BlockBeats News, August 5th, Nvidia is collaborating with SpaceX to develop a satellite for orbital computing called Starmind AI1. The plan is to deploy an AI data center in space, equipped with Nvidia's next-generation Rubin GPU and Vera CPU, providing data center-level computing power in the space environment.
SpaceX stated that the Starmind project aims to build an orbital AI computing network powered by solar energy. It will utilize high-speed laser links between Starlink satellites to connect to the ground, creating a "space data center" architecture. The project hopes to migrate some AI computing from the ground to orbit to alleviate the land, power, and cooling constraints faced by current AI infrastructure.
It was revealed that the deployed Starmind AI1 will have a height of approximately 30 meters, a wingspan of about 75 meters, a peak computing power consumption of up to 250 kilowatts, and an average of around 175 kilowatts. The overall design is more akin to a "flying computing facility."
However, space AI data centers still face many challenges, including radiation exposure, hardware reliability, communication latency, equipment maintenance, and launch costs. Currently, no company has proven that large-scale orbital data centers can surpass ground-based data centers in terms of cost-effectiveness and operational efficiency.
Nvidia's participation this time will further expand its AI infrastructure footprint. The Rubin GPU and Vera CPU were originally aimed at the next generation of AI data centers, and the Starmind project will attempt to deploy these high-performance computing chips in the orbital environment.
SpaceX's advantage lies in its in-house rockets and satellite network, which can reduce launch costs through Falcon 9 and future Starship missions. By integrating Starlink, SpaceX can achieve an integrated layout of computing, communication, and launch capabilities.
Reports indicate that SpaceX has previously proposed the construction of up to 1 million orbital data center satellites in the future. If this plan moves forward, it could alter the development path of AI infrastructure but may also lead to issues such as orbital congestion, space debris, and regulatory challenges.
Analysts believe that in the short term, Starmind is more likely to be applied in satellite image processing, scientific computing, national defense, and other scenarios. Whether it can evolve into a true "orbital cloud computing network" depends on cost, reliability, and scalability.
