🔍 Read the full analysis: How Much Does A Satellite Haiku Tell Us About AI In Space? on ThorstenMeyerAI.com
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Google says a satellite carrying its TPU chip used Gemma 4 to generate a haiku while passing over Svalbard in late October 2026. The test shows that a data-center-class TPU can run in orbit, but the prototype’s limited power and unresolved costs, heat, radiation and networking challenges leave large-scale space computing unproven.
Google says a small satellite carrying one of its data-center-class TPU chips generated a haiku with its Gemma 4 AI model during a pass over Svalbard, Norway, on October 30. The test is the first reported result from Project Suncatcher, Google’s effort to study whether solar-powered satellites could eventually support AI computing in orbit; it demonstrates a limited on-orbit test, not a working space data center.
Google’s prototype launched on October 1 aboard SpaceX’s Transporter-18 rideshare mission, alongside more than 100 other spacecraft, according to the source report. It was built with satellite company Planet on a standard Planet platform. After roughly three weeks of commissioning, Google said engineers sent a prompt to the satellite during a ground-station pass and received the poem: “Blue marble below / Data streams in silent grace / Starlight feeds my core.” The company described the response as its “Hello World” moment.
The hardware is small-scale. TechCrunch reported that the satellite carries four TPUs and can generate up to about 1 kilowatt of power, but can operate only one chip at a time. The source report says the chip ran in bursts of about 20 minutes during the test. Google also said engineers tried other space-themed prompts and asked the payload to report its technical status; a Google spokesperson told TechCrunch that it was “operating nominally.”
Google calls the response the first-ever AI response from a data center class TPU in orbit. That wording matters: it does not mean this was the first AI chip or AI model used in space. A Starcloud satellite carrying an Nvidia H100 launched in November 2025, and CNBC later reported that the company had run and queried Google’s Gemma model in orbit. Google’s claim concerns its TPU, while Suncatcher’s broader proposal focuses on connecting multiple satellites to work together.
Infographic 01 · What flew and what it did
One satellite, four chips, one haiku
Google’s Project Suncatcher prototype is the first test of whether its AI data center chips can work in orbit. Here is what is on board and how it got there.
- 4 Nov 2025Project Suncatcher announcedResearch post and preprint; plan: two prototypes with Planet by early 2027
- 1 Oct 2026Prototype launchedSpaceX Transporter-18 rideshare, built with Planet; peer-reviewed paper in Joule
- Oct 2026About three weeks of commissioningHost and TPU trays powered up, early machine learning tests
- Week of 26 Oct“Hello World”First prompt to Gemma 4; the haiku came down during a pass over Svalbard, Norway
- NextHeat and radiation tests in orbitHow the TPUs behave under thermal stress and real space radiation
- 2027Twin interconnected satellitesMore power and laser links between satellites (Google, TechCrunch)
Scale check
The satellite vs. one power deal on Earth
890 MW = 890,000 kW, about 890,000 times the prototype (my arithmetic). The prototype bar is drawn at a visible minimum; to scale it would be invisible.
Not the first AI chip in orbit
Starcloud-1, launched November 2025 (CNBC)
- Carried one Nvidia H100, a data center GPU
- Trained a small language model in orbit
- Ran Google’s open model Gemma up there in December 2025
- Google’s claim is narrower: the first response from a data center class TPU in orbit
Sources: Google, 30 Oct 2026 · Google, 1 Oct 2026 · Google Research, 4 Nov 2025 · TechCrunch, 30 Oct 2026 · TechCrunch, 1 Oct 2026 · Google and Constellation, 6 Oct 2026 · CNBC, 10 Dec 2025.
Powered by Thorsten Meyer AI
A Haiku Is Not a Data Center
The test establishes a narrow but meaningful technical point: a Google TPU and Gemma 4 operated together in orbit well enough to receive a prompt and return an output. That is a step beyond a proposal on paper. But generating a short poem is not evidence that a satellite can support the power, sustained workloads, cooling, reliability or data movement expected of a large AI computing facility.
The motivation is energy. Google’s research argues that solar panels in some orbits could be up to eight times more productive than on Earth. That could become attractive as demand from terrestrial data centers grows: the source report cites US Department of Energy estimates that data centers used about 4.4% of US electricity in 2023, with a projected share of 6.7% to 12% by 2028. Those figures describe US data centers, not a forecast of how much power space computing would save.
