Source Review: Sowers 2021 — Business Case for Lunar Ice Mining

Summary

Claim 1: "Business returns are positive for all 3 scenarios" (commercial stand-alone + 2 PPP)

Quote: Abstract. Verdict: Different conclusion (vs Jones 2020) Why: Sowers' positive-returns conclusion directly contradicts Jones 2020's 35-year-pessimism conclusion. The reconciliation is architectural: Sowers uses tent sublimation (φ ≈ 534) + commercial launch + fully robotic, while Jones uses strip mining + SLS-class launch + assumes lander-side costs not included. Both can be correct within their assumption sets.

Claim 2: "$4 billion initial investment" framed as "the cost of a luxury hotel in Las Vegas"

Quote: Sowers (interview, via figure-sowers extract). Verdict: Novel supporting Why: The most-cited optimistic commercial-architecture capex anchor in the field. Direct support for q5.c13. The "luxury hotel" framing is rhetorically effective but quantitatively the figure is 1-2 orders of magnitude below the SEI/Apollo-precedent ceiling, reflecting the architectural compression Sowers achieves.

Claim 3: "$2.5 billion for development, fabrication and delivery to the lunar surface" (capture tent ISRU)

Quote: Sowers (interview). Verdict: Novel supporting Why: Anchors the propellant-only ISRU portion of the full capex at $2.5B — useful sub-component reference. If true and applicable, this is below our pass-02 calc's ISRU+launch capex for the same milestone-M5.

Claim 4: "$550M of revenue per year generating a 9% return-on-investment and $2B in profit over the life of the project of 15 years"

Quote: Sowers 2021. Verdict: Consistent Why: The revenue and ROI numbers are internally coherent. For q5, these establish that if the capex anchor is right, the commercial case closes. Our calc focuses on capex; Sowers focuses on revenue-side closure. Complementary.

Claim 5: "On the Moon surface itself, the cost of propellant would be $500 per kilogram"

Quote: Sowers (interview). Verdict: Consistent Why: A $500/kg lunar-surface propellant price (vs $36,000/kg Earth-launched-to-lunar-surface) is consistent with the φ ≈ 534 tent-sublimation architecture. Directly supports the gear-ratio-pays-off argument in q4.

Claim 6: "10-year timeline for cislunar infrastructure realization"

Quote: Sowers (interview). Verdict: Merits investigation Why: Sowers' 10-year buildup timeline aligns with neither our BAU 25-year M1-to-M8 nor our IE 5-year. It sits in between, suggesting Sowers assumes an intermediate "moderately aggressive automation + commercial launch" regime that we don't model explicitly. Merits a follow-up to characterize this intermediate regime explicitly.

Cross-reference

• Full PDF body paywalled; abstract is verbatim from publisher page; key dollar figures retrieved from secondary coverage (21st Century Technology interview, SpaceQ Media coverage).
• The architectural difference from our calc (tent sublimation, robotic-only, propellant-only) is the basis for q5.c13's "consistent in direction but not directly comparable via simple capex scaling."
• Codex anti-hallucination check: all quoted text appears in either the publisher abstract or the figure-sowers extract.

Abstract

Verbatim: "The key to human expansion into space and space development, in general, is developing space activities that deliver value in an economic sense. In other words, the key to space development is making money in space, that is, profit. However, the search for money making space activities has proven elusive. We present a business case for a commercial company to mine lunar ice and process the ice into rocket propellant. We discuss the existing and future markets for propellant and an architecture for mining and processing propellant and the associated costs. We then examine 3 scenarios, 1 commercial stand-alone and 2 involving a public/private partnership (PPP) model with NASA. We provide a comparison with other similar analyses. Business returns are positive for all 3 scenarios, although the PPP models provide increased returns and share risk with the government. Once established, lunar-sourced propellant will dramatically reduce the cost of all beyond low Earth orbit space activities and potentially enable other profitable commercial ventures to emerge."

Key claims

• positive-returns-all-scenarios: "Business returns are positive for all 3 scenarios" (commercial stand-alone + 2 PPP variants)
• ppp-superior: "The PPP models provide increased returns and share risk with the government"
• price-deltas-by-orbit: Propellant cost from Moon surface: $500/kg vs $4,000/kg current LEO, $16,000/kg GEO, $36,000/kg lunar surface delivered from Earth (interview-derived figures, see figure-sowers/extract.md)
• 4b-initial-investment: $4 billion initial investment to complete the facility (Sowers' framing: "the cost of a luxury hotel in Las Vegas")
• 2.5b-capture-tent-cost: $2.5 billion for development, fabrication and delivery to lunar surface of the capture-tent ISRU system
• 550m-revenue-yr: $550M revenue per year; 9% ROI; $2B profit over a 15-year project life
• enabling-effect: "Once established, lunar-sourced propellant will dramatically reduce the cost of all beyond low Earth orbit space activities"

Reviewer notes

This is the canonical commercially-framed business case for lunar ice mining, distinct from the NASA-Langley breakeven analysis (jones-2020-breakeven) that gives a different conclusion. Load-bearing for q5: gives a concrete capex anchor ($4B initial, $2.5B for tent ISRU plant) which is the lowest-quartile of the published TEA spread and is at odds with NASA-OIG-scale Artemis costs (~$93B over a decade). The architectural choice — tent sublimation, fully robotic, commercial-launch — is consistent with the optimistic-φ branch (φ ≈ 534 per Metzger 2023's restatement of Sowers). Full paper text is paywalled; abstract is verbatim from publisher page. Key dollar figures retrieved from 21st Century Technology interview (figure-sowers/extract.md) and PwC / SpaceQ secondary coverage. Notably absent in the abstract: timeline-to-first-product, autonomous-operation lifetime, replacement schedule.