Data Centers in Space

As we enter 2026, it seems like the world is changing faster than ever as seen through big geopoltical events and technological innovations. On the latter note, a topic that seems rather “sci-fi” could become reality: data centers in space. Yes, that means “cloud computing” could soon literally be in the clouds (well, above the clouds).

​

Data centers, those warehouses housing the hundreds of thousands of servers needed for getting online, have been recently gaining a lot of attention for their critical role in the AI boom, and in particular for the tremendous cost in bulding and running them, both financially and in terms of energy, due to the intense requirements of AI models. The energy burden has been so intense that consumers have been seeing higher energy bills due to the overall increase in energy demand. As such, data centers in space could help alleviate some of the energy burden by leveraging nearly continuous access to solar energy, and using the vast iciness of space for cooling. While data centers in space aren’t functional yet, Google, with their moonshot Project Suncatcher and startups like Starcloud are getting in the planning. If this pans out, initial estimates show astronmical growth (pun not intended – see chart below).

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

 

 

The appeal of space data centers comes down to a few structural advantages:

 

• Continuous clean power from uninterrupted sunlight

• Cooling without water, using space’s natural heat dissipation

• Freedom from land, zoning, and grid bottlenecks

• Proximity to space-generated data, like imaging and communications

 

If this theme plays out, the winners extend beyond “space companies.” Beneficiaries include launch providers, satellite manufacturers, radiation-tolerant semiconductor designers, optical networking firms, and cloud platforms that can orchestrate workloads across Earth and orbit. On the other side, traditional data center operators in power- and water-constrained regions may see long-term pricing power erode, while utilities and municipalities counting on endless data center growth may face plateuing demand.

 

From an investment perspective, even modest adoption of these new data centers could ultimately alter the long-term economics of compute, energy, and infrastructure. We view this as a call option on the future of AI infrastructure with attractive “picks and shovels” opportunities in power management, networking, and specialized hardware that have value whether compute ultimately stays on Earth, moves to orbit, or (most likely) does both. But whether this “moonshot” works will ultimately be driven by profitability, performance, and a very literal constraint: how much space is left in space as low-Earth orbit fills up with accelerating satellite launches and decades of accumulated debris.

​

​​​​​​​​​​

​

​

​​

- Pedro Ramos
 

​

​