Artificial intelligence could be forcing the world’s biggest tech companies into an uncomfortable reality, Earth may not be enough anymore.
Every ChatGPT query, AI-generated image, recommendation engine and autonomous system runs inside massive data centres packed with high-performance chips.
These facilities already consume enormous amounts of electricity, water and land. Now, as AI adoption accelerates globally, the pressure on terrestrial infrastructure is becoming impossible to ignore.
That pressure could be driving one of the most ambitious ideas Silicon Valley has explored in decades: putting AI data centres in space.
According to a Reuters report, Google is in talks with SpaceX and other launch providers for missions linked to “Project Suncatcher”, an initiative focused on orbital data centres powered by solar energy. Google reportedly plans to launch a prototype by 2027 alongside satellite imaging company Planet Labs.
The proposal sounds like science fiction. But the forces pushing it forward are very real.
AI Energy Demand Explodes
The AI boom is rapidly becoming an energy story.
The International Energy Agency (IEA) said global electricity demand from data centres grew 17% in 2025 alone, far faster than overall electricity demand growth. AI-focused facilities expanded even faster due to the rise of large language models and generative AI systems.
In the United States, the Energy Information Administration expects total electricity consumption to hit record highs again in 2026 and 2027, partly because of AI and crypto-linked data centres. US power demand is projected to rise from 4,195 billion kilowatt-hours in 2025 to 4,379 billion kilowatt-hours by 2027.
The IEA also estimates that global data-centre electricity demand could exceed 1,700 terawatt-hours by 2035 in a high-growth AI scenario. That would account for roughly 4.4% of global electricity demand.
Researchers are already warning that concentrated AI infrastructure could destabilise regional power grids. A 2026 study published on arXiv projected electricity consumption by six major AI firms could nearly double from 118 TWh in 2024 to as much as 295 TWh by 2030.
This is the context behind orbital computing. The idea is not merely futuristic experimentation. It is increasingly being discussed as a possible infrastructure solution.
Why Space Looks Attractive
Orbital data centres promise something Earth struggles to provide consistently: uninterrupted solar energy.
Unlike terrestrial solar farms, satellites in orbit can receive sunlight almost continuously. Space also offers naturally cold conditions, potentially reducing the massive cooling requirements that conventional AI facilities face on Earth.
That matters because cooling has become a major environmental and operational challenge for AI infrastructure. Scientific American reported in 2025 that extreme heat and water scarcity are increasingly threatening data-centre operations worldwide. In several regions, competition over water resources is already intensifying.
Researchers estimate the water footprint of AI systems alone could reach between 312.5 billion and 764.6 billion litres in 2025. Their associated carbon footprint could range between 32.6 million and 79.7 million tonnes of CO2 emissions.
Space-based systems could theoretically reduce dependence on freshwater cooling and relieve pressure on crowded electricity grids. That is one reason multiple companies are quietly investigating the concept.
The World Economic Forum says, “space offers an environment naturally suited for high-efficiency, sustainable computing, providing solutions to the very challenges that limit terrestrial development.”
Meanwhile, SpaceX has reportedly filed plans involving as many as one million solar-powered data-centre satellites.
Technical Risks Remain Huge
Despite the excitement, orbital AI infrastructure remains highly experimental. Launching servers into orbit is expensive. Maintaining them is even harder.
A recent academic paper analysing orbital data-centre economics found current launch costs remain far above commercially viable levels for large-scale general computing. The study concluded that space-based computing may initially work only for niche use cases such as edge processing or communications-integrated computing.
There are also engineering risks.
Space hardware must survive radiation exposure, micrometeorites and orbital debris. Even tiny collisions can destroy satellites. Communication delays and bandwidth limitations between Earth and orbit also remain major hurdles for real-time AI workloads.
SpaceX itself has acknowledged these uncertainties. According to TechRadar, the company warned investors in its IPO filing that orbital AI infrastructure may never become commercially viable because of technological and operational complexity.
Environmental concerns add another layer of complexity.
While orbital systems could reduce land and water usage on Earth, rocket launches themselves generate emissions and atmospheric pollution. Experts also worry about increasing congestion in low Earth orbit and the long-term consequences of space debris.
Scientific American noted that space-based computing may simply shift environmental costs rather than eliminate them.
New AI Infrastructure Race
Even so, the race has already begun. The orbital computing push reflects a broader transformation happening inside the AI industry. Infrastructure is becoming as important as algorithms.
Last week, AI startup Anthropic agreed to use the full computing capacity of SpaceX’s “Colossus 1” facility in Memphis. Reuters reported the project involves more than 220,000 Nvidia GPUs and over 300 megawatts of computing capacity.
The scale of those numbers explains why technology companies are searching for radically new infrastructure models.
For years, AI competition revolved around software breakthroughs. Increasingly, it is becoming a contest over electricity, cooling, chips and physical compute capacity.
Orbital data centres may never fully replace terrestrial facilities. But their emergence signals something important: the AI industry is beginning to collide with the physical limits of Earth-based infrastructure. That collision is now shaping the next frontier of computing, one that may extend far beyond the planet itself.
Also Read: Elon Musk’s SpaceX Partners With AI Startup Cursor, Eyes $60 Billion Acquisition Option
