Fission Specific Power (6.5–6.7 W/kg)

Statement. The model sizes fission power at fFis = 0.15 t/kWe (6.67 W/kg). The 2025 NASA Fission Surface Power directive requires at least 100 kWe within a 15 t landed allocation, and the only design point sized to component level (Colozza 2020) reaches 6.47 W/kg for a 10-kWe system. The largest reactor ground-tested to date (KRUSTY) validated the physics at about 1 kWe electric only.

Evidence.

Requirement: NASA Fission Surface Power Directive, signed by Acting Administrator Sean Duffy, 2025-08-04. p2: "Minimum 100kWe power output / Assumed use of a heavy class lander (up to 15 metric tons mass available) / Readiness to launch by the first quarter of FY30 / A closed Brayton cycle power conversion system." p1: "surface power needs are at least 100kWe for long-term human operations including in-situ resource utilization." Arithmetic: 100,000 W / 15,000 kg = 6.67 W/kg required. This is a requirement on a not-yet-built system, not measured hardware.

Design: Colozza, A., 2020, Small Lunar Base Camp and ISRU Oxygen Production Facility Power System Comparison, NASA/CR-2020-220368, Table 22 (PDF p57, report p52): 10-kWe baseline Reactor 235 kg + Shield 547 kg + Balance-of-plant 763 kg = 1,545 kg total, giving 6.47 W/kg. The 1-kWe unit = 406 kg total gives 2.46 W/kg (shield 148 kg and balance-of-plant 122 kg do not scale down). Text (report p53): "the current electric Kilopower reactor design can be easily scaled from the test, validated a 1-kW design to a 10-kW electric system." A sizing-model design value, not flown.

Demonstrated envelope: Poston, D.I., Gibson, M.A., Godfroy, T., McClure, P.R., 2020, KRUSTY Reactor Design, Nuclear Technology 206:sup1, S13-S30, DOI 10.1080/00295450.2020.1725382. p2: "Kilopower systems are intended to provide between 1 and 10 kW(electric)"; p5: "The 5-kW(thermal) thermal capability of the KRUSTY reactor satisfied this goal. The electrical capacity/efficiency goal was partially met via two Stirling converters." This paper reports the ground test envelope (about 1 kWe class electric); it contains no system W/kg figure, so the 6.5–6.7 W/kg is sourced from the directive geometry and Colozza, never from KRUSTY.

Argument.

A1: Directive requirement 6.67 W/kg, and Colozza's 10-kWe design 6.47 W/kg agree to within about 3%.

A2: The directive itself notes prior work spanned "1kWe to 200kWe with no significant advancement towards flight system readiness," with the most recent effort, in 2023, focusing on a 40kWe concept. KRUSTY proved the reactor physics at about 1 kWe.

A3: Colozza Table 22 shows specific power rising from 2.46 W/kg at 1 kWe to 6.47 W/kg at 10 kWe because shield and balance-of-plant are a large near-fixed base. A single 6.67 W/kg constant is optimistic at pilot scale and realistic only once the reactor is large. This feeds the mass-model claim below.

---

0.05 t/kWe Solar

Statement. The model sizes continuous solar at fSol = 0.05 t/kWe (20 W/kg); we keep that value here. It prices generation and on-site storage only, and excludes transmission to a remote load (Argument A5). The value rests on three things: a real NASA array design at 120 kg for 10 kW, a Griffin-class CLPS lander payload budget of about 500 to 625 kg delivered to the surface, and a polar ridge site where the eclipse itself runs well under a full lunar night. 500 kg for a 10 kWe plant is about 120 to 154 kg of array plus about 350 to 380 kg for storage, power management, structure, and deployment.

That last 350 kg is where the argument lives, and it is a bet on technology NASA is now paying to develop, through the Watts on the Moon Challenge, whose scoring metric was mass adjusted for efficiency. Colozza prices the same switch in a NASA contractor report: holding the same oxygen plant at full power through the night costs 60,563 kg on batteries, and letting it hibernate on a 500 W keep-alive costs 2,740 kg, a factor of 22.1 at the same site with the same hardware model. Sizing our plant to carry full continuous load through the eclipse instead gives 0.530 t/kWe at Connecting Ridge and 0.983 t/kWe at the best single fixed site Speyerer and Robinson identify, which is 10.6 to 19.7 times heavier than what we publish, and Argument A2 states that case.

