TLDR
- BNP Paribas states orbital data centers will only be economically viable once launch costs drop below $300/kg — current costs range from $1,500 to $3,600/kg
- Building a 1-GW space data center today would cost over $100 billion, compared to $35 billion to $50 billion for a ground-based equivalent
- BNP anticipates Google, Amazon, and Meta will be among the first to conduct early-stage tests of orbital data centers if costs decrease
- Elon Musk asserts that within 30 to 36 months, space will become the “most economically compelling” location for AI infrastructure
- Nvidia-backed Starcloud launched the first Nvidia H100 GPU into space in November 2025
Orbital data centers are transitioning from science fiction concepts to topics of discussion in corporate boardrooms. Investment firm BNP Paribas has released an analysis on the concept; while it recognizes long-term potential, it notes the economics are not yet feasible.
With current launch costs ranging from $1,500 to $3,600 per kilogram, constructing a 1-gigawatt orbital data center would exceed $100 billion. A similar on-ground facility, by contrast, costs between $35 billion and $50 billion to build.
BNP analyst Nick Jones stated the bank does not consider orbital data centers a “feasible short- to medium-term solution.” He pointed to high launch expenses, costly space-rated hardware, and the difficulties of cooling systems and electrical infrastructure in orbit as key barriers.
For the economics to be favorable, BNP indicates launch costs need to fall below $300 per kilogram. This would represent a significant decrease from current rates.
If costs reach that threshold, BNP expects Google, Amazon, and Meta to be among the first companies to conduct early alpha tests of orbital data centers. The bank did not provide a specific timeline for when this might occur.
Why Space? The Energy Problem on Earth
The drive toward space is partially fueled by the energy requirements of artificial intelligence. Terrestrial data centers are using more power than ever before. As of 2023, U.S. data centers accounted for approximately 4.4% of total U.S. electricity consumption, per the Department of Energy.
McKinsey projects that by 2030, global data centers will need $6.7 trillion in investment to meet demand. Capital expenditures from major tech companies are forecast to reach $600 billion in 2026, with Amazon alone planning to spend $200 billion.
Elon Musk has made space-based data centers a central component of SpaceX’s strategy. He has stated that within 30 to 36 months, space will be the “most economically appealing location” for AI infrastructure. SpaceX’s declared objective is to deploy a constellation of one million satellites functioning as orbital data centers, each producing approximately 100 kilowatts of compute power per ton.
Musk’s reasoning focuses not mainly on reducing energy costs but on energy availability. He noted that electricity production outside China has remained largely stagnant, prompting questions about where new terrestrial data centers will obtain their power supply.
SpaceX Is Already Hiring
SpaceX has moved beyond mere discussion of the plan. Michael Nicolls, vice president of Starlink Engineering, posted on X that the company is recruiting for “numerous critical engineering positions” related to space-based data centers, including a Space Lasers Engineer role based in Redmond, Washington.
The firm declared it intends to acquire Musk’s AI company xAI, citing space-based AI as a long-term priority.
Early Tests Are Already Happening
In November 2025, Nvidia-backed startup Starcloud launched the first Nvidia H100 GPU into space via a SpaceX Falcon 9 rocket. The Starcloud-1 satellite weighed approximately 60 kilograms — roughly equivalent in size to a small refrigerator.
Starcloud’s long-term objective is to develop a 5-gigawatt orbital data center spanning around 4 kilometers in width, equipped with large solar and cooling panels.
BNP observed that over time, advancements in ground-link technology, cooling systems, and solar power could reduce the operational costs of orbital data centers to levels comparable to those of terrestrial facilities.