Google’s “moonshot” desire to expand its AI footprint is becoming more literal.
CEO Sundar Pichai said in a statement fox information interview Google will soon begin building an artificial intelligence data center in space, it was revealed on Sunday. Tech giants announce suncatcher plan Earlier this month, the goal was to find more efficient ways to power the energy-hungry center, in this case using solar energy.
“One of our moonshots is, how do we one day build data centers in space so that we can better harness the energy from the sun, which is 100 trillion times more energy than we generate on Earth today?” Pichai said.
Google will take the first step in building an off-terrestrial data center in early 2027 by partnering with satellite imaging company Planet to launch two pilot satellites to test hardware in Earth orbit. Pichai said that space-based data centers will become the new standard in the near future.
“But there’s no doubt in my mind that in about ten years it will be seen as a more normal way of building data centers,” he said.
Data Center Space Race
To be sure, Google isn’t the only company looking to the sky for answers to improve data center efficiency. Earlier this month, Y Combinator and Nvidia-backed startup Starcloud Sending the first artificial intelligence-equipped satellite into space. CEO and co-founder Philip Johnston predicts alien data centers will spawn 10 times lower Even taking into account emissions from launches, the carbon emissions are higher than those on Earth.
While the cost of satellites for testing AI hardware in space has fallen sharp reductionWhile the development of off-terrestrial data centers is within reach, the cost of building these solar centers remains an unknown, especially since Earth-based data centers are expected to require more than $5 trillion in capital expenditures by 2030, according to an April report McKinseyReport.
Google returns to conversation as AI frontrunner Recently released Gemini 3is one of several major hyperscalers pouring money into data centers to expand their computing capabilities. Google itself announced this month Invest $40 billion Texas data center construction.
At the same time, speculation about an AI bubble could create a data center oversupply that could make the data center space race a dangerous overinvestment.
“The risks are high,” McKinsey’s report said. “Overinvestment in data center infrastructure can lead to stranded assets, while underinvestment means falling behind.”
Using solar energy to power data centers is becoming increasingly attractive as concerns grow over the sustainability of expanding artificial intelligence computing, which requires large amounts of electricity. December 2024 U.S. Department of Energy Report A survey of domestic data center usage found that data center loads have tripled in the past 10 years and may double or triple again by 2028. The report said these data centers consumed more than 4% of the country’s electricity in 2023 and are expected to consume 12% of U.S. electricity by 2028.
According to its latest data, electricity use in Google’s data centers alone has more than doubled in the past five years, from 30.8 million megawatt hours last year to 14.4 million megawatt hours in 2020, when Google began tracking data center energy consumption specifically. Sustainability Report Released in June.
Google has been working to reduce the energy required by its growing data centers, reporting that the company will reduce data center energy emissions by 12% by 2024 despite its growing data center footprint. However, concerns about the long-term sustainability of data center expansion remain.
“There is still a lot we don’t know about the impact of artificial intelligence on the environment, but some of the data we have is worrying,” Golestan Radwan, chief digital officer of the United Nations Environment Programme, said in a statement after the project launched last year. notes Warning about the environmental impact of AI infrastructure expansion. “We need to ensure that AI has a positive net impact on the planet before deploying this technology at scale.”