The US Space Race is Around Space-Solarized Data Infrastructure

By Selva Ozelli Esq, CPA, Author of Sustainably Investing in Digital Assets Globally

This is the second article in a series of articles I am writing for Irish Tech News  to explore the financial, technical, legal aspects of utilizing space solar energized orbital data centers that are rapidly evolving into “AI Factories, designed specifically to convert massive amounts of electrical power into intelligence, measured in tokens”  around the world.

The US Space Race

My new series is a follow up to an interview ITN conducted with me in 2020 exploring how space solar energy could sustainably energize the tokenization of the global  financial markets which is projected to grow to multi-trillion dollars by the end of the decade.

The shift toward space-solarized data infrastructure is accelerating in the US rapidly following the historic March 1, 2026, drone strikes on AWS data centers in the United Arab Emirates and Bahrain which has extended during April and May. Executed by Iran’s Islamic Revolutionary Guard Corps (IRGC), these kinetic strikes marked the first time commercial hyperscale data centers were directly targeted and physically damaged in active warfare. The attacks caused prolonged service disruptions, exposed the vulnerability of terrestrial tech infrastructure, and proved that earth bound data centers are now prioritized military targets.

As detailed in the table below US technology and aerospace companies are increasingly looking to space-solarized solutions to address the immense energy and cooling demands of AI, with several key initiatives emerging.

US Tech and Aerospace Companies Focused on Space Solarized Data Centers

Two Distinct Approaches in Space Solarized Data Center Operations in the US

US technology and space companies in a race are aggressively pursuing orbital and space-solarized data centers and are tackling these operations through two distinct methodologies: orbital data processing (in-space edge compute) and space-based terrestrial power harvesting. Both approaches aim to bypass the escalating energy demands, cooling constraints, and land footprint limitations of Earth-based data center infrastructure.

The two approaches differ significantly in how they utilize space and solar resources. Here is a summary:

Terrestrial vs. Space-Based AI Compute

In-Orbit Data Centers (Space-Based Edge Compute Integration)

This method involves launching actual server racks, GPUs, and computing hardware directly into orbit typically in Low Earth Orbit- LEO or Lunar Data Centers focusing on processing space-gathered data at the point of collection rather than transmitting raw telemetry back to Earth for processing.

LEO satellites draw power directly from the sun without atmospheric interference, operating outside the day-night cycle to achieve constant energy and compute processing. Heat is dissipated into the vacuum of space using advanced thermal radiators.  Satellites use localized server arrays or tensor processing units (TPUs) to run real-time AI inference and cloud workloads in orbit.

US technology  companies as detailed in the table above like SpaceX (via its xAI data center satellite constellation proposals), Nvidia (with its Starcloud initiative),  Lone Star, Axiom Space, Aetherflux rebranded as Cowboy Space, Capella Space, and Planet Labs in partnership with Google’s Project Suncatcher—are actively integrating high-power chips (such as H100s)  to generate insights and train models directly in orbit. These US technology companies and aerospace firms are actively racing to build world AI computing infrastructure, aiming to construct dedicated “Level 2” orbital data centers. By mounting silicon chips like Nvidia H100s and TPUs in orbit, these firms hope to bypass Earth’s power and cooling limits, executing massive AI model training in the vacuum of space.

Jeff Bezos’ private space firm, Blue Origin, filed an FCC application for Project Sunrise—a 51,600-satellite constellation dedicated to orbital data processing. Meanwhile, Amazon Web Services (AWS) entered a strategic partnership with Starcloud to develop space-based cloud networks. Bezos noted on MarketWatch that the transition to orbital data centers will realistically take 10 to 20 years and to block SpaceX’s space-based ambitions in March 2026, Amazon Leo (formerly Project Kuiper) filed an FCC petition arguing that SpaceX’s proposed one-million-satellite constellation was a “lofty ambition” and “speculative placeholder” that could threaten space safety.  The legal challenge unfolded while Amazon itself faced regulatory pressure, trailing significantly behind its FCC milestone to deploy half of its intended ~3,200 broadband satellites, forcing the company to seek deadline extensions.

Meanwhile, on December 10, 2025 PowerBank Corporation (Canada) already launched the inaugural  DeStarlink Genesis-1 satellite, marking Orbit AI’s (Singapore)  first step toward building its space solarized Orbital Cloud network — an architecture where AI compute, connectivity, and blockchain (Ethereum) – verified processing occur directly in low-Earth satellites.

Terrestrial Power Harvesting (Space-to-Earth Energy)

The surging power demands of AI workloads have put immense strain on the domestic energy grid, forcing US technology companies to seek unconventional energy sources. Meta’s data center footprint consumed over 18,000 gigawatt-hours in 2024 alone.

Instead of putting servers in space, this approach keeps the data centers on the ground but uses  satellites in geosynchronous orbit (GEO) capture sunlight without the hindrance of weather, clouds, or nighttime, then transmit this energy to terrestrial receivers located directly at or near data center sites. This GEO positioning yields up to \(5 \times\) the energy of standard ground-based solar farms, bypassing terrestrial power grid constraints, eliminating interconnection bottlenecks, and providing instantaneous, 24/7 clean energy to highly demanding data center hubs.

Amazon Web Services (AWS)  indicated that it remains focused on terrestrial data center expansion and has entered a  partnership with Orbital Materials to focus on terrestrial data center infrastructure. The collaboration utilizes artificial intelligence to design and test new synthetic materials aimed at making traditional data centers more sustainable through targeted carbon removal and optimized cooling systems.

Meanwhile, with a first-of-its-kind commercial agreement, Space-based solar power (SBSP) in GEO is transitioning from theory as explained by Dr. Paul Jaffe to commercial implementation.    Meta announced an agreement with startup Overview Energy to secure up to 1 gigawatt (GW) of space-based solar power capacity for its AI data centers. By beaming continuous energy via near-infrared lasers from geosynchronous orbit, the technology bypasses grid constraints by illuminating terrestrial solar farms at night to provide round-the-clock generation.

Selva Ozelli Esq, CPA is an international  legal  expert  and author of Sustainably Investing in Digital Assets Globally.  Her  writings are translated into 45 languages and republished in over 200 global publications.  She is recognized as an expert media/TV commentator on global digital asset regulation, tax,  technology and aerospace matters.

See more breaking stories here.