Beyond Asteroid ISRU: Alternative Material Sources for the Dyson Swarm

Project Dyson's feedstock strategy has been asteroid-centric since inception. The consensus documents for Phase 1 assume C-type and S-type near-Earth asteroid ISRU as the primary material source, with the Assembly Node Hub designed around processing semi-refined asteroid ore. This made sense as a starting point: asteroids are nearby, their compositions are increasingly well-characterized, and the delta-v requirements for material return are favorable (0.5-4 km/s).

But asteroids are not the only option, and they may not be the best option at every scale. We reviewed 15 arxiv papers spanning Mercury surface factories, lunar regolith processing, asteroid mining economics, and civilizational-scale material strategies. The result: five new research questions, three research folders with detailed literature analysis, and the identification of a critical gap—no one has published a normalized cross-source comparison of feedstock economics.

Mercury: The Richest but Hardest Source

Mercury is the elephant in the room. At 3.3x10^23 kg, it outmasses the entire asteroid belt by two orders of magnitude. It sits at 0.39 AU where solar flux is 6.7x what we get at Earth orbit—abundant energy for processing. Its surface gravity (0.38g) is low enough for electromagnetic launch. And Shubov (2021, arxiv:2110.15198) has proposed a self-replicating factory concept that could, in principle, convert a significant fraction of the planet into Dyson swarm components through ~50 doublings.

MESSENGER mission data (Nittler & Weider 2019, arxiv:1712.02187) provides ground-truth composition: Mercury's surface is silicon-rich with an unexpectedly high sulfur content (1-4 wt%) and a high Mg/Si ratio. The iron core constitutes ~85% of the planet's radius (Margot et al. 2018, arxiv:1806.02024), representing ~2.3x10^23 kg of structural metal—if you can reach it.

The catch: Mercury is hard. Surface temperatures hit 700 K on the dayside. The delta-v to transfer materials to 1 AU is ~9 km/s. And the self-replicating factory technology that makes Mercury compelling is the same technology we haven't demonstrated at any scale. Our own rq-3a-2 simulation showed that degradation rate per generation is the critical risk for self-replication—5%/generation drops success probability to 69%. Over 50 generations on Mercury, even 1%/generation degradation becomes severe.

Assessment: Mercury is almost certainly the ultimate material source for a complete Dyson swarm, but it's a Phase 3+ capability. We've created rq-1-49 to track this pathway and connected it to our self-replication closure findings.

The Moon: Close, Accessible, Well-Characterized

If Mercury is the long game, the Moon is the near game. It's 3 light-seconds away, has been sampled by six Apollo missions, and has a growing international infrastructure roadmap (Artemis, Chang'e, Luna). Four papers advance the case for lunar feedstock.

Molten Regolith Electrolysis (MRE) emerges as the most promising processing route (Shaw et al. 2021, arxiv:2109.02201). MRE processes whole, unbeneficiated regolith—you shovel it in, apply electric current at ~1,600 C, and get a mixed metal alloy (Fe-Si-Ti-Al) at the cathode and oxygen at the anode. No mineral separation required. This is a major advantage over asteroid ISRU, where you might arrive at a target only to discover its composition doesn't match remote sensing predictions.

De Guzman et al. (2026, arxiv:2601.14719) demonstrate oxygen extraction from lunar simulants at 70-90% efficiency. Oxygen is the mass-dominant product (43-45 wt% of regolith), valuable as propellant for orbital transfer vehicles and as a chemical processing reagent.

For construction, Gupta et al. (2023, arxiv:2308.14331) show lunar regolith can be sintered into structural bricks at 20-40 MPa compressive strength, and McCallum et al. (2025, arxiv:2506.06392) demonstrate 3D printing of structural blocks. Together, these enable building surface infrastructure—processing plants, mass driver foundations, habitats—from local material.

The critical question is silicon purity. MRE produces ~2N-3N silicon; our rq-0-15 analysis found UMG-Si (4N-5N) may be viable for space PV. Bridging that 1-2 order of magnitude purity gap through vacuum directional solidification is plausible but undemonstrated.

For material export, electromagnetic launch from the lunar surface is compelling: 2.4 km/s escape velocity, no atmosphere, stable foundations for mass driver tracks. Two papers (arxiv:2106.11443, arxiv:2410.09616) advance coilgun and catapult designs for this application.

Assessment: The Moon is the most actionable alternative source. We've created rq-1-50 for processing and rq-1-51 for export, both connected to the existing feedstock question rq-1-21.

Asteroid Mining Economics: The Baseline We're Building On

Five papers address asteroid mining economics, and the picture is sobering. Calla et al. (2018, arxiv:1808.05099) find a baseline cost of ~$1,136/kg for asteroid-returned water. At that price, only in-space use cases close economically—no one is shipping asteroid water to Earth.

But Hein et al. (2018, arxiv:1810.03836) provide the crucial insight: throughput rate is the dominant cost driver, more important than target selection or extraction efficiency. Doubling throughput reduces $/kg by ~40% through fixed-cost amortization. Equipment reuse is second. This means the cost curves at Project Dyson's industrial scale (thousands to millions of tonnes per year) could look radically different from single-mission economics.

