SpaceX filed plans with the Federal Communications Commission on Friday for a million-satellite data center network. The company merged with xAI on Monday, combining Elon Musk’s space and artificial intelligence ventures. The FCC accepted the filing on Wednesday, established a public comment schedule, and Chairman Brendan Carr shared it on X. Musk detailed the rationale for orbital data centers on Patrick Collison’s podcast.
The filing submitted by SpaceX outlines a network composed of one million satellites functioning as data centers in orbit. This proposal represents an extension of satellite-based infrastructure into computational operations for artificial intelligence applications. SpaceX, known for its Starlink constellation, positions this network as a means to host data processing capabilities directly in space, leveraging orbital advantages for power generation and scalability.
The FCC welcomes and now seeks comment on the SpaceX application for Orbital Data Centers.
The proposed system would serve as a first step towards becoming a Kardashev II-level civilization and serve other purposes, according to the applicant. pic.twitter.com/TDnUPuz9w7
— Brendan Carr (@BrendanCarrFCC) February 4, 2026
One week after the filing, SpaceX and xAI completed their formal merger on Monday. This union integrates SpaceX’s expertise in rocket launches and satellite deployment with xAI’s focus on advanced artificial intelligence development. The merger creates a single entity capable of pursuing integrated projects that span terrestrial launch operations and space-based computing infrastructure.
The Federal Communications Commission processed the SpaceX application on Wednesday by accepting it officially and initiating a schedule for public comments. This procedural step allows interested parties to review and respond to the proposal. FCC Chairman Brendan Carr, who has served in his role throughout recent administrations, posted the filing document directly on the platform X, drawing public attention to the submission.
Elon Musk elaborated on the concept of orbital data centers during an episode of Stripe co-founder Patrick Collison’s podcast “Cheeky Pint,” which included guest Dwarkesh Patel. Musk presented the core economic argument centered on power generation. He explained that solar panels in space benefit from uninterrupted sunlight exposure without atmospheric interference or nighttime cycles, enabling higher efficiency.
Musk stated, “It’s harder to scale on the ground than it is to scale in space.” He quantified the power advantage, noting, “Any given solar panel is going to give you about five times more power in space than on the ground, so it’s actually much cheaper to do in space.” This comparison highlights the reduced operational costs for energy-intensive tasks like AI training, where electricity constitutes a major expense.
New season of Cheeky Pint starts tomorrow! First up: @dwarkesh_sp and I sit down with @elonmusk to discuss space GPUs and much more. pic.twitter.com/dJ4dLluTMr
— John Collison (@collision) February 4, 2026
Dwarkesh Patel addressed limitations in the power-focused rationale during the discussion. He pointed out that electricity represents only one component of data center expenses, which also include cooling, maintenance, hardware procurement, and network connectivity. Patel specifically questioned the practicality of servicing graphics processing units that fail amid prolonged AI model training sessions in orbit, where physical access requires complex retrieval missions.
Musk responded by designating 2028 as a pivotal year for the technology’s adoption. He asserted, “You can mark my words, in 36 months but probably closer to 30 months, the most economically compelling place to put AI will be space.” This timeline projects a shift driven by accumulating advantages in cost and capacity within the next three years from the podcast date.
Musk extended his forecast further, predicting, “Five years from now, my prediction is we will launch and be operating every year more AI in space than the cumulative total on Earth.” This statement envisions annual orbital AI deployments surpassing all historical terrestrial installations combined by approximately 2029.
Current projections place global on-ground data center capacity at 200 gigawatts by 2030. This scale equates to roughly one trillion dollars in infrastructure value when constructed and operated on Earth. Such investments underscore the magnitude of existing terrestrial commitments, against which orbital alternatives would compete.
SpaceX’s core revenue from orbital launches aligns directly with the requirements for deploying and replenishing satellite-based data centers. The recent SpaceX-xAI merger positions the combined company to internalize launch costs within its AI operations. The conglomerate plans an initial public offering in the coming months, which will provide capital for expanded initiatives.
