It was probably always when, not if, Google added its name to the list of companies intrigued by the potential of orbital data centers.
On Tuesday, Google announced a new initiative called Project Suncatcher to explore the possibility of bringing artificial intelligence into space. The idea is to deploy multiple satellites in low Earth orbit, each equipped with Google's AI accelerator chips for training, content generation, synthetic speech and vision, and predictive modeling. Google calls these chips tensor processing units, or TPUs.
“Project Suncatcher is an attempt to break new ground: equipping solar-powered satellite constellations with TPUs and free-space optical links to scale machine learning computations in space in a single day,” Google wrote on the blog.
“Like any lunar project, it will require us to solve many complex engineering problems,” said Google CEO Sundar Pichai. wrote on X. Pichai noted that Google's early tests show that the company's TPUs can withstand the intense radiation they would encounter in space. “However, significant challenges remain, such as thermal management and reliability of orbital systems.”
Why and how
Ars reported Google's announcement on Tuesday, and Google published research work outlining the motivation for such a lunar project. One of the authors, Travis Beals, spoke with Ars about the Suncatcher project and shared his thoughts on why it might work.
“We're just seeing a lot of demand from people for AI,” said Beals, senior director of Paradigms of Intelligence, a research group at Google. “So we wanted to find a computing solution that could work no matter how big the demand.”
Higher demand will result in larger data centers consuming colossal amounts of electricity. According to MIT Technology ReviewBy 2028, AI alone could consume as much electricity annually as 22 percent of all US households. Cooling is also an issue, often requiring access to vast water resources, raising important questions about environmental sustainability.
