Google has confirmed the launch date of the first Suncatcher project test satellite: 1 October 2026
On 1 October 2026, Google is launching a test satellite called MVP with four TPU chips from Vandenberg aboard a Falcon 9 rocket — the first step of the Suncatcher project toward solar-powered orbital AI data centers.
Google has confirmed the exact date and form of the first test launch of the Suncatcher project: an experimental satellite designated MVP will launch on 1 October 2026 aboard a SpaceX Falcon 9 rocket from Vandenberg Space Force Base in California. The refrigerator-sized satellite was built and tested by Planet Labs, and Google integrated four of its own TPU AI accelerators into it. The solar panels supply roughly 1 kilowatt of electricity, which according to The New York Times corresponds to the power draw of a microwave oven or a hair dryer. Google VP James Manyika says he expects the output to be sufficient to process basic AI queries directly from orbit.
Suncatcher is a project announced last year as Google's long-term attempt to design AI data centers located in orbit and powered by solar energy — an alternative to ground-based data centers, whose energy consumption has drawn criticism. SpaceX and Blue Origin are working on similar concepts. Google originally planned to launch two of its own satellites only in 2027, but decided to accelerate the plan and use for the first test a satellite that Planet Labs already had ready.
Google manager Travis Beals estimates that matching the output of a single ground-based data center with a capacity of 1 gigawatt would require approximately 10 000 similar satellites. The main obstacles include the lack of air for cooling chips in space, hardware damage from cosmic radiation, and the need to reduce the cost of launching cargo into orbit to approximately 200 dollars per kilogram for the concept to be economically viable. According to sources, Jeff Bezos estimates that orbital data centers could become cheaper than ground-based ones in as much as 20 years.
Why it matters
This is an early technical trial of a concept that, according to Google and its competitors (SpaceX, Blue Origin), is intended to eventually address the growing energy demands of AI computing by moving part of the infrastructure into space. However, the parties involved themselves state that economic viability is far off — according to Jeff Bezos's estimate, as much as 20 years away — and fundamental obstacles still need to be solved, such as cooling, radiation resistance of chips, and the cost of launching cargo into orbit. In practice, then, this is for now a research milestone with no immediate impact on the availability or price of AI services.
What was added since the original report
Verified updates
-
The satellite is being launched on a SpaceX Falcon 9 from Vandenberg Space Force Base in California; Google VP James Manyika claims the satellite has enough power for basic AI queries from orbit; Estimate: approximately 10 000 satellites would be needed to match the output of a single gigawatt-scale ground-based data center; Technical obstacles: lack of cooling air in orbit, chip damage from cosmic radiation, high launch costs
- The satellite is being launched on a SpaceX Falcon 9 from Vandenberg Space Force Base in California
- Google VP James Manyika claims the satellite has enough power for basic AI queries from orbit
- Estimate: approximately 10 000 satellites would be needed to match the output of a single gigawatt-scale ground-based data center
- Technical obstacles: lack of cooling air in orbit, chip damage from cosmic radiation, high launch costs
Check the original
Event sources
confirmed by 2 independent sources · 2 publishers, 2 independent. We count feeds from the same owner only once.