Evidence.

Evidence 1 (Colozza 2020 — system mass tables). Colozza, A., 2020, Small Lunar Base Camp and ISRU Oxygen Production Facility Power System Comparison, https://ntrs.nasa.gov/citations/20200001622.

Day and night are "approximately 354 h" each, and the system must "survive the nighttime, also approximately 354 h" (p8, report p. 2); the 708.33 h cycle length those halves come from is Table 2, p. 4. Battery specific energy 200 Wh/kg at a depth of discharge of 0.80 (p7, p48, p68; pp. 42-43), a constant regardless of operating time (p. 62). Usable specific energy is 200 Wh/kg × 0.80 = 160 Wh/kg. Table 14, p. 34 prints battery charge efficiency as a flat 99 and discharge efficiency as "Up to 99 (rate dependent)."

The complete photovoltaic mass boundary is stated as a seven-term sum on p. 46: array blanket, array structure, battery charge controller, wire, junction and breaker box, battery, and thermal control.

• Full continuous load through the night, battery, Figure 49 (report p. 61): battery 58,270.1 kg, thermal control 1,788.0 kg, array blanket 331.5 kg, structure 114.1 kg, charge controller 29.4 kg, junction and breaker box 16.5 kg, wire 13.6 kg, total 60,563.2 kg — rows close on the printed total exactly.
• Full continuous load, regenerative fuel cell, Figure 51 (report p. 62): 11,789 kg.
• Daytime-only production with a 500 W nighttime keep-alive, battery, Figure 53 (report p. 64): battery 1,041 kg, oxygen plant mass increase 865 kg, array blanket 520 kg, structure 179 kg, thermal control 66 kg, wire 47 kg, junction and breaker box 21 kg, charge controller 1 kg, total 2,740 kg.

Storage share falls from 96.21% to 38.00% on batteries and from 75.18% to 15.30% on the regenerative fuel cell when ISRU production is idled during eclipse phases. Totals fall by factors of 22.1 and 5.8.

Evidence 2 (Pappa 2021 — RSA design). Pappa, R.S. et al., 2021, Relocatable 10 kW Solar Array for Lunar South Pole Missions, https://ntrs.nasa.gov/citations/20210011743.

P. 15: "The current example RSA has a total mass of 120 kg." Figure 4, p. 20 breaks that down as 50 kg upper assembly, 36 kg composite telescoping mast, and 34 kg tripod, summing exactly. Against a 10 kW array output the quantum is 83 W/kg, or 0.012 t/kWe. The included items are blanket, cross arms, winch-and-cable deployment, mast, and tripod. The paper labels its own Figure 4 an "Order-of-Magnitude Summary of Masses, Resulting Loads, and Packing Volume" and states on p. 6 that its design parameters "are intentionally optimistic to provide understanding of the lower limit" to array mass and packaging volume. The same page imposes a requirement that the array blanket "must be elevated 10 m above the surface" and calls that elevation a major design driver; the mast itself is 16 m, the extra six metres being the hanging blanket, so the 10 m elevation and the 16 m mast are two different quantities and both are correct.

Evidence 3 (Gläser et al. 2014 — illumination). Gläser, P. et al., 2014, Illumination conditions at the lunar south pole, Icarus 243:78-90, DOI 10.1016/j.icarus.2014.08.013.

The authors define three separate quantities on p. 6, journal page 83, and they are not interchangeable. Accumulated illumination "is the average of all visible fractions of the solar disk over the considered time period"; accumulated percentage in light is the summed time with any part of the disk visible; longest continuous time in darkness is a third value. From pp. 7-8, journal pp. 84-85:

Height	Spot	Accumulated Illumination	Accumulated Time in Light	Longest Darkness
2 m	1	0.8812	0.9200	4.58 days
2 m	2	0.8793	0.9227	4.62 days
10 m	3	0.9255	0.9561	3.08 days
10 m	4	0.9194	0.9565	3.17 days

Table 1, p. 10 (journal page 87), prints the illumination column independently as 0.8812, 0.8793, 0.9255, and 0.9194. The abstract's headline figures, "receiving sunlight for 92.27% of the time at 2 m above ground and 95.65% of the time at 10 m above ground," are the accumulated time in light at spots 2 and 4. The energy-weighted duty at those same two spots is 0.8793 and 0.9194. The abstract also states that at these locations "the longest continuous periods in darkness are typically only 3-5 days."

