Google launches satellite to test AI chips in space

Prototype carries four TPUs to test chip performance in orbit

Google sent a refrigerator-sized satellite into space carrying four Tensor Processing Units, the specialty chips the company designed for machine learning and already deploys in data centers on Earth. The prototype, part of a project Google calls Project Suncatcher, will run a version of the company’s Gemma AI model for 15 minutes at a time, using the open-weight model to answer simple queries.

In a blog post, Google said the mission will gauge how the TPUs handle “the physical stress of spaceflight and the radiation and thermal extremes of space.”

Travis Beals, senior director and lead of Project Suncatcher, called the satellite “a very minimal test” to make sure the chips can run in space. A SpaceX rocket carried it into orbit, and Planet, the aerospace and satellite imagery company, will get it up and running. Beals said Google will then fire up and test the TPUs, with the goal of keeping the mission operational for a year.

Beals framed the project as a way to tap the solar system’s dominant energy source. “The sun puts out almost all of the power in our solar system. All of the other power sources that humanity has tapped into are just a tiny fraction of a percent,” he said. “So in some sense, this project is about tapping into the best way to use solar power to run AI compute.”

The satellite will travel in a sun-synchronous orbit where its solar panels will almost never be in the shade, eliminating the need to carry heavy batteries or backup power on the spacecraft.

Use of AI has grown sharply over the last few years, fueling demand for data centers. At Google’s annual developer summit in May, CEO Sundar Pichai said demand for AI services exceeds supply and he expects capital expenditures this year to be $180 billion to $190 billion — more than six times as large as in 2022 — as the company builds computing infrastructure and develops chips.

At a time when opposition to the proliferation of power-hungry data centers on Earth is growing, Google is one of a handful of companies advancing plans to put them in orbit, where there is virtually unlimited free energy and no protesters. Last November, venture-backed startup Starcloud launched a spacecraft carrying an Nvidia H100 chip and demonstrated a version of Google’s Gemini AI from space. SpaceX has said it expects to start deploying “orbital AI compute satellites” as early as 2028.

Google, in partnership with Planet, envisions clusters of satellites eventually orbiting together to form space-based data centers. Each satellite would carry dozens of TPU chips and communicate with the others — and Earth — via lasers. Next year, the company plans to put two more satellites into orbit to test their connection.

Heat remains a central challenge. Computer chips run hot, and on Earth the heat can be dissipated with wind or water — an option that does not exist in the vacuum of space. Brandon Lucia, a professor of electrical and computer engineering at Carnegie Mellon University, said the idea sounds “very sci-fi,” but much of the technology has been proven at a research level.

“In a satellite in particular, using more power to do the computations means dumping more heat into the confined environment inside of the satellite,” Lucia said.

Google called cooling the chips “a crucial research challenge” in its blog post, outlining approaches including using a combination of pipes and radiators to wick heat away. Beals said the radiators are among the heaviest components on the current mission — a drawback because heavier objects are more expensive to launch — and his team is working on ways to cool chips more efficiently.

Lucia also pointed to major questions about upkeep. When a chip malfunctions in a terrestrial data center, a technician can walk in and fix it, which is harder when the facility is circling hundreds of miles above Earth. “Those all have their cost and complexity amplified by a factor of 10, maybe a factor of 100. And so there has to be a big payoff,” he said.

There are environmental questions as well. While space data centers would run on solar power, the rockets that launch them run on carbon-based fuels, and the satellites would eventually re-enter Earth’s atmosphere, potentially adding pollution as they burn up.

Elon Musk has said space is the “only way to scale” data centers to meet future demand for AI. But the economics remain uncertain: launches are costly, repairs are difficult, and communication by laser across hundreds of miles and through Earth’s atmosphere is slower. Constellations of many satellites would be needed to equal the computing capacity of a terrestrial data center.

“It doesn’t help a lot if this is technically possible, if it’s always going to be too expensive to be practical,” Beals said.

Beals said it is difficult to know when space-based data centers will become economically viable. “I don’t see this being something where it’s cheaper to do this in the next five years. I think it will take longer than that,” he said. He added that space data centers will not replace terrestrial ones any time soon, with some use cases better suited to Earth and others to space. “In the really-long run, perhaps the set of things you do in space gets larger and larger until that’s where most of the compute is happening. But I think for a long time you’ll have both coexisting.”

Beals called the whole project a “moonshot.” “It’s going to take a long time,” he said. “It’s not going to be a completely straightforward journey.”