SpaceX targets late 2027 launch for Nvidia-powered AI satellites with 120-kilowatt compute output

2 Sources

Share

SpaceX accelerated its timeline to deploy AI satellites in late 2027, powered by Nvidia chips delivering 120 kilowatts of compute output. The company filed to launch up to one million data center satellites and signed deals with Google and Anthropic for compute access.

SpaceX has accelerated its orbital computing initiative, now targeting the fourth quarter of 2027 to launch its first Starmind AI satellites powered by Nvidia technology. CEO Elon Musk confirmed the timeline shift, moving up from the "as early as 2028" window disclosed in SpaceX's IPO prospectus, with plans to expand the program throughout 2028.

1

Nvidia Powers SpaceX's Space-Based AI Infrastructure

Nvidia announced that SpaceXAI, the entity operating SpaceX's AI infrastructure business, will build its first-generation Starmind satellite around an optimized Nvidia Vera Rubin NVL72 rack-scale system. This computing architecture already powers SpaceXAI's terrestrial AI factories running the Grok AI model.

1

The integration extends Nvidia's reach beyond ground-based data centers into orbital AI computing, marking a shift in how companies approach AI infrastructure expansion.

SpaceXAI will deploy Nvidia Vera CPUs to handle orchestration, code execution, and data processing surrounding AI model inference. The Vera chip features 88 Nvidia-designed Olympus cores and delivers up to 1.2 terabytes per second of memory bandwidth. SpaceXAI president Mike Nicolls explained that the chip "gives us the CPU performance and memory bandwidth to run enormous amounts of orchestration, code and data processing while keeping GPUs doing what they do best."

1

Technical Specifications of AI-Focused Satellites

The first-generation satellite, designated AI1, measures 70 meters tip to tip and sustains 120 kilowatts of compute output on average. Solar panels rated at 150 kilowatts power the space-based systems, while thermal management relies on a liquid-radiator architecture to handle the intense heat generated by continuous compute operations.

1

SpaceX CFO Bret Johnsen confirmed that Nvidia components will power the initial units.

The late-2027 launches represent proof-of-concept flights rather than revenue-generating service. These missions will validate the orbital computing technology before SpaceX scales the program. The company has filed regulatory applications to deploy up to one million data center satellites, signaling massive ambitions for space-based AI infrastructure.

1

Commercial Partnerships Drive Orbital Computing Forward

SpaceX has secured deals with Google and Anthropic to sell compute access from its orbital infrastructure. These partnerships position SpaceX's orbital computing initiative as a commercial service competing with traditional ground-based data centers. SpaceXAI is expanding its AI infrastructure with the broader Nvidia Vera Rubin platform as it moves toward gigawatts of computing capacity, according to Nvidia.

1

Ian Buck, Nvidia's vice president of hyperscale and high-performance computing, stated that SpaceXAI is "taking this architecture from massive AI factories to the next frontier of computing in orbit."

1

The collaboration leverages SpaceX's existing Starlink satellite network experience, which operates approximately 10,200 broadband and mobile satellites across 167 countries delivering connectivity to consumer, enterprise, and government customers.

2

SpaceX operates three segments: Space, which manufactures reusable rockets; Connectivity, running the Starlink network; and AI, which operates a vertically integrated platform spanning Grok, AI solutions, and computational infrastructure.

2

The orbital computing push integrates these capabilities, positioning SpaceX to capture demand from organizations requiring massive-scale data processing beyond Earth's atmosphere.

Today's Top Stories

© 2026 TheOutpost.AI All rights reserved