Source Review: CSIS MacDonald 2026 — Economics of a Permanent Moonbase

Overall verdict: consistent Two-sentence summary: CSIS-credentialed historical-precedent analysis anchoring Apollo at $250-300B in today's dollars and SEI 1989 at ~$1T in today's dollars. The most authoritative pessimistic-ceiling reference for the cost of permanent lunar surface presence under non-commercial-architecture assumptions.

Key claims

• "Apollo total cost: $250-$300 billion in today's terms" — Consistent. Robust historical anchor (multiple cross-references in space-policy literature). For our calc this is the "this is what we know it cost to do crewed lunar landings under Cold-War-era contracting" ceiling.
• "Apollo 17 marginal cost: $580-$670 million per astronaut day" — Consistent. Useful as cost-per-utility-time benchmark.
• "SEI 1989 projected $500 billion in 1989 dollars (~$1T today)" — Consistent (with reframe). The SEI projection is the canonical "what a permanent lunar presence would cost with 1989-architecture assumptions" anchor. As I argued in pass-03 reconcile (and Codex amended), the gap between SEI's $1T and our BAU $150-400B is explained by institutional-cost-overhead (Apollo-style contracting, schedule slip, single-vendor markup) more than by launch-cost delta alone.
• "There have yet to be significant nongovernmental revenues for lunar activities" — Consistent. Pessimistic-commercial framing that balances Sowers / Handmer optimism.
• "A mixed ownership model — government control of core systems with private ownership of supplementary infrastructure — may be necessary for sustainability" — Consistent. Aligns with the LunA-10 architectural framing.

Cross-reference

• The SEI $1T anchor is one of the most-cited pessimistic ceilings in space-policy discourse.
• Codex correctly amended my earlier framing of "launch-cost delta explains the gap" — institutional-cost-overhead is the dominant factor.
• Codex anti-hallucination check: all quoted figures verified against extract.md.

Abstract

A CSIS economic analysis of permanent lunar-base economics, anchored on the historical record of Apollo and Constellation/SEI. MacDonald gives concrete totals: Apollo cost ~$250-300 billion in today's dollars; Apollo 17 marginal cost ~$580-670 million per astronaut day; Space Exploration Initiative (SEI, 1989) projected ~$500 billion in 1989 dollars (>$1 trillion today). The argument: the historical pattern is that crewed lunar surface activity is taxpayer-dominated, with no significant nongovernmental revenues yet, and this is unlikely to change soon. A mixed ownership model — government control of core systems with private ownership of supplementary infrastructure — is advanced as the likely sustainability pattern given current revenue limitations.

Key claims

• apollo-250-300b-today: "Apollo total cost: $250-$300 billion in today's terms"
• apollo-17-marginal-580-670m-per-day: "Apollo 17 marginal cost: $580-$670 million per astronaut day"
• sei-1989-500b: "Space Exploration Initiative (1989) projected $500 billion in 1989, which is over $1 trillion in today's dollars"
• no-significant-private-revenues: "There have yet to be significant nongovernmental revenues for lunar activities"
• mixed-ownership-likely: "A mixed ownership model — government control of core systems with private ownership of supplementary infrastructure — may be necessary for sustainability"

Reviewer notes

CSIS-credentialed, MacDonald is a credentialed space economist. Load-bearing for q5: gives the historical Apollo/SEI cost anchors needed to scope "what does a permanent moonbase actually cost in historical precedent dollars." The Apollo and SEI numbers are robust historical anchors (multiple cross-references in the space-policy literature). Notably: MacDonald is pessimistic about commercial-revenue trajectories — this is the policy-economist counterweight to the optimistic-commercial branch (Sowers, Casey Handmer). The mixed-ownership prediction matches institutional precedent. The $1 trillion (today's dollars) SEI projection from 1989 is a useful upper-bound anchor for "if you tried to do this with a 1989-style architecture, this is the bill." Useful for synthesis pass; not the right anchor for the optimistic-architecture case.