🔍 Read the full analysis: How Google’s Satellite Haiku Puts AI In Space Into Perspective 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 haiku from orbit this week, a first test of its Project Suncatcher concept. The prototype shows that limited AI computing can run in space, but it does not establish that orbital data centers are practical or economical.
Google says a small satellite carrying one of its data-center-class TPUs generated a haiku while passing over Svalbard this week, marking an early working test of Project Suncatcher, the company’s plan to study AI computing in orbit. The demonstration confirms that the prototype could receive a prompt and return an AI-generated response; it does not show that space-based data centers are ready to compete with facilities on Earth.
The satellite launched on October 1 aboard SpaceX’s Transporter-18 rideshare mission, alongside more than 100 other spacecraft, according to the source report. Built with satellite company Planet on a standard Planet platform, it carries four Google TPUs. Google said the craft made contact and was operating as expected after reaching orbit, then spent about three weeks in commissioning and early tests.
For the demonstration, engineers sent a prompt asking Google’s Gemma 4 model for a haiku about being in space. Google said the satellite returned the three-line poem during a ground-station pass over Svalbard, Norway. The company described it as the first response from a data-center-class TPU in orbit. Engineers also tested space-themed limericks and asked the payload to report technical status, according to the account.
The hardware is still highly constrained. TechCrunch reported that the satellite can generate up to about one kilowatt of power and can operate only one TPU at a time. The chip runs in bursts: the report described roughly 15-minute intervals before launch and about 20 minutes during the recent test. Those limits make this a technology demonstration, not an orbital facility capable of serving large-scale AI workloads.
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 Haiku, Not an Orbital Data Center
The result matters because it moves Project Suncatcher from a research proposal to a small in-orbit computing test. It provides evidence that a Google TPU and AI model can operate together in space and return output to Earth. That is a meaningful engineering milestone, but the poem itself says little about the economics, reliability or scale of a future system.
Google’s stated rationale is access to solar energy. Its research says solar panels in a suitable orbit could be up to eight times more productive than on Earth. That could become relevant as electricity demand from data centers grows: the U.S. Department of Energy estimates that data centers used about 4.4% of U.S. electricity in 2023 and could account for 6.7% to 12% by 2028. Those figures describe a potential pressure that helps explain the research; they do not establish that moving computing into orbit would ease it.
The cost calculation remains a major obstacle. Google’s paper estimates that, at current launch prices, putting a kilowatt of computing capacity in orbit would cost about $14,700 per year in launch costs. The estimate falls to about $810 per year if launch costs reach $200 per kilogram, near the electricity cost cited for U.S. data centers. The lower figure depends on a future launch price, and the source says reaching it would require roughly 1,800 Starship launches. These are research estimates, not operating costs measured from this prototype.
Products Worth Considering
Logitech MX Keys S Wireless Keyboard, Graphite
Dell UltraSharp 27 Thunderbolt Hub Monitor (U2726DER)
Apple MacBook Air 13-inch with M5 chip, 16GB memory, 1TB SSD, and 12MP Center Stage camera (Midnight)
As an affiliate, we earn on qualifying purchases.
From Research Paper to First Response
Google introduced Project Suncatcher in a research post and preprint on November 4, 2025. The concept is a constellation of solar-powered satellites carrying TPUs and connected by high-speed laser links, so multiple spacecraft could work together. Google initially described a learning mission with Planet involving two prototype satellites by early 2027. The first prototype’s October launch came ahead of that stated milestone, but it is a smaller step than a functioning multi-satellite cluster.
The haiku is not the first reported example of AI hardware running in orbit. The source material says startup Starcloud launched an Nvidia H100 in November 2025, and CNBC later reported that the company had trained a small language model and run Google’s Gemma model there. Google’s narrower distinction is its claim about a response from a data-center-class TPU. Its longer-term plan also differs: Suncatcher proposes closely grouped satellites linked to work together, rather than relying only on computing performed separately on individual spacecraft.
Google’s paper describes a possible cluster of 81 satellites at an altitude of about 650 kilometers, with neighboring craft 100 to 200 meters apart. It also reports a bench test reaching 1.6 terabits per second with a pair of optical transceivers. That laboratory result is not a demonstration of the same performance between satellites in orbit. The company’s proposed architecture still depends on proving that close formation and fast links can work reliably in space.
“the first-ever AI response from a data center class TPU in orbit”
— Google, describing the milestone
Power, Heat and Repair Risks
The haiku test does not answer whether orbital AI systems can run continuously, handle demanding workloads or return enough value to justify launch and operations costs. Google’s next tests will examine thermal stress and radiation, but the source material provides no results from those trials yet. It also does not specify a long-term service life or a plan for repairing failed hardware in orbit.
Other unresolved issues include power supply, heat removal in a vacuum, communications with ground stations and the reliability of laser links between moving satellites. The source report says roughly 47,000 objects are already tracked in orbit, adding a crowded-orbit concern to the engineering questions. No evidence in the described demonstration settles how a large constellation would manage those risks, or what its full construction and operating costs would be.
The claim that this was the first AI response from a data-center-class TPU is attributed to Google. It should not be read as proof that this was the first AI model, chip or computing experiment in orbit; the source material describes earlier Starcloud activity. The precise comparison depends on the hardware and milestone being discussed.
Twin-Satellite Tests in 2027
Google says it plans further tests of the prototype, including exposure to radiation and temperature extremes. The next stated milestone is a pair of interconnected satellites in 2027. According to TechCrunch, the two craft are expected to generate more power and test optical links between satellites.
Those tests will provide a more relevant measure of Suncatcher’s central idea: whether separate spacecraft can share data quickly enough to operate as a larger computing system. Until Google reports their results, the current haiku remains evidence of a successful, limited prompt-and-response test—not proof that orbital data centers can match terrestrial infrastructure in capacity or cost.
Key Questions
What did Google’s satellite do?
Google says a satellite carrying a TPU ran its Gemma 4 model to answer a prompt with a haiku about space. The response was sent down during a ground-station pass over Svalbard, Norway.
Was this the first AI model in space?
Not according to the source material. It describes earlier Starcloud work involving an Nvidia H100 and Google’s Gemma model in orbit. Google’s stated distinction is a response from a data-center-class TPU.
Can the satellite run a full AI data center?
No. The prototype generates up to about one kilowatt and can operate one TPU at a time in short bursts. It is a test platform, not a full-scale data center.
Why is Google testing AI computing in orbit?
Google is investigating whether solar-powered satellites could provide energy for computing and whether linked satellites could work together. The company’s research estimates potential solar advantages in some orbits, but the cost and engineering case remains unproven.
What is the next milestone?
Google has said the next major step is to test two interconnected satellites in 2027, alongside further tests of the current hardware under radiation and thermal stress.
Source: ThorstenMeyerAI.com