Evidence 4 (Speyerer & Robinson 2013 — persistently illuminated regions). Speyerer, E.J. & Robinson, M.S., 2013, Persistently illuminated regions at the lunar poles, Icarus 222:122-136, DOI 10.1016/j.icarus.2012.10.010.

P. 1: "regions where the lunar surface remains illuminated for nearly 94% of the year with the longest eclipsed period lasting only 43 h ... small illuminated peaks (tens of meters across)." The 43 h is not a single-site eclipse. On p. 10, journal page 131, the paper gives it as the longest period during which all three of Stations 1, 2, and 3 were simultaneously in shadow. Those stations are spread over about 4 km of the Shackleton crater rim, and the roughly 0.94 annual illumination requires all three plus about 100 m of local mobility: 0.921 without mobility, 0.938 with it, and 0.937 in the conclusions. Taken in pairs the longest simultaneous eclipse is 60 h and 63 h. The single best 0.01 km² site, at 89.740°S / 201.2°E, receives 0.716 illumination and faces a longest eclipse of 145 h. So the 43 h describes a three-node relocatable architecture with mobility; the 145 h describes one array standing in one place.

Evidence 5 (CLPS/VSAT payload basis — open). CLPS landed payload, "Griffin class": about 500 to 625 kg delivered to the lunar surface, and a VSAT-class deployable array targeting about 154 kg. These two figures are the arithmetic basis of fSol = 0.05 t/kWe. Resolved by the CLPS task order payload statement or the lander vendor payload user's guide for the delivery mass, and by a NASA VSAT project mass statement for the 154 kg. Quinn notes this one needs work.

Pappa's own Appendix F prices a power extension harness within the 120 kg envelope above, but that harness runs about 8 m — from the elevated blanket down to the tripod base — and is stated to stow inside the cross-arm channel "without a separate volume penalty" (p. 12). It is a local, mast-height harness with no separate mass line, not a substitute for the km-scale base-to-load transmission cabling in Evidence 8 and 9, which the array design does not address at all.

A second, unresolved figure entered the record by way of a 2026-08 outside communication (Chris Tolton, CEO, Orbital Mining Corporation): that an industry VSAT provider has already built and tested a 10 kWe VSAT system at roughly 400 kg, more than double the CLPS-budget figure this claim uses. Astrobotic's public material describes a "state-of-the-art 10 kW VSAT system" under development as part of its LunaGrid service, and Lockheed Martin's public material separately describes VSAT hardware reaching 19.8 m; neither carries a system mass figure. Neither of the two full VSAT technical reports held in this corpus is the built industry system Tolton describes: Pappa 2021 is a preliminary NASA Langley design study, not built hardware, and Belbin 2024 is the NASA-built Government Reference Design demonstrator, which by its own text reports no mass figures of any kind, at any level of assembly, because it exists to test deployment mechanics with mock, non-functional arrays.

The 400 kg figure is unconfirmed — a secondhand recollection, not a citation — and is excluded from the mass model until traced to a primary document. Candidate sources: Astrobotic's own VSAT/LunaGrid technical filings (NASA contract deliverables, SBIR/STTR award descriptions) or Lockheed Martin's VSAT program documentation. Honeybee Robotics' LUNARSABER tower is a further VSAT-adjacent candidate, though the one source in this corpus that names it is a program-update briefing with no mass figure, not a technical report.

Evidence 6 (NASA Watts on the Moon Challenge). Managed by NASA's Glenn Research Center under the Centennial Challenges program at Marshall Space Flight Center, ran 2020–2024. The challenge "seeks solutions for energy distribution, management, and/or storage that address NASA technology gaps and can be further developed for space flight and future operation on the lunar surface," and frames the problem this claim turns on: "Solar energy is abundant on the surface of the Moon, but extended night hours (350 consecutive hours) and the extreme environmental temperature change from daylight to nighttime, create problems for solar power use."

Phase 2 required competitors to draw power from an intermittent NASA power source, deliver it continuously to a NASA load bank, operate in simulated lunar temperature and vacuum with no additional generation, and survive a 48-hour test representing six hours of solar daylight against 18 hours of darkness.

