Grid-connected sites, dedicated power supply, coastal facilities, stratospheric platforms, and low-Earth-orbit systems are assessed using a common service framework. Documented facts, industrial targets, exploratory assumptions, and calculated results are kept distinct.
Research manuscript / 01
Where Should AI Compute Be Built in 2029–2030?
Compares terrestrial, stratospheric, and orbital compute through power delivery, heat rejection, data transport, equipment replacement, and commissioning constraints. It asks which conditions make each architecture competitive for a particular workload and delivery horizon.
Scope
The question behind the study.
- Which sites can deliver usable power on the required schedule?
- When do transport, heat rejection, maintenance, and replacement change the preferred location?
- Which assumptions would make an orbital or stratospheric architecture competitive?
Reading the results
Assumptions stay visible.
This is a theoretical, prospective study based on public sources and first-order models. Hypothetical cost and deployment scenarios are conditional calculations. They are not measured operating performance, quoted commercial prices, or commissioning guarantees.