Google Launches First Orbital AI Test Satellite on October 1

Google is launching an experimental satellite named MVP on October 1, 2026, to test AI hardware in space. The refrigerator-sized satellite, built by Planet Labs and carrying four Tensor Processing Units, will run for 15-minute intervals due to cooling limits. This first step in Project Suncatcher aims to validate orbital computing before a planned 2027 dual-satellite launch.
Key points
- Launch scheduled for October 1, 2026, via SpaceX Falcon 9 on the Transporter-18 rideshare mission.
- Satellite 'MVP' contains four TPUs powered by 1 kilowatt of solar energy, enough for simple AI queries.
- Chips can only operate for 15-minute bursts due to vacuum cooling limitations; no airflow exists in space.
- Project Suncatcher, announced in May 2025, seeks to leverage near-constant solar power in low Earth orbit.
- Future plans include linking satellite clusters via high-bandwidth lasers for larger AI workloads by 2027.
Background
This launch follows Google's May 2025 announcement of Project Suncatcher, a 'moonshot' to explore orbital AI infrastructure. It aligns with broader industry concerns about terrestrial data center power demands, as noted by Elon Musk after SpaceX acquired xAI. Recent archive coverage highlights Google's simultaneous €13 billion investment in Finnish data centers, underscoring a dual strategy of expanding terrestrial infrastructure while testing orbital alternatives to mitigate energy and cooling constraints.
How outlets are covering it
Mashable and Ars Technica emphasize the experimental nature of the MVP satellite, noting it is a precursor rather than a full data center. The New York Times highlights the rigorous vibration testing conducted in San Francisco, where technicians confirmed the hardware survived launch simulations. Google's official blog focuses on the long-term vision of linking satellite constellations via lasers, while Ars Technica points out that the satellite was adapted from a Planet Labs design to accelerate the timeline. All sources agree that cooling remains the primary technical hurdle, with Google testing heat pipes and radiators to manage heat in a vacuum.
Why it matters
This launch represents a critical proof-of-concept for orbital AI computing, potentially reducing reliance on terrestrial power grids. Success could validate space-based infrastructure for handling massive AI workloads, addressing energy and cooling bottlenecks that currently limit terrestrial data center expansion. It also signals a shift in how major tech companies approach infrastructure scalability, moving from ground-based to orbital solutions.
What to watch
Google expects the MVP satellite to operate for one year, collecting data on radiation resistance and thermal performance. The next milestone is a 2027 launch of two custom satellites to test inter-satellite laser connectivity. Full-scale orbital data centers are not expected for years, as the company must first resolve engineering challenges related to cooling and hardware durability in space.
- Google is launching an AI data center into space on Oct. 1 mashable.com
- Google Takes the A.I. Data Center Race to Outer Space The New York Times
- Behind Project Suncatcher, our moonshot to put AI in space blog.google
- Google’s first Suncatcher orbital data center test launches October 1 Ars Technica
- Elon Musk Responds With Two Emojis as Google CEO Sundar Pichai Says Project Suncatcher Heads to Space: 'C Benzinga
Want the full story? Read the original reporting
Read on mashable.com