{"id":2260,"date":"2026-09-25T13:11:00","date_gmt":"2026-09-25T04:11:00","guid":{"rendered":"https:\/\/onepress.co.kr\/index.php\/briefing\/2026-09-25-google-suncatcher-orbit-en\/"},"modified":"2026-09-25T13:11:00","modified_gmt":"2026-09-25T04:11:00","slug":"2026-09-25-google-suncatcher-orbit-en","status":"publish","type":"briefing","link":"https:\/\/onepress.co.kr\/index.php\/briefing\/2026-09-25-google-suncatcher-orbit-en\/","title":{"rendered":"Google is preparing its first AI test satellite with four TPUs"},"content":{"rendered":"

2026-09-25 13:11 KST<\/strong><\/p>\n

Google said on 24 September that a compact satellite built with Planet will carry four AI accelerator chips into low Earth orbit next week.<\/p>\n

At roughly 1 kW and about one conventional server\u2019s compute, the payload is a test article, not the start of large-scale orbital AI service.<\/p>\n

What will be tested<\/h2>\n

The roughly ten-minute ride to orbit can expose the spacecraft to about 10g and components to 50\u2013100g. In orbit, radiation can flip bits and vacuum removes airflow, forcing heat pipes and radiators to reject heat. The mission will collect data across those conditions.<\/p>\n

What ground tests showed<\/h2>\n

The team shook the satellite on three axes and ran AI workloads while exposing Trillium TPUs to a proton beam at UC Davis. Google says the chips survived more total ionizing dose than expected over five years, but that is a company test, not proof of long-term orbital operation.<\/p>\n

Why compute in space<\/h2>\n

Google estimates near-continuous sunlight in low orbit could provide up to eight times the solar power available on Earth. Power alone does not make a data centre: launch cost, radiator area, radiation faults, unserviceable hardware and communications still matter.<\/p>\n

\"Google
AI-generated illustration of the Project Suncatcher test concept; it is not a photograph of the actual satellite.<\/figcaption><\/figure>\n

How the roadmap changed<\/h2>\n

In 2025 Google described two Planet-built prototypes by early 2027. It now separates an earlier single-satellite TPU test from a 2027 two-satellite laser-link test. They are different milestones.<\/p>\n

Is the launch fixed<\/h2>\n

EFE and independent launch schedules list 1 October from Vandenberg on Transporter-18. Rocket dates can move with weather and technical reviews; as of 25 September, launch and orbital success had not occurred.<\/p>\n

What to watch next<\/h2>\n

After launch, watch for separation, communications, error rates, power, radiator temperature and workload results. Even success leaves power, networking, debris and economics to solve before any large orbital data centre.<\/p>\n

Primary sources and independent checks<\/h2>\n

Official release: Google official update<\/a><\/p>\n

Earlier official research plan: Google 2025 research plan<\/a><\/p>\n

Independent report: EFE independent report<\/a><\/p>\n

Read in other languages<\/h2>\n

\ud55c\uad6d\uc5b4<\/a> \u00b7 English<\/a> \u00b7 Espa\u00f1ol<\/a> \u00b7 \u65e5\u672c\u8a9e<\/a> \u00b7 Ti\u1ebfng Vi\u1ec7t<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"

Google plans to launch the first Project Suncatcher test satellite, carrying four TPUs, on SpaceX Transporter-18 on 1 October. This is a precursor test of radiation, vibration and vacuum cooling\u2014not an operating space data centre.<\/p>\n","protected":false},"featured_media":0,"template":"","meta":[],"class_list":["post-2260","briefing","type-briefing","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/onepress.co.kr\/index.php\/wp-json\/wp\/v2\/briefing\/2260","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/onepress.co.kr\/index.php\/wp-json\/wp\/v2\/briefing"}],"about":[{"href":"https:\/\/onepress.co.kr\/index.php\/wp-json\/wp\/v2\/types\/briefing"}],"wp:attachment":[{"href":"https:\/\/onepress.co.kr\/index.php\/wp-json\/wp\/v2\/media?parent=2260"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}