🔍 Read the full analysis: A Satellite Haiku Isn’t The Whole Story About AI In Space 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
A Google- and Planet-built satellite running a Google TPU generated a haiku after a prompt was sent during a pass over Svalbard, Google said on October 30. The test shows that a data-center-class TPU can run briefly in orbit, but the prototype has limited power and does not establish that space-based AI computing is economically or technically ready to scale.
Google says a small satellite built with Planet has generated an AI-written haiku in orbit using a Google TPU, an accelerator designed for data-center workloads. The October 30 test is an early result for Project Suncatcher, Google’s effort to assess whether solar-powered satellites linked by lasers could eventually perform AI computing in space; it does not show that orbital data centers are ready for commercial use.
The prototype launched on October 1 aboard a SpaceX rideshare mission and spent about three weeks undergoing commissioning, according to Google’s account. During a ground-station pass over Svalbard, engineers sent a prompt asking Gemma 4, Google’s open AI model, to write a haiku about space. Google said the satellite returned the poem, then handled other tests, including space-themed limericks and a request to report its technical status.
TechCrunch reported that the Planet-based satellite carries four TPUs and can generate up to about 1 kilowatt of power. That limit means it can operate one chip at a time, in short bursts. Google said the spacecraft was operating as expected; the company described the prompt-and-response test as its “Hello World” moment. The result establishes a limited demonstration, not sustained processing capacity or a working network of orbital computers.
Google’s next stated tests will examine how the TPUs perform under thermal stress and radiation. If those tests proceed as planned, the next milestone is two interconnected satellites in 2027, according to Google. TechCrunch reported that the pair is intended to generate more power and test optical links between spacecraft.
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
From One Chip to Orbital Clusters
The test matters because it moves Project Suncatcher from a research proposal toward hardware trials, while leaving the hardest questions unanswered. Google’s long-term design depends on multiple satellites working together through high-capacity laser links. A single satellite returning a short poem during a communications pass does not test the performance, reliability or coordination needed for a large AI workload across a cluster.
Google’s case for moving computing off Earth centers on solar power. The company says solar panels in a suitable orbit could be up to eight times more productive than on Earth. That potential is relevant as electricity demand from data centers grows: the U.S. Department of Energy estimates data centers used about 4.4% of U.S. electricity in 2023 and projects a range of 6.7% to 12% by 2028. Those figures describe pressure on terrestrial power systems; they do not show that orbital computing is already a practical substitute.
For companies and the public, the immediate takeaway is a demonstration rather than a new service. The prototype’s limited power and brief operating windows are far from the scale of Earth-based data centers. Whether orbital systems could ease energy constraints depends on launch costs, satellite lifetime, cooling, communications and maintenance, among other unresolved factors.
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 32GX850A-B 32-Inch 4K UHD Glossy OLED Gaming Monitor
As an affiliate, we earn on qualifying purchases.
Suncatcher’s Earlier Research Plan
Google introduced Project Suncatcher in a research post and preprint on November 4, 2025. Its concept was a constellation of solar-powered satellites carrying TPUs and flying close enough to exchange data over laser links. The company initially described a learning mission with Planet involving two prototype satellites by early 2027. The first prototype reached orbit earlier, on October 1, 2026, and Google published a peer-reviewed version of its research in the journal Joule on that date.
Google’s effort is not the first reported case of an AI chip or model operating in orbit. CNBC reported that startup Starcloud launched a satellite carrying an Nvidia H100 GPU in November 2025 and later ran and queried Google’s Gemma model there. Google’s stated distinction is narrower: it calls its test the first AI response from a data-center-class TPU in orbit. Its larger proposed difference is an architecture of satellites connected to operate together, rather than computing on separate spacecraft.
The economics in Google’s research remain conditional. The paper estimates orbital launch costs at about $14,700 per kilowatt per year at current prices, compared with about $810 if launch costs fall to $200 per kilogram. Google’s paper says that lower cost would be close to what U.S. data centers pay for power and estimates reaching the proposed scale would require roughly 1,800 Starship launches. Those estimates rely on future launch costs and a large deployment, not on the cost of operating this prototype.
““the first-ever AI response from a data center class TPU in orbit””
The Limits Beyond the Haiku
Google has not yet shown that a cluster of satellites can perform sustained, coordinated AI computing in orbit. The next two-satellite test is planned for 2027, but its schedule and results remain subject to change. It is also unclear how well the equipment will withstand prolonged radiation exposure, manage heat in a vacuum, or be repaired if hardware fails.
Cost and communications are unresolved too. Google’s estimates depend on launch prices falling substantially, and the source material does not establish that this will happen or that the projected economics include every operating cost. A large constellation would also need reliable links to satellites and ground stations amid existing orbital traffic; the supplied material cites about 47,000 tracked objects in orbit, without detailing how a future Suncatcher network would manage collision risks. The haiku alone does not answer these engineering or business questions.
Thermal Tests and Two-Satellite Links
Google says it will test the prototype’s TPU trays under thermal stress and radiation exposure. The results should help establish whether the chips can continue operating in orbit beyond brief demonstrations. Google has named twin interconnected satellites in 2027 as the following milestone; TechCrunch reports they are expected to test optical links and offer more power than the first spacecraft.
That mission would be a more direct test of Suncatcher’s central idea: satellites sharing computing work through high-speed links. Until Google reports results from those trials, the project remains an early feasibility effort. The company has not announced a commercial launch date or shown that its projected launch economics can be achieved.
Key Questions
What did Google’s satellite do?
Google says the satellite received a prompt during a ground-station pass over Svalbard and used its onboard TPU to generate a haiku with the Gemma 4 model. Engineers also tried other prompts and asked the payload to report its technical status.
Is this the first AI to run in space?
No. CNBC reported that Starcloud launched a satellite with an Nvidia H100 in November 2025 and later ran and queried Google’s Gemma model in orbit. Google’s claim is specifically about a response from a data-center-class TPU.
Can the satellite run AI continuously?
The reported prototype is limited to about 1 kilowatt of power and can operate one of its four TPUs at a time, according to TechCrunch. The source material describes short operating bursts, not continuous data-center-scale computing.
What is Google testing next?
Google says it plans to test how the TPU hardware handles thermal stress and radiation. The company has named two interconnected satellites in 2027 as its next milestone, which would test links between spacecraft.
Is orbital AI computing economically viable?
That has not been established. Google’s research estimates launch costs under current and hypothetical lower-price scenarios, but the lower estimate depends on launch costs falling to $200 per kilogram. The prototype does not demonstrate that a commercial orbital system can meet those assumptions.
Source: ThorstenMeyerAI.com
