The Emergence of Orbital AI Data Centers: A New Frontier in Computing

Jeff Bezos’ aerospace company, Blue Origin, and Elon Musk’s SpaceX are both advancing technologies for artificial intelligence (AI) data centers in space. Blue Origin has been developing the necessary technology for over a year, while SpaceX plans to utilize an upgraded version of its Starlink satellites to host AI computing payloads. This initiative comes as demand for electricity and water to cool terrestrial servers continues to escalate. Bezos has predicted that gigawatt-scale data centers could be operational in space within the next 10 to 20 years, asserting that solar energy availability in space would allow these facilities to outperform Earth-based counterparts.

New York PostAn AI Model Has Been Trained in Space Using an Orbiting Nvidia GPUCNBC

In a significant development, the Nvidia-backed startup Starcloud successfully trained an AI model in space using its Starcloud-1 satellite, which was launched on November 2, 2025. This satellite is equipped with an Nvidia H100 graphics processing unit, marking the first instance of a high-powered GPU operating in space. Starcloud's CEO, Philip Johnston, emphasized that the satellite's capabilities demonstrate the feasibility of space-based data centers, which could alleviate the increasing strain on Earth's digital infrastructure. Johnston stated that these orbital centers could achieve energy costs ten times lower than those of terrestrial data centers.

CNBCAn AI Model Has Been Trained in Space Using an Orbiting Nvidia GPU

Starcloud's operations include running an open large language model (LLM) named Gemma, developed by Google, and training another model called NanoGPT, based on the complete works of Shakespeare. The company aims to build a 5-gigawatt orbital data center that would harness constant solar energy, thus avoiding the limitations of Earth's day-night cycles and weather conditions. Johnston noted that these advancements not only address technological ambitions but also align with environmental responsibilities, as terrestrial data centers consume vast amounts of water and produce significant greenhouse gas emissions.

CNBC

Despite the promising developments, skepticism remains regarding the practicality of orbital data centers. Critics point out the high costs and technical challenges associated with cooling systems in the vacuum of space, where traditional air-based cooling is ineffective. Starcloud is exploring alternative cooling solutions, including air-based or liquid-based systems, and plans to deploy large radiators in space to manage heat dissipation.

An AI Model Has Been Trained in Space Using an Orbiting Nvidia GPU

Starcloud's achievements have garnered attention from industry experts. Tris Warkentin from Google DeepMind remarked on the significance of running Gemma in space, highlighting the robustness of open models. Meanwhile, Starcloud's Chief Engineer, Adi Oltean, described the initiative as a crucial step toward reducing the burden on Earth's energy supplies by leveraging abundant solar energy in space.

An AI Model Has Been Trained in Space Using an Orbiting Nvidia GPU

Looking ahead, Starcloud plans to launch a second satellite, Starcloud-2, which will feature additional GPUs and enhanced capabilities. This ongoing development signals a potential shift in how computing resources are managed, with implications for both commercial and military applications, such as real-time intelligence and disaster response.

CNBC

• “We will be able to beat the cost of terrestrial data centers in space in the next couple of decades,” — Jeff Bezos, Founder of Blue Origin

New York Post

• “Anything you can do in a terrestrial data center, I'm expecting to be able to be done in space.” — Philip Johnston, CEO of Starcloud

CNBC

• “seeing Gemma run in the harsh environment of space is a testament to the flexibility and robustness of open models.” — Tris Warkentin, Product Director at Google DeepMind

CNBC

• “This is a significant first step toward moving almost all computing off Earth to reduce the burden on our energy supplies and take advantage of abundant solar energy in space,” — Adi Oltean, Chief Engineer at Starcloud

An AI Model Has Been Trained in Space Using an Orbiting Nvidia GPU