Ever since advanced technologies like artificial intelligence (AI) and machine learning (ML) came to the forefront, people have questioned their environmental cost. These technologies consume substantial power for operation and require vast quantities of water to cool the systems on which they run. As these technologies grow more popular, more questions are being raised about their sustainability. But it appears Google has found an innovative way to address this problem.
Google has partnered with Planet, an aerospace company, to build and launch satellites for their Project Suncatcher. This project is aimed at harnessing solar energy in space to power machine learning systems, utilizing the cold background of space for cooling.
Giving more details about Project Suncatcher, Google CEO Sundar Pichai said:
"Inspired by our history of moonshots, from quantum computing to autonomous driving, Project Suncatcher is exploring how we could one day build scalable ML compute systems in space, harnessing more of the sun’s power (which emits more power than 100 trillion times humanity’s total electricity production)".
Project poses complex engineering challenges
Pichai acknowledged that the project is not as simple as it sounds:
"Like any moonshot, it’s going to require us to solve a lot of complex engineering challenges. Early research shows our Trillium-generation TPUs (our Tensor Processing Units, purpose-built for AI) survived without damage when tested in a particle accelerator to simulate low-Earth orbit levels of radiation. However, significant challenges still remain, like thermal management and on-orbit system reliability".
The company plans to launch two prototype satellites by early 2027.
Planet Co-Founder and CEO Will Marshall described the new initiative as an experiment to test Google's TPUs in space, aiming to leverage the sun's constant power and the extreme cold for cooling. He stressed that the ultimate objective is to relocate power-intensive computing infrastructure from Earth to establish a future compute-in-space architecture.
