🔍 Read the full analysis: Google’s Orbital Haiku: What The AI Experiment Gets Right—and Leaves Open 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 its Project Suncatcher prototype has generated an AI response in orbit using a Google TPU and the Gemma 4 model. The test shows that a small satellite can run an AI workload, but it does not establish that space-based data centers are practical: power, heat, radiation, repairs, connectivity and launch costs remain unresolved.
Google says a satellite carrying one of its TPU chips has generated an AI-written haiku while orbiting Earth, a small-scale test of the company’s Project Suncatcher plan to explore computing in space. The response came from Gemma 4, Google’s open AI model, during a ground-station pass over Svalbard; it demonstrates an orbital AI inference test, not a functioning space data center.
The prototype, built with satellite company Planet, reached orbit on October 1 aboard SpaceX’s Transporter-18 rideshare mission. Google said the satellite made contact and was operating as expected. After roughly three weeks of commissioning, engineers sent it a prompt asking for a poem about space. Google described the response as its “Hello World” moment and said the downlinked haiku confirmed the test had succeeded.
According to TechCrunch’s account of the hardware, the satellite carries four TPUs but can produce only about one kilowatt of power. The system can run one chip at a time, in short bursts; the report described runs of about 20 minutes during this week’s test. Google said engineers also tried space-themed limericks and asked the payload to report its technical status. A Google spokesperson told TechCrunch that the satellite was “operating nominally.”
The test is a limited demonstration, not evidence that orbital computing is ready to take on workloads at terrestrial data-center scale. Google’s next stated tests will examine how its TPUs perform under thermal stress and radiation exposure. The company has also outlined a later mission involving two linked satellites, which would test whether spacecraft can exchange data over optical links.
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 Test, Not a Space Data Center
The result matters because it moves Project Suncatcher from research plans to a working, if modest, orbital experiment. A chip generating a response in space shows that some AI computation can take place on a satellite and that engineers can send prompts to the payload and receive results. It does not show that a large model can be trained in orbit, or that a network of satellites can match the capacity and reliability of an Earth-based data center.
Google’s rationale is energy supply. The company argues that solar panels in a suitable orbit could be substantially more productive than panels on Earth. That possibility attracts interest as electricity demand from computing grows. But the prototype’s roughly one-kilowatt power ceiling is far removed from the scale of commercial data centers. The experiment’s near-term value is in testing hardware and operations, while its wider significance depends on whether the technology can scale at a price and reliability that make sense.
For readers, the distinction between a successful demonstration and a viable service is central. The haiku is an accessible proof that the system returned an answer; it says little by itself about sustained computing capacity, costs, or the environmental trade-offs of building and launching many satellites.
Popular right now
De'Longhi Rivelia Automatic Espresso Machine with Milk Frother
De'Longhi 15-Bar Pump Espresso Machine with Milk Frother & Temperature Control
Dyson V8 Cyclone Cordless Vacuum
As an affiliate, we earn on qualifying purchases.
From Research Paper to Orbit
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 and connected by high-speed laser links. In the company’s research, closely spaced satellites could work together more like a distributed computer than a collection of isolated spacecraft. Google later published a peer-reviewed version of the research in the journal Joule on October 1, 2026, the day the prototype launched.
The project’s schedule and scale have shifted from the earlier plan. Google had described a “learning mission” with Planet involving two prototype satellites by early 2027. Instead, one smaller prototype launched in October 2026. The company says a future milestone is a pair of interconnected satellites in 2027; TechCrunch reported that the pair is expected to produce more power and test optical links.
Google is not the first company to put an AI-capable chip in orbit. CNBC reported that startup Starcloud launched a satellite with an Nvidia H100 GPU in November 2025 and later ran and queried Google’s Gemma model in orbit. Google’s narrower claim is that its test produced an AI response using a data-center-class TPU. Its proposed distinction is the plan to link satellites into clusters for shared computing, rather than operate AI workloads only on individual spacecraft.
The economics remain speculative. Google’s paper estimates current launch costs at about $14,700 per kilowatt-year of orbital capacity, compared with about $810 if launch prices fall to $200 per kilogram. The paper’s lower-cost scenario would require a major decline in launch costs; the source estimates roughly 1,800 Starship launches to reach the scale under discussion. These are research estimates, not demonstrated costs for an operating orbital data center.
““the first-ever AI response from a data center class TPU in orbit.””
— Google, describing the orbital test
The Engineering and Cost Gaps
The test does not establish how well TPUs will operate over extended periods in orbit, where equipment faces radiation and temperature conditions unlike those in a terrestrial data center. Google has said those are subjects for future testing. It is also unclear how the company would address maintenance or component failures when a satellite cannot be repaired like equipment in a ground facility.
Other open questions include how reliably a large satellite cluster could maintain high-speed links, how often it could communicate with ground stations, and whether it could handle sustained workloads rather than brief runs. The source material cites about 47,000 tracked objects in orbit, but does not provide a date or explain what object categories are included in that figure; it should not be read as a complete count of collision risks.
Cost projections depend heavily on launch prices falling. Google’s estimates compare present launch costs with a much lower hypothetical price, and the prototype does not validate either the projected economics or the launch cadence required for a large constellation. No commercial deployment plan, service price, or operating timeline has been established in the material provided.
Tests Before the 2027 Pair
Google says it will continue evaluating the prototype, including tests of TPU performance under orbital heat and radiation conditions. The results have not yet been reported. The next planned milestone is a two-satellite mission in 2027, according to Google; TechCrunch says the spacecraft are intended to test optical links and carry more power than the current prototype.
Those tests would address a more demanding question than whether one satellite can return an AI response: whether separate spacecraft can exchange enough data to operate together on a shared workload. Until the company reports those results, Project Suncatcher remains an early research program. The haiku confirms a narrow technical milestone, while the commercial case for computing at scale in orbit remains open.
Key Questions
What did Google’s orbital satellite do?
Google says a prototype satellite used a Google TPU to run Gemma 4 and return a short poem after engineers sent it a prompt during a ground-station pass.
Was this the first AI system in orbit?
No. The source material says Starcloud launched a satellite with an Nvidia H100 in 2025 and later ran and queried Google’s Gemma model there. Google’s claim is specifically about an AI response from a data-center-class TPU in orbit.
Is Google operating a data center in space?
No. The test involved one small satellite, with about one kilowatt of available power and limited runs on one TPU at a time. It is an early experiment, not a commercial orbital data center.
What is Google testing next?
Google says it will examine TPU performance under thermal stress and radiation exposure. A two-satellite mission planned for 2027 is intended to test links between spacecraft, according to the company and TechCrunch’s report.
What remains uncertain about Project Suncatcher?
It is not yet clear whether linked satellites can support sustained workloads, how they would be maintained, or whether launch and operating costs could compete with computing on Earth. Google’s cost figures are research estimates, not proof of commercial viability.
Source: ThorstenMeyerAI.com
