Key Takeaways
- Launch expenses need to plummet below $300/kg from current rates of $1,500ā$3,600/kg to make space-based data centers financially feasible, according to BNP Paribas
- Building a 1-GW orbital computing facility today would exceed $100B compared to $35Bā$50B for terrestrial infrastructure
- BNP anticipates Google, Amazon, and Meta will spearhead preliminary testing once economic barriers diminish
- Elon Musk projects space becoming the premier location for AI infrastructure within 30ā36 months
- Nvidia-supported Starcloud deployed the inaugural Nvidia H100 GPU to orbit in November 2025
The concept of establishing data centers in orbit is transitioning from theoretical possibility to serious strategic consideration. Financial institution BNP Paribas recently released comprehensive research examining this prospect, concluding that substantial promise exists alongside significant economic hurdles.
Current launch economics present a formidable barrier. With per-kilogram costs ranging from $1,500 to $3,600, constructing a 1-gigawatt orbital computing facility would exceed $100 billion in total investment. By comparison, equivalent ground-based infrastructure requires $35 billion to $50 billion.
BNP’s analyst Nick Jones indicated the firm doesn’t consider orbital data centers a “viable near- to medium-term solution.” He highlighted prohibitive launch expenses, costly space-qualified components, and complex challenges surrounding thermal management and power distribution in orbital environments.
Economic feasibility hinges on reducing launch costs to under $300 per kilogramārepresenting a dramatic decrease from present rates.
Should this cost threshold be achieved, BNP predicts Google, Amazon, and [[LINK_START_0]]Meta[[LINK_END_0]] will emerge as frontrunners in conducting preliminary experimental deployments of orbital computing infrastructure. The analysis didn’t specify projected timeframes for these developments.
The Terrestrial Energy Constraint Driving Space Ambitions
Interest in orbital solutions stems largely from AI’s voracious energy appetite. Terrestrial data facilities are consuming unprecedented electricity volumes. Department of Energy data shows US data centers represented approximately 4.4% of national electricity usage as of 2023.
McKinsey projects global data center infrastructure will demand $6.7 trillion in capital investment through 2030 to satisfy growing requirements. Major technology corporations anticipate capital expenditures reaching $600 billion in 2026, with Amazon individually committing $200 billion.
[[LINK_START_1]]Elon Musk[[LINK_END_1]] has positioned space-based computing as fundamental to SpaceX‘s strategic roadmap. He’s projected that within 30 to 36 months, orbital environments will represent the “most economically compelling place” for AI infrastructure deployment. SpaceX envisions launching one million satellites functioning as distributed orbital computing nodes, each delivering approximately 100 kilowatts of computational capacity per ton.
Musk’s rationale centers less on operational cost savings and more on power accessibility. He’s emphasized that electrical generation capacity outside China has remained essentially stagnant, creating fundamental questions about energy sourcing for expanding terrestrial data center networks.
[[LINK_START_2]]SpaceX[[LINK_END_2]] Actively Recruiting for Orbital Computing Initiative
[[LINK_START_2]]SpaceX[[LINK_END_2]] has progressed beyond conceptual planning stages. Michael Nicolls, who serves as vice president of Starlink Engineering, announced via X that the organization is actively recruiting for “many critical engineering roles” supporting space-based data center development, including a Space Lasers Engineer position located in Redmond, Washington.
The company disclosed plans to acquire Musk’s artificial intelligence venture xAI, explicitly identifying space-based AI infrastructure as a strategic long-term objective.
Experimental Deployments Already Underway
In November 2025, Nvidia-supported startup Starcloud successfully deployed the first Nvidia H100 GPU to orbit aboard a [[LINK_START_2]]SpaceX[[LINK_END_2]] Falcon 9 launch vehicle. The Starcloud-1 satellite weighed approximately 60 kilogramsācomparable to a compact refrigerator.
Starcloud’s ultimate vision encompasses a 5-gigawatt orbital computing complex spanning approximately 4 kilometers, incorporating extensive solar collection and thermal management systems.
BNP observed that sustained innovation in ground communication technologies, thermal control systems, and solar power generation could eventually narrow the operational cost differential between orbital and terrestrial data center facilities.
Get 3 Free Stock Ebooks
Discover top-performing stocks in AI, Crypto, and Technology with expert analysis.
- Top 10 AI Stocks - Leading AI companies
- Top 10 Crypto Stocks - Blockchain leaders
- Top 10 Tech Stocks - Tech giants
