Google to Build Orbital Data Centers for AI by 2027
Google has announced an ambitious plan to move part of its artificial-intelligence infrastructure off the planet. The company will launch the world’s first space-based data centers — an orbital network of satellites designed to power the next generation of AI systems.
From Earth to orbit: the next computing frontier
According to The Guardian, the project aims to deploy around 80 satellites equipped with solar panels and high-efficiency processors at an altitude of 650 kilometers. These orbiting servers will support large-scale AI model training and cloud computing without consuming terrestrial energy or producing heat emissions on Earth.
The first prototypes are scheduled for launch in 2027, marking the beginning of a long-term effort to create a distributed orbital cloud. Google says the initiative will help handle the rapidly growing energy demands of AI training while improving global latency and data availability.
Why the sky is cheaper than the ground
The economics of space computing are shifting fast. The cost of satellite launches has dropped by more than 70 % over the past decade, and reusable rocket technologies make orbital deployment increasingly viable.
Industry experts predict that by the mid-2030s, the construction cost of space-based data centers could rival terrestrial ones — while operational expenses would be up to 20 times lower thanks to natural cooling, unlimited solar energy, and radiation-assisted thermal management.
As NVIDIA CEO Jensen Huang recently noted, “In ten years, most large-scale data centers will be built in orbit.” NVIDIA itself plans to launch its first orbital AI facility, Starcloud-1, as early as next month, accelerating competition in this emerging industry.
Energy efficiency and sustainability
AI’s carbon footprint is becoming a pressing concern. Training a single frontier-scale model can consume as much electricity as an entire city. Google’s orbital project could serve as a turning point for sustainable computation: continuous solar exposure in low-Earth orbit offers a clean, nearly limitless power source.
By decoupling data-center expansion from terrestrial grids, Google aims to reduce both environmental impact and dependency on land-based infrastructure — a crucial factor as global energy systems strain under the growth of cloud and AI workloads.
The geopolitical and security dimension
Orbital computing introduces new strategic questions. Space-based data centers could bypass regional energy restrictions and regulatory frameworks, creating a “sovereignty gray zone” for data jurisdiction. Governments are already debating how to classify orbital compute clusters — as satellites, infrastructure, or both.
At the same time, secure inter-satellite communication and quantum encryption protocols are expected to become mandatory to protect sensitive AI workloads from potential interception or space-weather interference.
A race beyond gravity
Analysts describe the Google–NVIDIA rivalry as the beginning of an “AI space race.” Both corporations envision orbital compute nodes linked to terrestrial facilities via laser-based broadband. For Google, the advantage lies in scalability and energy independence; for NVIDIA, in hardware optimization and chip control.
By 2030, space-based infrastructure could form the backbone of global AI networks — extending beyond simple data storage toward autonomous orbital training systems.
Conclusion
The idea of a data center in space once sounded like science fiction. Now it is becoming a strategic necessity. With AI workloads doubling every six months, Google’s orbital initiative may mark the first step toward a new paradigm — where computing quite literally leaves the Earth to find more room to think.
Editorial Team — CoinBotLab