Source Review: Per Aspera — Realities of Space-Based Compute (2025)

Verdict: Consistent Confidence: high

Per Aspera's 30-50 t/MW per-MW orbital DC mass figure is one of three independent engineering analyses (mine 40 t/MW, luminix 42 t/MW) converging within ±25%. The bottleneck ordering (thermal → eclipse → comms → radiation → structural) is consensus-aligned with academic analyses. The 2030 "crawl phase" characterization supports q6.c7 stall-regime framing. Per Aspera's "100 MW LEO cloud by 2028 is science fiction" framing is defensible and brackets the lower edge of TAI-C transition.

Abstract

Per Aspera's "realities" assessment of space-based compute provides the most quantitative single per-MW orbital data center mass breakdown available in the public literature. For a 100 kW system: ~700 kg solar panels + ~500 kg batteries + ~1,000 kg radiators + additional structure/cooling/electronics, totaling 3-5 metric tons. Extrapolated, this gives roughly 30-50 t/MW of orbital data center mass. The post identifies the principal bottlenecks in priority order: thermal rejection, power intermittency (30% eclipse duty), comms bandwidth, radiation tolerance, and structural/deployment. The 2030 horizon is characterized as the "crawl phase" — kilowatt-scale, niche applications only. The post emphasizes that "anyone waving a business plan for a hundred-megawatt LEO cloud by 2028 is still pushing science fiction."

Key claims

• mass-per-100kw: 100 kW system mass ~3-5 t (700 kg solar + 500 kg batteries + 1000 kg radiators + structure/electronics).
• mass-per-mw-extrapolated: ~30-50 t/MW orbital data center mass.
• bottlenecks-priority: 1) thermal rejection, 2) eclipse duty cycle,
  3. comms bandwidth, 4) radiation tolerance, 5) structural/deployment.
• radiator-area: ~100-200 m² radiator per 100 kW system.
• 2030-horizon-crawl: 2030 phase is kilowatt-scale niche applications; 100 MW LEO cloud by 2028 is "science fiction."
• launch-cost-prereq: $500/kg LEO is the prerequisite for the 2030s "walk" phase scaling.

Reviewer notes

Tier C: expert-blog analysis with first-principles engineering grounding. The 30-50 t/MW figure is the single most useful per-MW orbital DC mass estimate I located; it bounds the demand calc tightly. Bottleneck ordering aligns with consensus engineering literature. The "crawl/walk/ run" phasing through 2030/2035 is a useful organizing scheme for q6's demand decomposition. Tier C scalar review applies.