SpaceXAI announced plans to deploy Nvidia’s Vera central processing units to support its expanding AI infrastructure, including the Grok AI system. The integration will accelerate tool orchestration, code execution, data processing, and simulations, according to a Monday statement.
The deployment will leverage the Nvidia Vera Rubin platform, which combines accelerated computing, NVLink interconnects, Spectrum-X Ethernet networking, BlueField data processing units, and Nvidia’s software stack. SpaceXAI intends to scale its computing capacity using this architecture as part of its infrastructure roadmap.
The company is developing its first-generation Starmind AI satellite, designed around an optimized Nvidia Vera Rubin NVL72 system. This adaptation extends the platform’s use beyond terrestrial data centers to orbital deployment, reflecting a shift toward space-based AI computing.
Nvidia’s Vera CPUs feature 88 Olympus cores and LPDDR5X memory, delivering up to 1.2TB/s of memory bandwidth. The processors are reported to complete tasks up to 1.8 times faster than x86 CPUs on certain workloads, positioning them as a high-performance alternative for AI infrastructure.
Mike Nicolls, President of SpaceXAI, stated that Vera’s performance and memory bandwidth enable efficient handling of large-scale orchestration, code, and data processing while allowing graphics processing units to focus on their core functions.
Ian Buck, Nvidia’s Vice President of Hyperscale and High-Performance Computing, emphasized the Vera Rubin platform’s role in enabling scalable, high-efficiency AI deployments across diverse environments.
