The AI space race has officially started. Recently, Google CEO Sundar Pichai revealed that the company is working to launch its first experimental data centers into orbit by 2027, marking the start of a long-term initiative to scale machine learning beyond Earth. As reported by Business Insider, internally known as Project Sunbather, the project aims to leverage the immense power of the sun to fuel AI workloads. In an interview with Fox News Sunday, Pichai said, “Google will start by sending “tiny, tiny racks of machines” aboard satellites to test feasibility. “We are taking our first step in ’27,” Pichai said. “In a decade, it’ll be normal to build extraterrestrial data centers.”

Why Google is planning for the space

Google’s plan for the space comes at a time when scrutiny is mounting over the environmental toll of AI. Data centers consume a lot of electricity, water and rare minerals, while producing significant greenhouse gas emissions and electronic waste. Google hopes to reduce the strain on Earth’s ecosystem by moving some of its infrastructure off the planet.

By moving some of this infrastructure off-planet, Google hopes to reduce the strain on Earth’s ecosystems. “One of our moonshots is: How do we one day have data centers in space so that we can better harness the energy from the sun, which is one hundred trillion times more energy than we produce on Earth today,” Pichai explained.

Global concerns over AI’s footprint

The United Nations Environment Program (UNEP) has warned that AI’s environmental impact could be severe if left unchecked. Sally Radwan, UNEP’s chief digital officer, said in November: “We need to make sure the net effect of AI on the planet is positive before we deploy the technology at scale.”Concerns include:

Google’s next move

Pichai has also confirmed that Google hopes to have its custom TPU chips operating in orbit by 2027. The company sees this as the first step toward scaling AI infrastructure in space, with the potential to normalize extraterrestrial data centers within a decade.