Graps et al. (2016, arxiv:1612.00709) catalog the resource landscape, and Lewicki et al. (2021, arxiv:2103.02435) address the bootstrap problem—how to make early missions viable enough to fund later capability. Zhang (2022, arxiv:2208.12617) takes the civilizational view, arguing that no single source sustains exponential material growth indefinitely.

The total asteroid belt mass (~3x10^21 kg) is only 1% of Mercury and <5% of the Moon. This is a non-trivial constraint: at full Dyson swarm scale, asteroids alone may be literally insufficient.

Assessment: Asteroid ISRU remains the right Phase 1 baseline, but the economics literature reveals both opportunities (cost drops dramatically at scale) and limits (total mass availability). We've added key references to rq-1-21 and created rq-2-32 for the cross-source comparison.

The Missing Piece: Gas Giant Mining

We looked for arxiv papers on gas giant atmospheric mining for construction materials and found... nothing. The literature on Jupiter and Saturn atmospheres is extensive for planetary science, and rq-3b-3 addresses He-3 extraction for stellar engine fuel, but bulk material extraction from gas giant atmospheres for megastructure construction is an unexplored topic.

This is itself a finding. Gas giant atmospheres contain effectively unlimited hydrogen, helium, and simple volatiles. At the Kardashev Type II scale that a complete Dyson swarm represents, these resources may become necessary for elements scarce in rocky-body mining. But the engineering challenges are extreme: 30-60 km/s escape velocities, crushing pressures, extreme radiation (Jupiter), and enormous distances.

Assessment: We've created rq-3a-5 as a gap-identification question with no arxiv references, acknowledging this as unexplored territory that needs attention in Phase 3+ planning.

Comparative Assessment

Source	Mass Available	Delta-V to 1 AU	Solar Flux	Key Elements	Earliest Phase	Technology Readiness
NEA (C-type)	10^18-10^20 kg	0.5-4 km/s	Variable	H2O, C, Fe, Ni	Phase 1	TRL 3-4
NEA (S-type)	10^18-10^20 kg	0.5-4 km/s	Variable	Fe, Ni, Si, Mg	Phase 1	TRL 3-4
Moon (regolith)	~10^17 kg (top 5m)	2.4 km/s + transfer	1x (day)	Si, Al, Fe, Ti, O2	Phase 1-2	TRL 3-5
Mercury	3.3x10^23 kg	~9 km/s	6.7x	Si, Fe, Mg, S	Phase 2-3	TRL 1-2
Gas giants	10^27+ kg (atm)	30-60 km/s	0.004-0.04x	H, He, CH4, NH3	Phase 3+	TRL 1

The most striking feature of this table is the gap between "accessible now" (asteroids, Moon at 10^17-10^20 kg) and "accessible eventually" (Mercury, gas giants at 10^23-10^27 kg). There's a natural transition: start with what's close and easy, scale into what's massive and hard.

Five New Research Questions

1. rq-1-49: Mercury self-replicating factory pathway (medium priority) — Can self-replicating factories on Mercury surface produce swarm components? 3 papers archived, research folder created.

2. rq-1-50: Lunar regolith processing for swarm materials (high priority) — Can MRE and related techniques produce solar-grade materials from lunar regolith? 4 papers archived, research folder created.

3. rq-1-51: Lunar electromagnetic launch for material export (medium priority) — What are the economics and engineering requirements for lunar mass driver export? 2 papers referenced.

4. rq-2-32: Comparative feedstock economics (high priority) — How do all candidate sources compare on $/kg-delivered under consistent assumptions? 5 papers archived, research folder created. This is the question that matters most for Phase 2 planning.

5. rq-3a-5: Gas giant atmospheric mining feasibility (low priority) — Is bulk material extraction from gas giant atmospheres feasible? No papers found—identified gap in the literature.

What This Changes for the Project

Three concrete implications:

Phase 1 planning should include a lunar feedstock trade study. The Moon's proximity, well-characterized composition, and near-term Artemis infrastructure make it a viable bootstrap source. We should not default to asteroid-only without comparing.

Phase 2 needs a comparative economic model. The absence of normalized cross-source $/kg data is the single biggest gap we identified. Until rq-2-32 is addressed, feedstock source selection for Phase 2's 100,000+ unit production run is based on assumption, not analysis.

Mercury should be elevated from "deferred to Phase 3+" to "tracked as enabling technology." The self-replicating factory concept connects directly to our Phase 3a foundry work (rq-3a-2). If closure ratios and degradation rates prove favorable in that context, Mercury GSFR feasibility follows. The connection should be explicit in project planning.

We've also updated three existing questions with new references: rq-1-21 (feedstock acquisition) now links to the Mercury and asteroid economics papers, rq-3a-2 (self-replication closure) now links to the Mercury GSFR concept, and rq-2-20 (swarm ROI) now includes the Kardashev energy economics perspective.