Google's Project Suncatcher satellite reached orbit on October 1, and contact is confirmed

Google's Project Suncatcher satellite reached orbit on October 1, and contact is confirmed

Google's prototype satellite for Project Suncatcher launched into orbit on October 1 aboard SpaceX's Transporter-18 rideshare mission, and Google says its team "has confirmed contact with the satellite and it is operating as expected". The satellite was built in partnership with Planet, according to the post Travis Beals published on October 1, and it carries Google's own TPUs.

Project Suncatcher wordmark and flower logo on a yellow, orange and blue gradient

Google's own title card for Project Suncatcher, the research programme the prototype satellite belongs to. Image: Google

Beals is senior director of Paradigms of Intelligence at Google. He calls the launch "the first step in a long-term research moonshot exploring whether space could one day host scalable machine learning infrastructure", which is a long way from running a model in orbit and is not claimed as one.

The experiment now is survival. "Over the coming weeks, we'll gather in-orbit data on how our TPUs handle the physical stress of spaceflight and the radiation and thermal extremes of space", Beals writes. Confirmed contact is a first-contact result, so nothing yet says how the chips behave once they have been up there a while.

The research behind the mission is now peer reviewed

"Our peer-reviewed paper is now available in Joule, detailing the research behind this mission", Beals writes. Before flight the team ran Trillium TPUs in a proton beam facility at UC Davis's Crocker Nuclear Laboratory while the chips executed AI workloads, watching for errors such as bitflips, and reported that they "can survive a radiation total ionizing dose greater than what they would receive during a five-year space mission".

We covered that five-year radiation dose and the rest of the pre-launch testing on October 1, before Google confirmed the satellite was up. "Some things can only be tested in space", Beals writes, which is the whole argument for flying one at all.

The reason for going up is power, not compute

Google announced Project Suncatcher last year, and it is a bet on sunlight. In low Earth orbit a satellite can sit in near-constant sun and generate up to eight times more solar power than on Earth, Google says, and the eventual idea is to link constellations of satellites so they can carry larger workloads between them. Google puts its next milestone for that in 2027.

A single prototype proves nothing about an orbital cluster, and the problems Google names are radiation, vibration on the way up and cooling in a vacuum, where there is no air to carry heat away.

What Google has not published

There is no TPU count for the satellite, no orbital parameters, no launch time and no description of what the hardware will be asked to compute. Google gives no mass, no power budget and no mission duration, and it has not said when the in-orbit data will be released or in what form.

Google names Planet as the partner the satellite was built with and says nothing about how the work was divided, so how much of the spacecraft around the TPUs is Planet's is not on the record. The paper in Joule covers the research behind the mission rather than the flight hardware.

Transporter-18 is a rideshare, so the Suncatcher prototype went up alongside other operators' payloads rather than on a flight of its own. For now the publishable result is the smallest possible one: the thing is in orbit, it answered, and Google says it is behaving.

Leave a Comment

Your email address will not be published. Required fields are marked *