the growing energy demands and environmental concerns linked to Earth-based data centers.
According to Travis Beals, Google’s Senior Director for Paradigms of Intelligence, space could offer the ideal environment for scaling AI computing. The company’s blog post and accompanying preprint paper outline the vision and the technical challenges of building a satellite-based computing network. In orbit, solar panels can generate power almost nonstop, potentially making them up to eight times more efficient than those on the ground.
One major obstacle is communication. To match the performance of terrestrial data centers, satellites must exchange data at extremely high speeds, tens of terabits per second. Achieving this would require satellite constellations flying in tight formations, possibly just a few kilometers apart, which poses additional risks due to space debris and collision hazards.
Another challenge is durability. Hardware in space faces intense radiation exposure, but Google says its latest Trillium TPUs have passed radiation tests, showing they can withstand a five-year mission without permanent damage.
Cost is also a key consideration. Currently, launching and maintaining satellites is expensive. However, Google’s cost projections suggest that by the mid-2030s, the price of running space-based data centers could be roughly on par with the energy costs of Earth-based facilities.
To move the concept closer to reality, Google has announced a partnership with Planet, a company known for Earth imaging satellites. Together, they plan to send prototype satellites into orbit by 2027 to test the hardware and study performance in real conditions. While still a long way from deployment, Project Suncatcher signals Google’s growing ambition to redefine the limits of AI infrastructure and explore new frontiers in sustainable computing.

Microsoft Unveils MAI-Image-1, Its First In-House AI Image Generator

Google Aims to Build Solar-Powered AI Data Centers in Space

Adobe previews AI “sneaks” that reshape video, lighting, and voice

Enterprise AI Vendor Tray.ai Launches Agent Gateway to Tame Agent Tool Development

ChatGPT to Add PayPal Payments Next Year