AI server farms are hitting power constraints: the International Energy Agency projects global data-centre electricity use could roughly **double to ~945 TWh by 2030**, with AI a primary driver, creating local grid bottlenecks and siting delays.
This page describes AI data centres **in orbit**, powered by sunlight and linked with free-space optical networks to beam **data (not energy)** down to Earth.
The concept has moved from speculation to live programs: Google Research’s **Project Suncatcher** laid out a technical blueprint in November 2025, while NVIDIA-backed **Starcloud** is flight-testing a GPU payload to validate space-compute operations: - Space AI
# Why pair AI with “space solar” Above the atmosphere, sunlight is continuous (no clouds, no night), so orbital platforms get **steady DC** and shed waste heat directly to space via radiators.
This avoids land, water, and grid bottlenecks on Earth and aligns with megaconstellation launch economics. Suncatcher’s paper argues this may be the **most scalable** long-run path for ML energy, subject to comms and reliability breakthroughs:
- suncatcher_paper.pdf 
- If you want to satiate AI’s hunger for power, Google suggests going to space - arstechnica.com
# Relation to Space Solar This is **Space Solar for compute**: harvest sunlight in space and apply it **in situ**, rather than beaming energy down. Many enabling parts overlap with space-solar power (thin-film arrays, large deployables, power electronics), but the “last mile” is **photons of data**, not microwaves of power.
For Earth, the benefit is reduced grid strain; for astronomy and SETI, it reframes technosignature expectations (bright optical networks, not thermal beaming) - services.google.com
# Open challenges
Inter-satellite **Tb/s fabrics**, radiation-hard accelerator roadmaps, **launch cadence** for continual replacement, secure **laser downlinks** in adverse weather, and policy for **space traffic management** at data-centre scale. Press coverage underscores both ambition and uncertainty as teams move from concepts to hardware - theguardian.com - spacedaily.com
# Further reading
Project Suncatcher explainer and paper; ecosystem reporting and early-trial coverage - research.google - services.google.com - arstechnica.com - space.com - blogs.nvidia.com
# See - Space AI