🔍 Read the full analysis: Google’s Space AI Haiku: What It Suggests And What It Leaves Unanswered on ThorstenMeyerAI.com
Get hardware and tech essentials 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 generated a response from Gemma 4 after launching on Oct. 1, 2026. The test shows that a small AI workload can run in orbit, but does not establish that space-based data centers are practical or cost-effective. Google says further tests will examine radiation and heat, with two linked satellites planned as a later milestone.
The satellite was built with satellite company Planet and launched on a SpaceX rideshare mission, according to Google and the source report. Following about three weeks of commissioning, engineers sent a prompt during a ground-station pass as the spacecraft went over Svalbard, Norway. Google said the response came back as a three-line poem: “Blue marble below / Data streams in silent grace / Starlight feeds my core.” The company described the exchange as its “Hello World” moment.
The spacecraft carries four TPUs, but its reported power supply of up to about 1 kilowatt means it can operate only one chip at a time. TechCrunch reported the system ran in bursts of about 20 minutes during the test period. Google said engineers also asked the payload for space-themed limericks and for a report on its technical status. A Google spokesperson told TechCrunch that the satellite was “operating nominally.”
Google calls the test the first AI response from a “data center class TPU in orbit.” That description is narrower than saying this was the first AI computer in space: the source report says startup Starcloud launched a satellite carrying an Nvidia H100 GPU in November 2025, and CNBC later reported it had run and queried Google’s Gemma model in orbit. Google’s claimed distinction is its TPU hardware and its longer-term plan to connect satellites into a computing cluster.
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
Testing Google’s Orbital Computing Plan
The haiku is a small demonstration with a larger purpose: Google is testing whether AI computing could eventually move beyond terrestrial data centers. The company’s proposal depends on satellites powered by solar panels and linked to one another, rather than on a single spacecraft running isolated jobs. If that model works, it could offer another way to supply computing capacity as demand for AI and data-center electricity grows.
But the current test does not establish that orbital computing can compete with facilities on Earth. The satellite’s roughly 1-kilowatt power limit and short operating bursts are far removed from the needs of a large AI data center. The demonstration confirms a narrow technical milestone—a model returned a response from orbit—not the performance, reliability, or economics of a production-scale system.
The cost estimates make the gap especially clear. Google’s research paper estimates that, at current launch prices, a kilowatt of capacity in orbit would cost about $14,700 per year in launch costs. The paper estimates about $810 per year if launch costs fall to $200 per kilogram. Those figures are estimates tied to launch assumptions, not a complete operating-cost comparison. Google’s paper says reaching the lower-cost scenario would require roughly 1,800 Starship launches.
Products Worth Considering
Logitech MX Mechanical Full-Size Wireless Mechanical Keyboard, Tactile Quiet Switches, Backlit Keys, Bluetooth, USB-C, Multi-Device, Graphit
Logitech MX Master 4 for Mac, Ergonomic Bluetooth Mouse with Haptic Feedback, Ultra-Fast Scrolling, USB-C Charging, for MacOS, iPadOS, Windo
LG UltraGear 27-Inch QHD OLED Gaming Monitor, 240Hz, 0.03ms, G-SYNC Compatible, FreeSync Premium Pro, HDMI 2.1
As an affiliate, we earn on qualifying purchases.
From Suncatcher Proposal to Orbit
Google introduced Project Suncatcher in a research post and preprint paper on Nov. 4, 2025. It proposed solar-powered satellites carrying TPUs, arranged in close formations and connected through laser links so they could work together. Google’s initial stated milestone was a learning mission with Planet involving two prototype satellites by early 2027.
The first prototype reached orbit earlier, on Oct. 1, 2026, aboard SpaceX’s Transporter-18 rideshare mission, according to the source material. Google said it had established contact and the spacecraft was operating as expected. The company also published a peer-reviewed version of its research in the journal Joule that day. The haiku test followed several weeks of commissioning and early machine-learning checks.
Google’s paper argues that solar panels in a suitable orbit could be up to eight times more productive than on Earth. The proposed system is aimed at the energy demands of computing: the source material cites U.S. data centers as using about 4.4% of U.S. electricity in 2023, while the Department of Energy projected a range of 6.7% to 12% by 2028. These figures describe the U.S. electricity context; they do not show that orbital facilities would be cheaper or ready to replace terrestrial data centers.
““As the satellite passed by Svalbard, Norway, the downlinked poem confirmed success.””
— Google, describing the test
The Technical and Cost Hurdles
The test leaves major engineering questions open. Google says its next work will examine how TPUs handle thermal stress and radiation exposure. The source material also identifies heat rejection in a vacuum, repair difficulties, communications with ground stations, and collision risks from tracked objects in orbit as unresolved challenges. No repair plan or long-term reliability result is reported for this prototype.
It is also unclear how well multiple satellites could coordinate demanding workloads in practice. Google’s paper sketches clusters of 81 satellites flying within a radius of about one kilometer at an altitude of roughly 650 kilometers, with neighboring spacecraft separated by 100 to 200 meters. The paper reports a bench test reaching 1.6 terabits per second with a pair of optical transceivers, but that laboratory result is not the same as demonstrating a functioning, interconnected orbital cluster.
Nor has Google established that its projected launch-cost reductions will occur or that they would make the full system competitive. The available figures focus on launch costs per kilowatt and depend on a much lower launch price than today’s. The source material does not provide a complete estimate covering satellite construction, replacement, operations, communications, insurance, or end-of-life disposal.
Thermal Tests and Linked Satellites
Google says it will run further tests on the existing satellite to assess TPU performance under heat and radiation conditions in orbit. The next stated milestone is a pair of interconnected satellites in 2027. According to the source report, those spacecraft are expected to generate more power and test optical links between satellites, a step toward evaluating whether separate machines can work together.
The linked-satellite test will be more informative than the haiku about the central Suncatcher proposition: whether orbital hardware can sustain computing and transfer data between spacecraft. Google has not provided, in the source material, a detailed timeline beyond the 2027 milestone or a commitment to build a full orbital data center. Until those tests and cost assumptions are demonstrated, the project remains an early research effort rather than an alternative ready for commercial deployment.
Key Questions
What did Google’s satellite do?
Google says its Gemma 4 model generated a haiku after engineers sent a prompt to a satellite carrying a TPU chip. The exchange took place during a ground-station pass over Svalbard, Norway.
Was this the first AI system in orbit?
No. The source report says Starcloud launched a satellite with an Nvidia H100 GPU in November 2025, and CNBC later reported that it had run and queried Gemma in orbit. Google describes its result more narrowly as the first AI response from a data-center-class TPU in orbit.
Can this satellite power a data center?
No such capability has been demonstrated. The prototype produces up to about 1 kilowatt and can operate one of its four TPUs at a time, in short bursts. Google’s proposed system would require many more satellites and further testing.
Why is Google testing AI computing in space?
Google’s research explores whether solar-powered satellites could provide computing capacity and whether linked spacecraft could operate as a cluster. The company cites the productivity of solar panels in suitable orbits and rising electricity demand from data centers as part of the rationale.
What happens next for Project Suncatcher?
Google says it plans additional tests of the current satellite under thermal stress and radiation exposure, followed by a planned two-satellite mission in 2027 to test inter-satellite links. The results and the cost of scaling the system remain unknown.
Source: ThorstenMeyerAI.com
