NASA's 40 kWe lunar fission program, measured by what exists

What the work claims

The paper is a program status report, not a test-results paper. Its claims are institutional: that the FSP project, run out of NASA Glenn for the Science Technology Mission Directorate with the Department of Energy and three industry teams, has completed its Phase 1 concept-design studies and is readying a Phase 2 request for proposals that would lead to an engineering demonstration reactor flown to the Moon.¹ The target system is a high-assay low-enriched uranium (HALEU) fueled reactor delivering at least 40 kWe for 10 years with no maintenance, with a mass goal under 6,000 kg and a stowed envelope of 4 m by 6 m.¹ Every one of those numbers is a requirement or a goal levied on future contractors. None of it is measured flight hardware, and the paper never pretends otherwise: the mass figure is a goal, the power figure is a requirement, and the mission is a plan.

What the paper documents as actually accomplished is a set of paper studies and test campaigns: three industry design contracts completed in September 2023, a NASA-DOE government reference design put through three design analysis cycles between April and November 2023, and an ongoing set of technology maturation efforts in power conversion, heat rejection, PMAD, and reactor materials.¹

How it works

A fission surface power unit converts the heat of a sustained nuclear chain reaction into electricity through a thermodynamic power conversion loop, then rejects the remaining heat through radiators. The FSP design space, as described in the paper, spans two conversion families: Stirling engines and closed-cycle Brayton turbines. Both take reactor heat at the hot end and export electricity plus waste heat; the radiator, the shielding, and the power management and distribution (PMAD) subsystem are comparable in mass and complexity to the reactor itself for a lunar surface unit of this class.¹

The paper is most useful as a map of the technical risk the program itself has identified. The technology maturation list names: organics in the power conversion system and heat-pipe fluids under radiation; advanced-manufactured turbines and heater heads; radiator dust tolerance (lunar dust degrades radiator surfaces and optical properties); integrated heat-pipe Stirling tests; liquid-metal pumps; high-voltage power transmission in the PMAD; metallic hydride moderators, specifically yttrium hydride; plus instrumentation, control, and shielding work.¹ Read that list carefully: only two items (moderators and shielding) are nuclear. The rest are ordinary power-engineering problems, and the program is treating them as co-equal risks.

The acquisition structure is three-phase. Phase 1 awarded three $5M single-year concept studies in 2022 to Intuitive Machines with X-Energy, Lockheed Martin with BWXT, and Westinghouse with Aerojet Rocketdyne (now L3 Harris); all three completed in September 2023 against exactly three levied requirements: produce 40 kWe for 10 years, survive launch and vibration loads, and keep the radiation dose at or below 5 REM per year above background at 1 km from the reactor.¹ In parallel, Los Alamos, Idaho, and Oak Ridge national laboratories with Glenn built a government reference design: a transportable 40 kW system in three pallets deployed from a lander, iterated through three design analysis cycles (DAC1 April 2023, DAC2 August 2023, DAC3 November 2023).¹ Phase 1A continues risk-reduction work on the high-risk concepts the Phase 1 studies surfaced. Phase 2, the open competition to design, build, test, and deliver a flight unit, begins only when an RFP is released, and the paper states that timing is pending FY25 budgets.¹

The physics heritage is real but modest in scale. The paper recounts the chain: the 2012 DUFF test, which coupled heat pipes to a Stirling engine for the first time on a reactor and produced 24 watts of electricity in a six-month, roughly $1M experiment at the Nevada National Security Site; the 2018 KRUSTY demonstration, which the authors describe as a flight-like system operated at prototypic conditions in a relevant environment, rated Technology Readiness Level 5, the first US steady-state and transient space-reactor test since the 1960s; an earlier 2010 to 2015 technology demonstration unit that ran at full power in a Glenn vacuum chamber under representative lunar conditions; and the 2000s Prometheus project, a 200 kWe Brayton-based nuclear-electric-propulsion system with a 1300 K reactor that concluded in 2005 on cost and complexity grounds.¹

The steelman

The strongest honest case is that this program is doing the unglamorous parts right, in a field with a graveyard of over-promised reactors. The requirements discipline is genuinely notable: three contractual requirements, one of them a quantified safety boundary (5 REM per year at 1 km) that gives siting engineers something to design against, and none of them prescribing a reactor architecture. That minimalism let three very different industrial teams (a lunar-lander startup paired with a microreactor developer, a prime paired with a naval-reactor house, and a reactor vendor paired with a propulsion company) produce genuinely divergent concepts, and it means the government now holds a cross-section of the vendor base's ideas rather than a single point design.¹

