SpaceX has acquired xAI, Elon Musk’s artificial intelligence company, in a move that brings AI development directly under the same roof as rockets, satellites, and space-based communications. The acquisition signals an ambitious effort to rethink how large-scale AI systems are powered and deployed, with space positioned as the long-term solution to the growing demand for compute.
The combined organization brings together SpaceX’s launch capabilities, its global satellite network, and xAI’s model development under a single, vertically integrated structure. According to SpaceX, the goal is to accelerate progress in artificial intelligence while addressing the mounting physical constraints of terrestrial data centers, including power consumption, cooling requirements, and environmental impact.
Current AI systems rely heavily on large ground-based facilities that consume enormous amounts of electricity. SpaceX argues that global energy demand for AI cannot be met sustainably using only Earth-based infrastructure. As models grow larger and workloads more intensive, the company sees space-based compute as a necessary next step rather than a distant science fiction concept.
The central idea behind the acquisition is the development of orbital data centers powered by near-constant solar energy. In orbit, satellites can harvest solar power with minimal atmospheric loss, avoiding many of the cooling and maintenance challenges faced by ground facilities. SpaceX says this approach could eventually deliver AI compute at lower cost than terrestrial alternatives.
Starship plays a key role in this vision. SpaceX notes that no previous launch system has been capable of placing the massive payloads required for space-based data centers into orbit at scale. While recent years have set records for orbital launches, total payload mass has remained relatively modest. Starship is designed to change that equation by delivering hundreds of tons per launch at high frequency.
This year, SpaceX plans to begin launching more powerful next-generation Starlink satellites using Starship, dramatically increasing capacity per launch. The same launch cadence and payload scale would be required for AI-focused satellite constellations. SpaceX envisions launch rates approaching one flight per hour, with each mission carrying roughly 200 tons, enabling millions of tons of infrastructure to be placed in orbit annually.
According to internal estimates, launching one million tons of satellites per year, each generating around 100 kilowatts of compute power per ton, would add approximately 100 gigawatts of AI compute capacity annually. Over time, SpaceX sees a path toward deploying even larger amounts of compute, potentially reaching terawatt-scale capacity as launch costs decline and manufacturing scales.
The company believes that within a few years, space-based compute could become the lowest-cost option for training large AI models and processing massive datasets. That cost advantage, SpaceX argues, would unlock faster progress in AI research, scientific discovery, and advanced engineering.
The acquisition also fits into SpaceX’s broader multi-planetary ambitions. Starship’s ability to transport heavy cargo beyond Earth orbit opens the door to sustained operations on the Moon and eventually Mars. SpaceX outlines a future where lunar facilities manufacture satellites using local resources, reducing reliance on Earth launches and enabling even larger AI constellations deeper in space.
Space sustainability remains a stated focus of the plan. SpaceX says future AI satellite systems will build on existing design and operational practices used for Starlink, including controlled deorbiting and end-of-life disposal to limit orbital debris.
While the timeline for operational space-based AI data centers remains uncertain, the acquisition of xAI formalizes SpaceX’s intention to push AI infrastructure beyond Earth. By combining launch, communications, and model development, the company is positioning itself to pursue one of the most unconventional approaches yet to scaling artificial intelligence.
