Google’s prototype orbital compute satellite successfully launched today aboard a SpaceX rocket from California, marking the first time the tech giant has deployed its advanced chips into space.
Constructed by Planet Labs, the satellite aims to demonstrate the functionality of a Google Tensor Processing Unit, a competitor to Nvidia’s GPUs, within the harsh environment of space. This mission involves supplying a kilowatt of continuous power, managing thermal cooling, and rigorously testing a series of models to ensure reliability.
“We’ve done testing on the ground, but you know, there’s no test that’s completely as good as the real thing,” stated Travis Beals, the Google executive overseeing Project Suncatcher, the initiative to develop large-scale compute clusters in orbit.
Once commissioned, the satellite will operate its TPU in fifteen-minute intervals to prevent straining the satellite's power and thermal management systems. While this satellite utilizes a standard Planet Labs platform, the companies are developing a next-year demo featuring two purpose-built satellites designed to collaborate via laser communications links.
Although this SpaceX flight carries more than one hundred payloads, Google’s initiative distinguishes itself through its long-term scope. Beals describes this as a “long-term moonshot” focused on building infrastructure for future AI workloads, envisioning a network of 81 satellites processing tasks in parallel.
“The bandwidth and the latency between TPUs really, really matters when you’re trying to run a multi-rack workload…we’re trying to look ahead to not just what workloads exist today, but where they will be in five years,” Beals explained. The company notes that rockets capable of scaling orbital data centers cost-effectively do not yet exist, necessitating this forward-looking approach.
On Thursday, Google published a peer-reviewed white paper on orbital data centers in the journal *Joule*, offering a rigorous analysis of space-based compute logistics. The document highlights Google's economic outlook on space access, specifically analyzing SpaceX’s cost-reduction trajectory.
As a major investor in SpaceX, Google relies on the company to transport its spacecraft. The authors of the white paper argue that SpaceX has achieved a 20% annual price-reduction learning curve, projecting launch prices near $200 per kilogram by 2035.
To achieve these costs, the analysis estimates that Starship must transport a total of 370,000 tons of payload, requiring approximately 1,800 launches over the next decade, assuming a 200-metric-ton payload per mission. This represents a significant logistical challenge, considering Starship has never flown more than five times in a single year, despite Musk’s predictions of hourly flight rates by 2029.
However, Google’s updated research suggests its chips can withstand the radiation environment of space. After realizing the chips actually provide more shielding than anticipated, the company redid tests in a particle accelerator. While this resulted in slightly higher error rates in the logic circuitry, Google remains confident the hardware can support large-scale inference workloads for the satellite's five-year lifespan.
“The error rate is very low if you’re thinking about typical inference operations, right? Like one in a million,” Beals noted. Conversely, he highlighted that such error rates would become problematic for massive training runs involving thousands of chips running for months.
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.