🔍 Read the full analysis: AI In Space, Explained Through Google’s Haiku-Writing Satellite 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
A Google- and Planet-built satellite running Gemma 4 on a Google TPU generated and sent a haiku while passing over Svalbard, Google said on October 30, 2026. The small test shows that an AI chip can operate in orbit, but it does not establish that space-based computing is practical or economical at scale.
Google said October 30 that a small satellite built with Planet generated a haiku using Gemma 4 on a Google TPU and sent the result to a ground station during a pass over Svalbard, Norway. The test is the first reported result from Project Suncatcher, Google’s effort to assess whether solar-powered satellites could eventually support AI computing in orbit; it demonstrates a limited on-orbit operation, not a working space data center at commercial scale.
Google described the exchange as the satellite’s “Hello World” moment. Engineers sent a prompt asking for a haiku about space, and the satellite returned: “Blue marble below / Data streams in silent grace / Starlight feeds my core.” Google said staff also tested space-themed limericks and asked the system to report technical status. A Google spokesperson told TechCrunch the satellite was “operating nominally.”
The prototype launched on October 1, 2026, aboard a SpaceX rideshare mission, alongside more than 100 other spacecraft, according to the supplied account. After about three weeks of commissioning, Google said engineers powered up the host computer and TPU hardware and ran early machine-learning tests. TechCrunch reported that the satellite carries four TPUs but produces up to about 1 kilowatt of power, limiting operation to one chip at a time and short bursts. The company has not described the test as continuous or large-scale computing.
Google calls the milestone “the first-ever AI response from a data center class TPU in orbit.” That wording is specific to a TPU response: the source material says Starcloud had already put an Nvidia H100 GPU in orbit in November 2025 and later ran Google’s Gemma model there, as CNBC reported. Google’s stated distinction is its longer-term plan to connect satellites into computing clusters, rather than relying only on individual 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
Testing AI Beyond Earth
The test matters because it moves one part of AI computing into orbit: a data-center-class chip ran a model and returned an answer through a satellite-ground connection. If the approach can eventually operate at much greater scale, space-based solar power could offer another source of energy for computing at a time when data-center electricity demand is drawing attention. The test itself does not show that orbital computing can reduce costs, ease pressure on power grids, or serve customers.
Google’s paper argues that solar panels in a suitable orbit could be up to eight times more productive than on Earth. That is a company research estimate, not a result established by this single satellite demonstration. The prototype’s roughly one-kilowatt power ceiling also illustrates the gap between an experiment and the hundreds of megawatts used by large terrestrial data centers. The comparison is about scale, not equivalent computing capacity or energy costs.
For readers, the immediate takeaway is not that AI workloads are about to move off Earth. The result shows that an orbiting satellite can run a small AI task under test conditions. Whether this can become a reliable and affordable service depends on engineering and launch costs that remain unresolved.
Popular right now
Bose QuietComfort Ultra Wireless Noise-Cancelling Headphones with Spatial Audio, Black
Samsung 85-Inch QN70H Neo QLED 4K Smart TV (2026)
Ninja Foodi 10-Qt DualZone Smart XL Air Fryer, 6-in-1, DZ550
As an affiliate, we earn on qualifying purchases.
From Research Plan to Prototype
Google introduced Project Suncatcher in a research post and preprint on November 4, 2025. The proposal describes solar-powered satellites carrying TPUs, flying in close formation and communicating through optical, or laser, links. Google’s initial stated milestone was a learning mission with Planet involving two prototype satellites by early 2027. The first prototype launched earlier than that target, but its reported task remains a small operational test.
Google also published a peer-reviewed version of its research in Joule on October 1, 2026, according to the supplied source. In the paper’s cost estimates, a kilowatt of orbital capacity would cost about $14,700 per year in launch costs at current prices. The paper estimates that figure could fall to about $810 if launch costs reached $200 per kilogram, a rate the source compares with the cost of power at US data centers. The lower figure depends on a future launch price; it is not the prototype’s measured operating cost. The source says reaching that price would require roughly 1,800 Starship launches.
The proposal also requires satellites to exchange data at high rates. Google’s research sketches clusters of 81 satellites at about 650 kilometers altitude, with nearby spacecraft separated by 100 to 200 meters. The supplied material says Google reported a bench test of 1.6 terabits per second between a pair of optical transceivers. A bench result is not proof that a full satellite cluster can sustain that performance in orbit.
““the first-ever AI response from a data center class TPU in orbit””
Costs and Reliability Remain Open
The haiku test does not establish how reliably a TPU can run over long periods in orbit. Google says it still plans tests of thermal stress and radiation exposure. The source material does not provide results from those tests, a sustained computing benchmark, or evidence that the prototype has handled a commercial workload.
Several proposed system-level problems also remain. Heat is difficult to remove in a vacuum, spacecraft are hard to repair once launched, and links between satellites and ground stations must support useful workloads. The supplied account cites about 47,000 tracked objects in orbit, but does not specify the date or tracking criteria for that count; it should not be treated as a complete measure of collision risk. The available information does not resolve how Google would manage debris, maintenance, launch cadence, or the full cost of operating a large constellation.
Google’s cost projections depend on launch prices falling substantially. It is not clear when, or whether, the assumed price of $200 per kilogram could be reached, or what other costs would apply to building, launching and replacing the satellites. The company’s claim about solar productivity also does not by itself show that orbital computing would be cheaper or lower-impact overall.
Thermal Tests and Satellite Links
Google says the next step is to test how the TPU hardware performs under heat and radiation exposure. The company’s subsequent milestone is a pair of interconnected satellites in 2027, according to its stated plan. TechCrunch reported that the two spacecraft are expected to generate more power and test optical links between satellites.
Those tests would address whether separate spacecraft can exchange data as part of a computing system, a central feature of Project Suncatcher. Google has not provided a confirmed schedule for later missions or set out a date for commercial deployment. The next meaningful evidence will be performance and reliability results from the planned in-orbit tests, rather than another isolated AI-generated response.
Key Questions
What did Google’s satellite do?
Google said the satellite ran Gemma 4 on a Google TPU, generated a haiku in response to an uplinked prompt and sent the result to a ground station during a pass over Svalbard.
Was this the first AI system in orbit?
Not according to the supplied reporting. Starcloud launched a satellite carrying an Nvidia H100 in 2025 and later ran Google’s Gemma model there, CNBC reported. Google’s claim is narrower: it describes this as the first AI response from a data-center-class TPU in orbit.
Can the satellite run a full data center?
No. The prototype reportedly produces up to about 1 kilowatt and can operate one of its four TPUs at a time in short bursts. It is a test spacecraft, not a full-scale data center.
Why put AI computing in space?
Google’s proposal is based partly on the potential for solar panels in a suitable orbit to produce more power than on Earth. But launch expense, heat management, radiation, communications and repairs remain significant challenges, and the test has not shown that orbital computing is cost-effective.
What is Google testing next?
Google says it plans to assess the TPU’s performance under thermal stress and radiation. Its stated next milestone is two interconnected satellites in 2027 to test higher-power operation and optical links.
Source: ThorstenMeyerAI.com
