🔍 Read the full analysis: What Happens When Google’s Satellite Writes A Haiku? on ThorstenMeyerAI.com
Get business pricing on monitors, keyboards and dev gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Google says a satellite carrying four of its TPU chips used the Gemma 4 model to write and send back a haiku during a pass over Svalbard. The test marks an early step in Project Suncatcher, but the prototype can run only one chip at a time in short bursts; the cost and engineering challenges of scaling AI computing in orbit remain unresolved.
Google says an orbiting prototype satellite has run its Gemma 4 AI model on a Google TPU and returned a haiku to Earth, marking the first reported result from Project Suncatcher, the company’s research effort into solar-powered AI computing in space. The demonstration, made during a pass over Svalbard, Norway, shows that the system can run a small AI task in orbit, not that space-based data centers are ready to compete with facilities on Earth.
Google said engineers sent the satellite a prompt asking Gemma 4 to write a haiku about being in space. The satellite generated the three-line response and downlinked it during a ground-station pass. Google described the exchange as the mission’s “Hello World” moment. The poem read: “Blue marble below / Data streams in silent grace / Starlight feeds my core.”
The prototype, built with satellite company Planet, launched on October 1 aboard a SpaceX rideshare mission. After about three weeks of commissioning, Google said it had powered up the host computer and TPU trays and run early machine-learning tests. TechCrunch reported that the satellite carries four TPUs and produces up to about one kilowatt of power. Power and thermal limits mean it can run only one chip at a time, in bursts of roughly 20 minutes, according to the source material.
Google’s October 30 announcement calls this the first AI response from a data-center-class TPU in orbit. That description is narrower than saying AI has never run in space: Starcloud launched a satellite carrying an Nvidia H100 GPU in November 2025 and later ran and queried Google’s Gemma model there, CNBC reported. Google says the next phase will test how its TPUs handle radiation and thermal stress in orbit.
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
What the Haiku Test Proves
The result is a technology demonstration, not evidence that orbital computing is already a practical substitute for data centers on Earth. It establishes that a satellite can receive a prompt, run an AI model on a TPU and return an answer, while operating within tight power and heat constraints. That is a meaningful early test for a project whose longer-term goal is to connect multiple computing satellites into a system.
The proposal matters because AI computing is increasing demand for electricity and data-center capacity. The source material cites U.S. Department of Energy estimates that data centers used about 4.4% of U.S. electricity in 2023, with their share projected to reach 6.7% to 12% by 2028. Google argues that solar panels in a suitable orbit could produce substantially more energy than on Earth. But producing power in orbit does not by itself solve the challenges of building, cooling, connecting and maintaining computing hardware there.
Cost is a major test. In Google’s research, the estimated annual launch cost for one kilowatt of capacity in orbit is about $14,700 at current launch prices. The paper estimates that this could fall to about $810 per kilowatt-year if launch costs reach $200 per kilogram. That lower figure is an assumption about future launch prices, not the cost demonstrated by this prototype. The project’s commercial prospects depend on whether launch prices fall and whether the rest of the system can work at much larger scale.
Products Worth Considering
Jackery HomePower 3600 Plus Portable Power Station, 3584Wh, 3600W
DJI Mic 3 Wireless Microphone Kit (2 Transmitters, 1 Receiver, Charging Case)
Anker SOLIX C1000 Gen 2 Portable Power Station
As an affiliate, we earn on qualifying purchases.
From Research Plan to Orbit
Google introduced Project Suncatcher in a research post and preprint on November 4, 2025. It proposed clusters of solar-powered satellites carrying TPUs and linked by lasers, so several spacecraft could work together on computing tasks. Google’s initial milestone was a learning mission with Planet involving two prototype satellites by early 2027; the first single-satellite test reached orbit earlier, on October 1, 2026.
The company’s longer-range research sketches a cluster of 81 satellites flying within a radius of about one kilometer at an altitude of roughly 650 kilometers. Satellites would be about 100 to 200 meters apart, close enough for high-speed optical links. Google’s paper reports a bench test reaching 1.6 terabits per second with a pair of optical transceivers; that is a test result for the link hardware, not a demonstration of a functioning 81-satellite computing cluster.
The planned design distinguishes Suncatcher from earlier individual-satellite AI experiments. Google’s stated aim is to connect spacecraft and distribute work across them. Whether those links can support useful computing operations in orbit has not yet been established by the haiku test.
““the first-ever AI response from a data center class TPU in orbit.””
The Limits Still Untested
The haiku demonstration does not show how the hardware will perform over a long period, under sustained workloads, or as part of a multi-satellite system. Google has not yet reported the results of its planned radiation and thermal-stress tests, and the prototype’s short operating bursts reflect limits in power and heat management.
Other open questions include how reliably satellites can exchange large volumes of data, communicate with ground stations and be repaired if hardware fails. The source material also cites about 47,000 tracked objects in orbit, a reminder that crowded orbital conditions are part of the operating environment. The available account does not establish a deployment schedule for a commercial service, its full costs, or how it would compare with terrestrial computing across all operating expenses.
Google’s estimates for lower launch costs and future power economics are projections in its research, not outcomes of this flight. Nor does the reported optical-link bench test show that a large orbital cluster has been assembled or run as one computer.
Tests Before a Satellite Cluster
Google says engineers will continue testing the prototype, including its response to radiation and thermal conditions in orbit. The next stated milestone is a pair of interconnected satellites in 2027. According to the source material, that mission is expected to generate more power and test optical links between spacecraft.
Those results will help show whether Google can move beyond a single satellite running brief tasks toward the multi-satellite system described in Project Suncatcher. The timing and scope of later missions, and whether the cost assumptions can be met, remain open.
Key Questions
What did Google’s satellite do?
Google says the satellite ran its Gemma 4 model on an onboard TPU after receiving a prompt, then sent the generated haiku to a ground station during a pass over Svalbard.
Was this the first AI system in orbit?
No. The source material says Starcloud launched a satellite carrying an Nvidia H100 GPU in November 2025 and later ran and queried Gemma in orbit. Google’s claim concerns an AI response from a data-center-class TPU.
Can the prototype run a full data center workload?
The reported demonstration does not establish that. The satellite produces up to about one kilowatt and can run only one of its four TPUs at a time, in short bursts, according to the source material.
What is Project Suncatcher meant to do?
Google is researching whether solar-powered satellites linked by lasers could work together as an orbital AI computing system. The haiku test used one prototype satellite and did not demonstrate such a cluster.
What happens next?
Google says it will test the satellite’s performance under radiation and thermal stress. Its next stated milestone is a pair of interconnected satellites in 2027 to test optical links and generate more power.
Source: ThorstenMeyerAI.com
