🔍 Read the full analysis: Google’s First Data Center Satellite Tried Poetry. What Can AI Prove? on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Google says a satellite carrying its data-center-class TPU generated a poem using the Gemma 4 model during a pass over Svalbard. The test marks an early Project Suncatcher milestone, but the prototype has limited power and does not establish that orbital AI computing is commercially viable.
Google says a satellite carrying one of its data-center-class TPUs generated a poem with the Gemma 4 AI model and sent it to a ground station during a pass over Svalbard, Norway. The test is an early milestone for Project Suncatcher, Google’s effort to study whether solar-powered satellites could eventually support AI computing in orbit; it does not show that a space-based data center is practical at commercial scale.
The satellite launched on October 1, 2026, aboard SpaceX’s Transporter-18 rideshare mission, alongside more than 100 other spacecraft, according to the source report. Google said it confirmed contact and that the spacecraft was operating as expected. After several weeks of commissioning, engineers sent a prompt asking the model for a haiku about space. The resulting lines were: “Blue marble below / Data streams in silent grace / Starlight feeds my core.”
The spacecraft was built with satellite company Planet using a standard Planet platform. TechCrunch reported that it carries four TPUs but produces up to about 1 kilowatt of power. Power and heat constraints mean it can run only one chip at a time, in bursts. The source says earlier runs were around 15 minutes and the latest lasted roughly 20 minutes. Google described the poem as its “Hello World” moment and said engineers also tested space-themed limericks and asked the payload to report its technical status.
Google’s next stated tests will examine how TPUs perform under thermal stress and radiation exposure. The company has described a follow-on mission with two interconnected satellites in 2027, which would test optical links between spacecraft. The current satellite is a demonstration of a limited compute task, not a full data center or evidence that AI workloads can yet be shifted from Earth to 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 a One-Kilowatt Test Shows
The result matters because it moves Project Suncatcher from research plans to a working, if small, orbital test of Google’s own AI hardware. It shows that a TPU and an AI model can be operated in space and that a response can be returned to Earth. That is a concrete engineering milestone, but the poem itself says little about performance on demanding workloads, operating costs, or whether a network of satellites could reliably act as one computer.
The economic case remains unsettled. Google’s research estimates that, at current launch prices, a kilowatt of orbital capacity would cost about $14,700 per year in launch costs. In the paper’s scenario where launch costs fall to $200 per kilogram, that estimate drops to about $810 per kilowatt-year, close to the power cost cited for U.S. data centers. The lower figure depends on a major reduction in launch prices; it is a modeled estimate, not a price available today. The paper estimates that reaching its envisioned scale would require roughly 1,800 Starship launches.
For readers, the broader issue is the growing demand for electricity to run AI infrastructure. The source cites U.S. data centers using about 4.4% of national electricity in 2023 and a Department of Energy projection of 6.7% to 12% by 2028. Those figures help explain the attraction of solar power in orbit, but they do not establish that orbital systems will ease energy pressures on Earth. Launch costs, maintenance, heat management and communications must all be addressed first.
Popular right now
Dyson V8 Cyclone Cordless Vacuum
Shark Rocket Corded Stick Vacuum (HV301) – Gray/Orange
Ninja DualZone 6-in-1 Air Fryer, 8-Qt, Grey, DZ801
As an affiliate, we earn on qualifying purchases.
From Suncatcher Plans to Orbit
Google first set out Project Suncatcher in a research post and preprint paper on November 4, 2025. The concept is a constellation of satellites carrying TPUs, flying close together and connected by laser links so they could coordinate computing. Google’s initial stated milestone was a learning mission with Planet involving two prototype satellites by early 2027. The single-satellite test launched earlier than that target, but it does not test the planned multi-satellite system.
Google’s research argues that solar panels in a suitable orbit could be up to eight times more productive than on Earth. Its longer-term proposal sketches groups of 81 satellites at about 650 kilometers altitude, with neighboring spacecraft 100 to 200 meters apart. Google says a bench test of a pair of optical transceivers reached 1.6 terabits per second; that lab result is not the same as demonstrating those data rates across an operating orbital cluster.
Google’s claim is specifically about a response from a data-center-class TPU, not the first AI computation in orbit. The source reports that startup Starcloud launched a satellite carrying an Nvidia H100 GPU in November 2025 and later ran and queried Google’s Gemma model in orbit, according to CNBC. The distinction matters: Google is testing its own chip and a proposed connected-satellite architecture, while prior reported work involved AI computing on an individual satellite.
““the first-ever AI response from a data center class TPU in orbit.””
Limits Beyond the First Poem
The demonstration does not establish how the chip performs over long periods, how much useful AI work it can complete, or whether the satellite can withstand the combined effects of heat and radiation. Google has not reported results from the planned thermal and radiation tests. The current satellite’s roughly 1-kilowatt ceiling and short operating bursts also limit what it can demonstrate.
Other open questions concern the proposed system rather than this single test: whether satellites can maintain fast, reliable links while moving in orbit; how equipment could be repaired or replaced; how heat can be managed in a vacuum; and how operators would handle the hazards of orbital debris. The source cites about 47,000 tracked objects in orbit, but does not provide a date or define the tracked-object category for that count. The economics also depend heavily on future launch prices and on the cost of building, deploying and operating a constellation.
Google’s “first” wording should be read narrowly. The report describes earlier orbital AI work with an Nvidia GPU and the Gemma model. What remains to be demonstrated is whether Google’s approach—especially the linked cluster of satellites—can do sustained, useful computing at a cost that competes with Earth-based data centers.
Google’s Planned Orbital Tests
Google says it will continue testing the current satellite, with a focus on thermal stress and radiation exposure. It has also identified two interconnected satellites in 2027 as the next milestone, subject to the current work and mission plans. The follow-on is intended to test optical communications between spacecraft, a key part of the company’s proposal to connect satellites into a computing cluster.
Evidence from those tests should clarify whether the hardware can operate in orbit and exchange data as intended. It will not, by itself, resolve the economics of deploying a much larger constellation. Google has not provided a detailed commercial deployment schedule in the source material, and the scale-up depends on results from the hardware tests, launch costs and the ability to manage power, heat, communications and orbital risks.
Key Questions
What did Google’s satellite do?
It ran Gemma 4 on a Google TPU and generated a poem in response to an uplinked prompt, which Google says was sent back to a ground station during a pass over Svalbard.
Was this the first AI computation in orbit?
Not according to the source material. Starcloud previously launched a satellite carrying an Nvidia H100 and later ran and queried Google’s Gemma model in orbit, as reported by CNBC. Google’s claim concerns an AI response from a data-center-class TPU.
Can the satellite run a full data center?
No. The prototype has up to about 1 kilowatt of power and can operate only one of its four TPUs at a time in short bursts, according to the reported hardware details. It is a small test payload, not a full data center.
Why is Google testing AI computing in space?
Google’s proposal is based in part on the potential for solar panels to produce more energy in a suitable orbit than on Earth. The company is investigating whether that advantage could support AI computing, but launch costs and engineering challenges mean the commercial case remains unproven.
What happens after this test?
Google says it will test the TPU’s performance under heat and radiation exposure. Its next stated milestone is a planned pair of interconnected satellites in 2027, intended to test optical links between spacecraft.
Source: ThorstenMeyerAI.com