Second, the maturation portfolio treats the correct bottleneck. Every credible study of space fission says the reactor core is not the hard part at 40 kWe; keeping conversion machinery alive for ten unattended years in vacuum, dust, and radiation is. A program that spends its risk-reduction dollars on radiator dust tolerance, organic fluid radiolysis, additively manufactured turbomachinery, and high-voltage PMAD is spending where the failures will actually happen.¹

Third, the authors make a claim worth taking seriously: of the five critical technology elements in NASA's nuclear-electric-propulsion technology maturation plan, all but the electric propulsion subsystem are being worked under the FSP portfolio.¹ If true, surface fission is quietly subsidizing the power plant for future high-power deep-space propulsion, doubling the return on every maturation dollar.

Where a skeptic should push

The single most load-bearing assumption is that Phase 2 happens at all. The paper says the RFP timing is pending FY25 budgets, which means the flight demonstration rests on a political decision, and the historical base rate is poor: Prometheus died of cost and complexity, the affordable-FSP TDU never flew, and KRUSTY, the program's flagship accomplishment, was a kilowatt-class ground test rated TRL 5, six years ago.¹ TRL honesty matters here: nothing in this paper is hardware at flight scale. The 40 kWe unit is a set of requirements; the three industry designs are studies; the government reference design is analysis cycles; and the 10-year no-maintenance life, the figure that makes the system worth having, is asserted, not demonstrated, because no space power system of any kind has run ten years unattended on a planetary surface.

The mass goal deserves the same treatment. Under 6,000 kg for 40 kWe is roughly 6.7 W per kilogram at the system level, before anyone has bent metal, and the paper gives no measured basis for it. The KRUSTY reactor that anchors the program's credibility produced about 1 kWe class power; closing the gap to 40 kWe multiplies every integration problem that scale brings.¹ Finally, the safety requirement itself, while admirably quantified, is a design constraint, not a verified outcome: 5 REM per year at 1 km will have to be demonstrated for the actual as-built shielding configuration, on a surface where astronauts work nearby.

What a 40 kWe lunar reactor asks of the power chain

The non-obvious implication hides in the maturation list: for lunar fission, the differentiating engineering is not nuclear. Radiator dust tolerance (Section 5 of the paper) is the item that decides whether a ten-year unattended unit survives, because a dust-degraded radiator forces the conversion loop to run hotter or the reactor to derate, and either shortens life. The genuinely new discipline the paper implies is high-voltage surface transmission: distributing 40 kWe across a growing base camp at low voltage means heavy harness mass and large conduction losses in regolith-rated cable runs, so the program is maturing high-voltage PMAD for the surface grid. Anyone waiting for the reactor to solve lunar power is looking at the wrong subsystem; the power electronics and thermal rejection are the pacing items.¹

The threat is schedule fragility with a long dependency chain. A surface reactor anchors habitat load growth, ISRU throughput, and night survival for other assets; every year the RFP slips, solar-plus-storage architectures get another year of incremental improvement and another tranche of flight heritage, and the crossover point where fission stops being competitive arrives earlier. Prometheus is the template for how these programs end.¹

The opportunity is the NEP spillover, stated concretely in the paper's Figure 1: four of the five NEP critical technology elements are being matured under FSP.¹ A kilowatt-class ground test plus a funded 40 kWe surface program gives nuclear-electric propulsion its power conversion, heat rejection, and PMAD for free, which is the cheapest route to megawatt-class deep-space power anyone has proposed with actual money behind it.

The bottom line

Calibrated judgment, bounded to what the paper shows: the FSP project has demonstrated disciplined programmatics and requirements engineering, plus a genuine TRL 5 reactor physics demonstration from 2018. It has not demonstrated anything at 40 kWe, and its own authors gate the flight phase on a budget decision. My read: high confidence that the pre-formulation work described is real and competently scoped; moderate confidence that a Phase 2 award happens on any particular schedule; low confidence in the 6,000 kg mass goal and the ten-year unattended life until a funded design exists to test them. What would confirm the claim: an RFP release, an industry preliminary design review with a mass statement, and a dust-tolerance test on a flight-representative radiator. What would break it: another budget hold, or a Phase 1A finding that the conversion loop cannot meet life requirements at lunar temperatures.