AI Computing Power Demand Soars, Tech Industry Turns to Space Base Construction
With the rapid spread of generative artificial intelligence, global computing power demand is skyrocketing. Facing challenges such as rising energy and cooling costs on Earth and limited land resources, tech giants are seeking new breakthroughs — constructing data centers in space is becoming the next focal point of industry imagination.
Several Silicon Valley leaders have recently discussed the possibility of "space computing power," suggesting that orbital platforms will provide almost unlimited energy and space for future AI development, becoming the strategic pinnacle of the next technological race.
Google Proposes "Project Suncatcher," TPUs May Enter Space as Early as 2027
Google CEO Sundar Pichai, in a recent podcast, spoke about the company's long-term technological explorations, stating they are studying the feasibility of deploying AI hardware in orbit. He mentioned that Google is advancing a research project called "Project Suncatcher," aiming to free machine learning systems from Earth's surface limitations, enabling scaled operations in the space environment.
Pichai stated: "When you realize the computing power AI will require in the coming decades far exceeds what current energy systems can support, sending equipment to space no longer seems implausible." He revealed that Google hopes to launch TPUs into orbit by 2027, laying the foundation for space-based computing power.
Industry observers believe that if Google can be the first to send dedicated AI chips into space, it will gain a first-mover advantage in the global computing power competition.
Musk Proposes Annual Deployment of "Hundreds of Gigawatts" of Computing Power
Another key promoter of the space computing power concept is SpaceX and Tesla founder Elon Musk. He recently suggested on social media that future large rockets could send "AI satellites" equipped with solar arrays into orbit, constructing hundreds of gigawatts of computing power systems annually—far exceeding the total power capacity of current global data centers.
Musk emphasized that solar energy in space can be captured almost losslessly and without day-night and weather constraints, offering the only long-term solution to AI's energy bottleneck.
Industry experts believe Musk's vision, while requiring significant technological breakthroughs, reflects the tech industry's concerns about paths for computing power expansion, as terrestrial resources may not be able to support the extreme demands of the AI era.
Tech Leaders Collectively Discussing Space Computing Power as a Trend
Besides Google and Musk, tech leaders such as Amazon founder Jeff Bezos, OpenAI CEO Sam Altman, and Salesforce CEO Marc Benioff have also expressed that data centers will enter orbit in the next 10 to 20 years.
They believe space provides not only sustainable energy but also avoids the limitations of ground-based cooling systems and land costs, offering an ideal operating environment for large AI models.
China Also Plans Space Computing Power Layout; 700-Kilometer Orbit May Be First to Implement
Recent public information shows that China is also planning to build a high-power orbital computing platform. The plans indicate future deployment of gigawatt-level data center systems in orbits between 700 and 800 kilometers, including space computing nodes, data relay systems, and ground control networks.
The first stage (2025–2027) will focus on breakthroughs in space energy, heat dissipation, and communication technologies, and conduct experiments on initial computing power constellation construction.
Space Computing Power Becomes New Frontier in Global Tech Competition
As AI energy consumption rapidly rises, ground expansion of computing power is increasingly limited by energy, cost, and environmental constraints, making space the new blue ocean for the next generation of technological revolutions. The synchronized layouts of Google, Musk, and China indicate that space computing power is no longer just an experimental concept but is accelerating toward reality.