The 3 km separation between source and load was a capability requirement met by analysis only. Competitors were ranked by "Total Effective System Mass," a metric adjusted for efficiency and not mass alone; NASA's own release credits the winner with both the lowest mass and the highest efficiency. No per-team mass or Total Effective System Mass score is published, so no Wh/kg or t/kWe can be derived for the winning hardware.

Sources: https://www.nasa.gov/prizes-challenges-and-crowdsourcing/centennial-challenges/watts-on-the-moon-challenge/, https://www.nasa.gov/news-release/nasa-awards-1-5-million-at-watts-on-the-moon-challenge-finale/, and the Phase 2 Rules PDF, its Level 3 Technical Guidance, and the Figure 1 load profile, all linked from https://www.herox.com/WattsOnTheMoon.

Evidence 7 (rough cross-check). Not a derivation: Colozza's photovoltaic-plus-battery component masses (Evidence 1) applied at the eclipse durations from Evidence 3 (Gläser) and Evidence 4 (Speyerer and Robinson). No published formula connects duration to mass at a site.

Eclipse basis	h	t/kWe	W/kg
Three stations spread over 4 km plus 100 m mobility	43.0	0.310	3.23
Best pair of those stations	63.0	0.443	2.26
Glaser Connecting Ridge 10 m spot 3, one fixed site	73.9	0.515	1.94
Glaser Connecting Ridge 10 m spot 4, one fixed site	76.1	0.530	1.89
Glaser Connecting Ridge 2 m spot 2, one fixed site	110.9	0.760	1.32
Glaser abstract upper bound, one fixed site	120.0	0.819	1.22
Speyerer best single 0.01 km² site	145.0	0.983	1.02

Evidence 8 (Kerslake 2007 — surface power transfer). Kerslake, T.W., 2007, Lunar Surface-to-Surface Power Transfer, NASA/TM-2007-215041.

Sizes DC cable, AC cable, RF beamed, and laser beamed power transfer subsystems for 1 to 50 kWe over 0.1 to 10 km, a range the paper ties directly to Lunar Architecture Team outpost siting, where an ISRU plant sits away from the habitat core specifically to limit dust contamination (p. 1) — the same siting logic behind the 3 km VSAT-to-user separation in Evidence 6.

At the 10 kWe point design, 1000 VDC, 1 km, the DC cable power transfer subsystem alone totals 766.5 kg (76.7 kg/kWe): power cable 453.7 kg, step-down converter 108.0 kg, step-up converter 81.2 kg, control cable 50.0 kg, spool and deployer 50.0 kg, remainder in connectors, thermal control, and secondary structure (Fig. 8, p. 10). At 5 km the same 10 kWe design totals 4,567.0 kg (456.7 kg/kWe), power cable alone 4,016.0 kg; at 10 km it totals 8,223.5 kg (822.3 kg/kWe) (Fig. 10, p. 11). Kerslake does not tabulate a 3 km case directly.

Raising voltage lowers this substantially: at 1 km and 10 kWe, total subsystem mass falls to a minimum of 550.5 kg (55.1 kg/kWe) at 2000 VDC against 766.5 kg at 1000 VDC and 1,946.3 kg at 160 VDC — the low-voltage, array-native regime VSAT-class hardware in this corpus operates in (Fig. 12-13, p. 12). AC cable at 1000 VAC, 1 km, 10 kWe totals 1,001.1 kg (100.1 kg/kWe), heavier than the equivalent DC point design (p. 13). Beamed alternatives run heavier still at these ranges: RF at 2.45 GHz totals 2,449.2 kg and at 5.8 GHz totals 1,228.7 kg at 10 kWe and 1 km; laser totals 1,777 kg (293 K) or 1,677 kg (325 K), essentially flat with distance (Figs. 43-46, pp. 30-31). This is a spreadsheet sizing-tool study, not built or tested hardware.

Evidence 9 (Csank et al. 2022 — lunar microgrid). Csank, J., Thomas, G.L., Granger, M., & Gardner, B., 2022, Powering the Moon: From Artemis Technology Demonstrations to a Lunar Economy, NASA Document ID 20220004165.