Cost is a central test. The source report says Google’s paper estimates an orbital kilowatt would cost about $14,700 a year in launch costs at current prices, compared with about $810 if launch costs fall to $200 per kilogram. The lower figure is a scenario, not the prototype’s demonstrated operating cost. The paper estimates reaching that launch-price level would require roughly 1,800 Starship launches. For companies and energy planners, the haiku is evidence of a functioning experiment; whether orbital computing can compete economically remains an open question.
Products Worth Considering
Jackery Explorer 1000 v2 Portable Power Station
Jackery Explorer 300 Portable Power Station, 292Wh
Bose TV Speaker Compact Soundbar with Bluetooth and HDMI-ARC, Black
As an affiliate, we earn on qualifying purchases.
From Suncatcher Plan to Prototype
Google introduced Project Suncatcher in a research post and preprint on November 4, 2025. Its concept is a constellation of solar-powered satellites carrying TPUs, flying close together and using laser links to share data. Google’s initial stated milestone was a learning mission with Planet involving two prototype satellites by early 2027. The first prototype instead launched in October 2026, but the source material does not establish that the original two-satellite mission has already taken place.
Google’s peer-reviewed research appeared in the journal Joule on October 1, 2026, the launch date, according to the source. The paper sketches clusters of 81 satellites at about 650 kilometers altitude, with neighboring spacecraft roughly 100 to 200 meters apart. Google says a ground test of a single pair of optical transceivers reached 1.6 terabits per second. That was a bench test, not a demonstrated link between operational satellites in orbit.
The distinction from earlier projects is the proposed network design. Starcloud and other efforts have explored running AI workloads on individual spacecraft; Suncatcher’s concept is for interconnected satellites to act more like one computing cluster. Google says its next major milestone is a pair of linked satellites in 2027, which would test optical communications in orbit.
““first-ever AI response from a data center class TPU in orbit””
— Google, describing the first result
Costs, Heat and Links Remain Open
The demonstration does not settle whether orbital AI computing can be made reliable, affordable or scalable. The prototype’s power output is limited, and the source report identifies heat management in a vacuum and exposure to radiation as issues Google still plans to test. The satellite’s short operating bursts do not show how a larger system would handle sustained workloads.
Other unresolved questions include how to repair or replace failed hardware in orbit, how consistently satellites can communicate with ground stations, and whether tightly spaced satellite clusters can maintain high-speed optical links while operating. The source report also cites about 47,000 tracked objects in orbit; it gives no further details on how that figure was measured or what share presents a specific risk to Suncatcher.
Google’s cost estimates depend heavily on future launch prices and assumptions about a much larger system. No operating cost for the prototype, commercial timetable, or evidence of an economic advantage over terrestrial data centers is provided in the source material. The haiku confirms a prompt was returned; it does not answer those questions.
The Planned Two-Satellite Test
Google says it will continue tests of the current prototype, including measurements of TPU performance under thermal stress and radiation exposure. The next stated milestone is a pair of interconnected satellites in 2027. TechCrunch reports those spacecraft are expected to generate more power and test optical links between satellites, a key part of the proposed cluster design.
That mission would provide a more direct test of whether separate spacecraft can exchange data at the rates Google’s concept requires. Until those results are reported, and until costs and environmental limits are better established, Project Suncatcher remains an early research effort rather than a demonstrated alternative to data centers on Earth.
Key Questions
What did Google’s satellite do?
Google says the satellite received a prompt during a pass over Svalbard and returned a haiku generated by Gemma 4, running on a Google TPU.
Was this the first AI in space?
No. The source report describes a Starcloud satellite launched in November 2025 with an Nvidia H100 and says CNBC later reported that it ran and queried Gemma in orbit. Google’s narrower claim is about a response from a data-center-class TPU.
Can the satellite run a large AI data center?
No such capability has been demonstrated. The prototype produces up to about 1 kilowatt and can run one of its four TPUs at a time. Google’s proposed larger system would require many interconnected satellites.
What is the next milestone?
Google says it plans a pair of interconnected satellites in 2027 to test optical links, alongside continued checks of TPU performance under radiation and thermal stress.
Source: ThorstenMeyerAI.com
