🔍 Read the full analysis: AI In Space: Lessons From Google’s First Data Center Satellite on ThorstenMeyerAI.com
Prime made for students and young adults
- Fast, free delivery for dorm and study essentials
- Prime Video and Amazon Music included
- Member-only deals
TL;DR
Google says a satellite built with Planet has produced an AI response from a data-center-class TPU in orbit, using its Gemma 4 model. The test shows that a small workload can run in space, but it does not establish that orbital data centers are practical or economical; power, heat, radiation, repairs and launch costs remain major hurdles.
Google says a satellite carrying its Tensor Processing Units (TPUs) has run the Gemma 4 AI model in orbit and returned a response to Earth, marking a small-scale test of Project Suncatcher, the company’s research into space-based AI computing. The satellite was built with Planet and launched on October 1; Google reported the response on October 30. The test confirms that a TPU can run an AI task in orbit, but it does not show that a space-based data center can yet compete with facilities on Earth.
After about three weeks of commissioning, engineers sent a prompt during a ground-station pass as the spacecraft passed over Svalbard, Norway. Google said the request asked Gemma 4 to write a haiku about space. The poem was downlinked to Earth, which the company described as its “Hello World” moment. Engineers also tested space-themed limericks and asked the payload to report its technical status.
The prototype uses a standard Planet satellite platform and carries four TPUs, according to TechCrunch. The spacecraft can generate up to about 1 kilowatt of power, limiting operation to one chip at a time and short bursts. TechCrunch reported bursts of roughly 20 minutes during the latest tests. Google said the satellite was “operating as expected”; a spokesperson told TechCrunch it was “operating nominally.”
Google’s next tests are intended to examine how TPUs perform under thermal stress and radiation exposure. The company has described a follow-on milestone involving two connected satellites in 2027. TechCrunch reported that the pair is expected to generate more power and test optical links between spacecraft. Those plans are future work, not results demonstrated by the current prototype.
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 experiment tests whether some computing work could eventually be done where solar power is abundant, rather than relying entirely on Earth-based data centers. Google’s research says solar panels in a suitable orbit could be up to eight times more productive than on Earth. That is a company estimate about a potential advantage, not a finding that this prototype has delivered cheaper or more efficient computing.
The economics remain unresolved. Google’s paper estimates that launch costs at current prices would make a kilowatt of orbital capacity cost about $14,700 a year. If launch costs fell to $200 per kilogram, the paper estimates the annual figure could reach about $810, near the electricity cost for US data centers. The paper’s scenario for deploying the needed infrastructure would require roughly 1,800 Starship launches. Those estimates depend on future launch prices and system designs; the present satellite is far too small to validate them.
The issue matters as electricity demand from computing grows. The source material cites US Department of Energy estimates that data centers used about 4.4% of US electricity in 2023, with the share projected to reach 6.7% to 12% by 2028. Moving computing to orbit is one proposed response, but the prototype does not show that it can relieve grid demand at meaningful scale.
Popular right now
Dyson V8 Cyclone Cordless Vacuum
Dyson Car+Boat Cordless Handheld Vacuum
Shark Rocket Corded Stick Vacuum (HV301) – Gray/Orange
As an affiliate, we earn on qualifying purchases.
From Research Plan to Orbit
Google introduced Project Suncatcher in November 2025 in a research post and preprint, outlining a concept for solar-powered satellites carrying TPUs and communicating through laser links. Its initial public milestone was a learning mission with Planet involving two prototype satellites, then expected by early 2027. The first satellite reached orbit earlier, on October 1, 2026, aboard SpaceX’s Transporter-18 rideshare mission. Google also published a peer-reviewed version of its research in the journal Joule that day.
Google’s distinction should be stated narrowly. The company calls this the first AI response from a data-center-class TPU in orbit; it is not the first AI chip or AI workload in space. CNBC reported that startup Starcloud launched a satellite carrying an Nvidia H100 in November 2025 and later ran and queried Google’s Gemma model in orbit. Google’s proposed difference is a network of closely spaced satellites linked by lasers, intended to work together like a computing cluster, rather than isolated AI workloads on individual spacecraft.
Google’s research sketches clusters of 81 satellites operating at about 650 kilometers altitude, separated by roughly 100 to 200 meters. That design remains a research concept. The source material says a bench test reached 1.6 terabits per second with a pair of optical transceivers, but a ground test does not establish that a large orbital cluster can sustain those speeds under operational conditions.
““the first-ever AI response from a data center class TPU in orbit.””
Costs and Orbital Risks Remain
The result is a demonstration of a narrow task, not evidence that orbital AI is commercially viable. It is not yet clear how reliably the hardware will perform over time, how much protection it will need from radiation, or how heat can be managed in a vacuum. The current prototype’s power and operating limits also differ greatly from the demands of a large computing cluster.
Other open issues include repairs, replacement and communication between satellites and ground stations. Google’s planned twin-satellite test has not yet demonstrated that optical links can support a distributed workload in orbit. The source material also cites about 47,000 tracked objects already in orbit, but does not specify a date or category breakdown for that count. The risks to a future constellation, and how it would be managed, remain unquantified here.
Google’s launch-cost estimates depend on prices that may change, while the required launch cadence and full operating costs are not established by the current mission. The satellite’s successful response cannot resolve those questions.
Tests Before a Two-Satellite Mission
Google says it will continue evaluating the current satellite, including tests of TPU performance under radiation and thermal stress. The next stated milestone is a pair of interconnected satellites in 2027, which would test whether spacecraft can exchange data through optical links while operating in orbit. The timing and detailed capabilities of that mission remain subject to the company’s plans.
Evidence from those tests would help show whether the concept can progress beyond one chip responding during a ground-station pass. For now, the confirmed result is limited: a compact satellite ran Gemma 4 on a TPU and returned an AI-generated response. Whether this approach can deliver useful computing at a competitive cost remains an open question.
Key Questions
What did Google’s satellite do?
It ran Gemma 4 on a Google TPU in orbit and returned an AI-generated response to a prompt sent during a ground-station pass. Google described the response as its “Hello World” moment for Project Suncatcher.
Was this the first AI system in space?
No. The source material reports that Starcloud launched a satellite carrying an Nvidia H100 in November 2025 and later ran AI workloads in orbit. Google’s narrower claim is the first response from a data-center-class TPU in orbit.
Can the satellite power a full data center?
No. The prototype generates up to about 1 kilowatt and can operate one of its four TPUs at a time in short bursts. It is a technology test, not a full-scale orbital data center.
Why is Google exploring computing in space?
Google’s research examines whether solar-powered satellites could provide energy for AI computing. The company estimates that solar panels in a suitable orbit could be more productive than on Earth, but the cost and operational case has not been established.
What is the next milestone?
Google has said it plans further tests of the current satellite and a two-satellite mission in 2027 to test interconnected operation. The current source material does not establish a launch date or the outcome of that future mission.
Source: ThorstenMeyerAI.com