Sizes a lunar microgrid connecting Habitat and ISRU sub-grids, siting the ISRU production site "about 3-5 km from power generation, often in or near a cold trap" — the same order of separation as the 3 km VSAT standoff in Evidence 6 and the km-scale ranges in Evidence 8. Across radial, ring, and mesh architectures sized up to 40 kWe (the Fission Surface Power capacity), the paper states that "cables dominate microgrid mass, ranging from about 70% to 95% of total," with cable mass scaling roughly linearly with power level and almost quadratically with transmission distance. A ring topology, single-line-fault tolerant, adds roughly 50% over radial; a mesh topology, dual-line-fault tolerant, roughly doubles mass over radial. The paper states its own figures are underestimates: switchgear, control and communication hardware, and cable-deployment equipment are excluded from the mass model.

Argument.

A1: The retired 0.4 W/kg is a battery-dominated 354 h full-night survival system: 354 h × 1 kW / 0.160 kWh/kg = 2,212 kg/kW = 0.452 W/kg, with Colozza's own full-carry result at 0.426 W/kg. The 2 to 7 W/kg illuminated-ridge band is an array-dominated system bridging a short eclipse, and it must not be averaged with the first. The nearest held solar specific powers are Pappa's 83 W/kg system and 300 W/kg blanket, Kornuta's 70 to 90 W/kg onboard arrays, and Turyshev's 30 to 165 W/kg, all array-boundary numbers an order of magnitude above the band, while the continuous-system derivation of Evidence 7 gives 1.02 to 1.94 W/kg, below it. Colozza itself is not polar — 30°N on a fixed array through a flat 354 h night; its 160 Wh/kg usable value, also used in A2 and A3, is licensed by 196 cycles over 15 years, a duty a polar site cycling every lunation may not meet, and no source we hold gives a cycle-life curve at that cadence.

A2: 200 Wh/kg × 0.80 = 160 Wh/kg usable; 354 h × 1 kW / 160 Wh/kg = 2,212 kg/kW = 0.45 W/kg, essentially the retired 2.5 t/kWe. Colozza's own full-night result: 60,563 kg of photovoltaic plus battery against a 4,853 kg reactor (three 10-kWe Kilopower units at 1.99 km standoff for the same 25.83 kW load, Colozza 2020 Figure 57/58, p. 67-68). The actual eclipse is shorter, since ridge geometry, not a calendar night, sets it: Gläser's Connecting Ridge site sees 73.9 to 110.9 h of continuous darkness, Speyerer and Robinson's best single site 145 h. No published study ties duration to mass at a site; Evidence 7 applies Colozza's component masses at these durations as a rough check, and still runs 0.515 to 0.983 t/kWe, above fSol. fSol itself comes from Evidence 5, not this check, and Evidence 5 remains unresolved. Error here runs toward more mass, not less.

A3: On the full system boundary, at 10 kWe continuous and 160 Wh/kg usable, the band inverts to 0.76 days at 7 W/kg and 2.98 days at 2 W/kg. On the narrower array-plus-battery boundary it inverts to 0.86 and 3.22 days, applying the 0.99 efficiency term consistently across both boundaries. A three-station 43 h architecture lands inside the band at about 3.2 W/kg while every fixed single site in the corpus, from Glaser's 3.08 days to Speyerer's 145 h, lands below it on the heavy side.

A4: 500 kg landed, of which 120 to 154 kg is a real NASA array design and the remaining 350 to 380 kg is allowance for energy storage, power management and distribution, structure, and deployment. Under this allocation solar is lighter than fission.

A5: Evidence 6 and Csank (Evidence 9) put ISRU siting at 3 to 5 km from the power source, the same dust-contamination logic Kerslake (Evidence 8) cites for the original Lunar Architecture Team outpost. Kerslake's DC-cable point design runs 766.5 kg at 1 km rising to 4,567.0 kg at 5 km for a 10 kWe load, before any voltage optimization; Csank's independent finding, that cables run 70 to 95% of microgrid mass at Fission Surface Power's 40 kWe scale, points the same direction at a different power level and topology. Added to the 500 kg CLPS allocation in Evidence 5, either result roughly doubles to sextuples total plant mass once transmission to a 3+ km load is included — before resolving whether the 120 to 154 kg array-only figure or the unconfirmed 400 kg industry figure is correct.

Tags. solar, power, power storage, nuclear, kilopower