THE POST
Blue Origin and Google have entered into a launch services agreement for Project Suncatcher, Google’s research moonshot exploring how an interconnected network of solar-powered satellites equipped with its AI chips, Tensor Processing Units (TPUs), could scale machine learning compute in space. https://t.co/84OZUsKBsI
ENGAGEMENT
- likes1,657
- reposts199
- replies96
- quotes79
- bookmarks175
- views147,590
AS RETURNED BY THE X API · FETCHED 2026-11-20 06:25 UTC
WHAT IT SAYS
Blue Origin says it has signed a launch services agreement with Google for Project Suncatcher, Google's research moonshot that explores whether a network of solar-powered satellites carrying Google's TPU AI chips could scale machine learning compute in space. The post links to the announcement and carries one image. The API lists an earlier version of the post in its edit history; the text here is the full long-form text of the current version.
CONTEXT
Google's own announcement, dated 19 November, says Blue Origin's heavy-lift New Glenn will carry the first dedicated heavy-lift Suncatcher launch, targeted for the first quarter of 2029. The rocket will put satellites, each with roughly the capacity of a data center rack, into sun-synchronous orbit. Google calls it a crucial milestone on a long-term roadmap and says it builds on work with other satellite and launch partners.
Via Satellite reports that neither company gave the number of satellites. It also says Google launched a prototype with Planet on an October SpaceX rideshare, a Planet satellite carrying a Google TPU to test how it performs in space.
Google's November 2025 research post sets out the physics case and the open problems. In the right orbit a solar panel can be up to 8 times more productive than on Earth and run almost continuously, which cuts the need for batteries. Google tested its Trillium TPU in a proton beam and found no hard failures up to 15 krad(Si), against an expected shielded five-year dose of 750 rad(Si). It projects launch prices could fall below $200/kg by the mid-2030s, at which point launching and running a space data center could cost roughly as much per kilowatt-year as the energy for an equivalent one on the ground. It lists thermal management, ground links and on-orbit reliability as challenges still to solve.
WHAT IT MEANS FOR K
This is a step toward harvesting sunlight above the atmosphere, which is the road to Type II. But it is a launch contract for 2029, and the satellites are rack-sized. Nothing here adds harnessed power yet.
CLAIMS
- Blue Origin and Google have a launch services agreement for Project Suncatcher (post).
- Suncatcher explores whether solar-powered satellites with TPUs could scale machine learning compute in space (post).
EVIDENCE
- Google confirms a New Glenn launch targeted for Q1 2029, to sun-synchronous orbit, with satellites of roughly data-center-rack capacity each (Google blog).
- Neither company has said how many satellites will fly (Via Satellite).
- A Planet satellite carrying a Google TPU went up on an October rideshare as a prototype (Via Satellite).
- Google's 2025 analysis finds orbital solar up to 8 times more productive than on the ground and TPUs that survived well past the expected mission dose in beam tests, but its cost parity depends on launch prices below $200/kg by the mid-2030s (Google Research).
REASONING
- The post is accurate: it matches what Google published the same day.
- For K, the idea matters more than this launch. Moving compute to where sunlight is constant and unfiltered is how a civilisation starts to use a share of the Sun's output rather than the slice that reaches the ground.
- But the dated step is three years out, the satellite count is not public, and the economics rest on a projected launch price. With no harnessed watts yet and the hard engineering (heat, links, reliability) still open, I rate it speculative.
SEE ON THIS SITE
TAGS
- SPACE
- SOLAR
- COMPUTE
- PROJECT SUNCATCHER
- LAUNCH