Elon Musk: SpaceX and Nvidia's Star Mind Computer to Enter Mass Production Next Year
Elon Musk stated that the Star Mind computer, jointly engineered by SpaceX and Nvidia, exceeds most expectations and will enter mass production next year, while praising the collaboration with Nvidia's engineering team.
Jensen Huang expressed his pleasure in working with Musk, noting that both engineering teams are jointly designing a new server architecture for deploying data centers in space, which may be applicable both in space and on the ground.
SpaceX had previously announced a collaboration with Nvidia to design the Starmind AI1 satellite computing payload, with each satellite equipped with Nvidia Rubin GPUs and Vera CPUs, providing data center-level computing capabilities in space, and plans to run AI workloads in low Earth orbit.
Musk confirmed that SpaceX will exclusively adopt Nvidia's architecture, believing that Vera Rubin is the best AI computer, and plans to start launching related satellites next year, aiming for significant scale by 2028.
The Star Mind/Starmind project aims to eliminate ground power, land, and cooling constraints through orbital satellites, utilizing solar energy and space radiation cooling, with a target scale of up to a million satellites.
In market mechanisms, the buyer is SpaceX's demand for orbital AI computing capabilities, while the seller is Nvidia's Rubin and Vera platforms; the event is driven by the joint design of both engineering teams, with funds and technology flowing from ground data center chips to optimize servers in space, benefiting Nvidia with large exclusive orders and SpaceX with expanded computing power, while ground data centers face power and cooling bottlenecks.
This collaboration optimizes Nvidia's rack-level systems for orbital environments and may also be applied to ground deployments, forming a shared architecture between space and ground.
Source: Public Information
ABAB AI Insight
SpaceX announced in August 2026 a collaboration with Nvidia to design the Starmind AI1 satellite computing payload, utilizing the Vera Rubin NVL72 architecture, with each satellite carrying dozens of Rubin GPUs; Musk later confirmed the exclusive use of Nvidia and set targets for launches next year and scaling by 2028.
The capital pathway reflects the re-engineering of Nvidia's ground rack-level systems into space-optimized versions, while leveraging SpaceX's launch capabilities and Starlink links to shift AI computing from ground power and cooling constraints to orbital solar energy and radiation cooling; Musk's public praise further enhances the visibility of the collaboration between the engineering teams.
This is similar to historical aerospace projects where chip manufacturers redesign hardware for extreme environments, and also resembles attempts to expand data centers from single ground facilities to distributed orbital nodes; currently, it is in the phase of prototyping from ground AI factories to orbital data centers.
Essentially, this represents a restructuring of the industrial chain: transforming server architectures originally designed for ground data centers into versions that adapt to space radiation, solar power, and vacuum cooling, shifting computing nodes from reliance on the terrestrial power grid to orbital deployment, thereby changing the physical distribution and energy sources of AI computing power.
ABAB News · Cognitive Laws
The best friend is also the largest chip order
Moving ground architecture to space is the true path to mass production
Exclusive collaboration locks in supply better than public announcements