Data Centers: Ashburn, then Abilene, then space

研报英文原文证据摘录

Data Centers: Ashburn, then Abilene, then space

Barclays | Data Centers

While some of these leaders are known for their grandiose projections, we acknowledge they

have also overseen growth in their other businesses that seemed improbable, to say the least.

We think the commercial satellite industry serves as a pertinent framework for evaluating the

prospect of data centers in space. For the ~30 years until 2020, projections about spaced-based

communications generally failed to meet expectations (Iridium, Globalstar, and ICO Global are

among the many bankruptcies in the industry). However, since 2020, there has been a new era

of decreasing costs and scaling capabilities. So the question is; are spaced-based data centers

at the beginning of ~30 years of failed promises and unmet expectations? Or can space-based

data centers capitalize on the momentum of commercial satellites, and alter the competitive

environment? We think the latter is more likely, but estimate it will still take 10+ years for DCs in

orbit to become competitive with terrestrial deployments.

The primary benefit of orbital data centers is access to power (more specifically, a magnitude of

power that would generally not be achievable on earth). Solar panels provide near-continuous

energy without terrestrial constraints (power infrastructure, pollution concerns, land

availability, local regulations, etc.). Orbital data centers conceivably could handle all but the

most latency-sensitive workloads. Some space-based communications currently offer 25-60

milliseconds of latency (for context, the blink of an eye is ~300ms). However, there is still a long

way to go before the economics and engineering make orbital data centers feasible at scale.