Google launches the first AI-powered satellite to test data centers in space.

Google is preparing to launch the first satellite equipped with artificial intelligence chips into orbit, scheduled for Thursday, October 1. The satellite, named MVP, will carry four Tensor Processing Units (TPUs), microchips developed by Google to accelerate machine learning computations. The goal is to test the resilience of this hardware to the extreme conditions of space, such as radiation, temperatures, and the physical stress of spaceflight.

This is the first test of Project Suncatcher, Google's initiative launched in 2025 that aims to bring AI data centers into space to harness solar energy. The satellite will operate for one year, providing the chips with only one kilowatt of energy from solar panels—enough for testing—before disintegrating upon re-entry into Earth's atmosphere. Google has not disclosed the total expenditure but estimates that costs will become equivalent to those on Earth by the mid-2030s.

The idea of placing data centers in space stems from the growing demand for computing power for artificial intelligence and the limitations of terrestrial infrastructure. Technology companies fear they may not have sufficient space or electrical power to continue building these enormous computing centers, while protests over their environmental impact have become daily occurrences. Low Earth orbit offers a crucial advantage: satellites can harness nearly constant sunlight, generating up to eight times more solar energy than on Earth.

Google is not the only company working on this frontier: Elon Musk, Jeff Bezos, and Sam Altman support the opportunity, but the technological limitations are significant. The main engineering challenges include resistance to space radiation (which can cause circuit errors), cooling the chips (in a vacuum, there is no air to dissipate heat), and the high costs of transporting equipment into orbit. Experts believe it will take at least a decade before the system becomes economically sustainable on a large scale, with much engineering work still to be done.