TradingKey - Nvidia (NVDA) announced Monday that SpaceX (SPCX)'s artificial intelligence division, SpaceXAI, will deploy Nvidia Vera CPUs to develop and run next-generation agentic AI applications. The chip will be used across multiple projects, including the Grok chatbot, helping handle tasks among tool calling, code execution, data analysis, and model inference, thereby reducing response times for AI agents.

Nvidia stated that Vera is its first CPU designed specifically for AI agents. SpaceXAI will use Vera to manage application-layer workloads and boost GPU utilization efficiency. As Grok-related business expands, SpaceXAI also plans to scale up its AI infrastructure based on the Nvidia Vera Rubin platform, pushing compute capacity toward the gigawatt level.

Driven by news, SpaceX's stock price briefly turned positive after falling over 2% in intraday trading. As of press time, it was down 0.28% at $136.58.

SpaceX stock price chart, Source: TradingView

Reportedly, SpaceXAI plans to extend this computing architecture from terrestrial data centers into space. The company's first Starmind AI satellite is slated to use an optimized Nvidia Vera Rubin NVL72 rack-scale system, bringing the accelerated computing architecture used in terrestrial AI factories into an orbital environment. Nvidia and SpaceXAI are currently adapting for space operating conditions—including power supply, thermal management, communication bandwidth, reliability, and hardware integration—aiming to maintain a unified computing architecture and software ecosystem.

This means SpaceXAI's AI infrastructure planning is no longer limited to terrestrial data centers, but has begun exploring orbital computing. Once the satellite is operational, certain AI computing tasks are expected to be completed in space. However, the relevant projects remain in the planning and development stages, and Nvidia has not yet disclosed a specific launch timeline for the Starmind satellite.

SpaceX has been accelerating the construction of its computing infrastructure recently. Earlier this month, the company and Tesla announced plans to build a semiconductor manufacturing facility in Texas, aiming to produce approximately 1 terawatt of computing capacity per year. The adoption of Nvidia's Vera and Vera Rubin platforms demonstrates that SpaceXAI is seeking to support Grok, terrestrial AI factories, and future orbital computing projects with a unified chip and system architecture.

Find out more