The AI1 Design Point: A Bounds-Based Analysis of SpaceX's Orbital Data-Center Satellite

Companion to the theory preprint "Thermodynamic Bounds and Mass-Trade Criteria for Heat Rejection in Orbital Data Centers" (doi:10.5281/zenodo.20650893), applying its radiator-area bounds to a reported industrial design point. On June 9–10, 2026, SpaceX announced AI1, an orbital data-center satellite reported at 150 kW peak and 120 kW sustained compute with up to 110 m² of deployable liquid radiators. We apply the radiator-area bounds of the contemporaneous, independently derived thermodynamic framework to the reported figures, using the reported compute powers as the radiative heat load. Treating the 110 m² as double-sided panel planform (220 m² emitting; SpaceX states the radiators radiate from both sides), the equilibrium radiator temperature is approximately 337 K at the sustained load and approximately 353 K under a continuous-peak hypothetical; a total-emitting-area reading instead requires 391 K to 412 K, which strongly disfavors a conventional subcritical loop using ammonia, the coolant secondary coverage describes as likely, though public information cannot rule the sustained case out. An illustrative combined stress case (emissivity 0.91 to 0.80; effective sink 220 to 260 K) removes about 40 kW of fixed-temperature capacity, consuming the full 30 kW peak-to-sustained headroom plus a further 10 kW, or equivalently raises the sustained-load equilibrium by about 22 K into unreported temperature and pressure limits. The reduced-order radiative model does not rule out the reported design point; allowable operating temperatures, coolant phase and pressure, internal gradients, the complete power budget, and the economics remain unreported and open. All radiator-model calculations and displayed derived values, excluding externally sourced thermophysical property data, are verified by the accompanying assertion suite. Every reported figure traces to quoted announcement coverage, with key statements drawn from direct Musk and Dahl quotations. The paper passed four rounds of adversarial peer review with independent Wolfram verification. Analysis and drafting by a multi-model AI workflow (Claude Fable 5; peer review and verification by GPT 5.5 with Wolfram; evidence base via Perplexity), directed by the author; see Acknowledgments.