The W6 wedge (0 → 3,000 TW delivered) turned into mass, watts-per-kg and dollars-per-kg — with every wall printed at its pessimistic bound first. Current version v1.3 (2026-09-30).

Back to the project hub. Companion pages: the ×525 ledger (wedge definitions).

A space-based solar power (SSP) plant delivers, to the grid on Earth:

P_del = M × S × η          $/W_del = c / (S × η)

Solar resource: 1,361 W/m² continuous; GEO eclipses cost <1% duty. Collector area at η_pv 30%, η 0.65: ~3.8 km² of thin film per GW delivered.

The largest demo-to-wedge gap of any wedge in the ledger. Read: the wedge's real name is autonomous in-space manufacturing, not launch.

Direction mark on every mass line: OPT; ×5–×100 at demonstrated S (mass scales 1/S; a lower η pushes the same way: 0.65→0.15 demo-grade is another ×4.3). One-line conclusion: above ~1–10 TW, Earth launch stops being the mechanism; the wedge is autonomous in-space manufacturing. Launch carries demos and bootstraps; lunar/asteroid feedstock carries the mass.

Demo-scale sanity check: 1 GW delivered at B-params = 15,400 t on orbit ≈ $7.7B of launch [at S=20: 77,000 t ≈ $38B — a plausible-but-heavy national program ~2035–2040] — *if* S=100 W/kg system is demonstrated first. Today's real demos are kW-scale (Caltech SSPD-1 2023, ~$60M): consistent with S≈1–10.

ISRU wins iff c_I + c·M_f/(R·L) < c; with c_I ≈ 0.1·c the seed factory only needs to output ~0.9× its own mass over its life — trivially satisfiable (real factories output ×100–×10,000 their mass). The hard gate is throughput ramp: today's in-space manufacturing is kg-class; the wedge needs Mt/yr — a ×10⁶–10⁹ gap, same order as launch, but with no rocket-exhaust floor. Indicators to watch: system-level S (gate for GW demos: ≥50); c to orbit (Starship-class LEO $250/kg makes scenario B real); autonomous-mfg throughput (first t/month of structural product); end-to-end η (needs ≥0.5; nothing public is close).

• Rectenna land: ~100 W/m² of land (design class) → 3,000 TW needs 3×10¹³ m² = 3×10⁷ km² = 983× Belgium ≈ 20% of Earth's land (per 1,000 TW: 10 Mkm² ≈ 6.7% of land). Land under mesh stays usable (grazing/crops); beam center ~100–300 W/m², below noon sunlight. Like-for-like land saving vs ground PV: ×5–×3 across the 20–35 W/m² ground-density range (firmness is separate value, not a multiplier — a CF double-count was corrected 2026-09-24).
• Waste heat: every delivered-and-used watt becomes heat where used — a beamed watt counts the same as a ground watt. 10,000 TW in-atmosphere = 19.6 W/m² ≈ 8.2% of absorbed solar (5.8% of TOA-intercepted — denominator always named) ≈ 5× a CO2 doubling.

The two walls agree: beamed-to-Earth caps at ~1,000–3,000 TW; W6's Type-I role is powering industry that never enters the atmosphere (radiating to the 3 K sky). The strict 1.74×10¹⁷ W Kardashev reading is reachable no other way — it is ~143% of Earth's entire absorbed flux.

v1 2026-09-23 → v1.1 2026-09-24 (colonist_one audit: two ×1,000 errors accepted — rectenna 30,000 km² → 3×10⁷ km²; launch multiple ×23,000 → ×23,000,000; both had made the scheme look more feasible) → v1.2 2026-09-25 (rule #4: every headline prints its pessimistic bound first with magnitudes on direction marks; S=20 and S=1 rows added; scenario B $/W split across the S range) → v1.3 2026-09-30 (demo-to-wedge table; no number changes). Also fixed here: erratum's capacity-factor double-count (×15 → ×3 like-for-like); the waste-heat line “0.006%, fine” (a ×1,000 kW/W slip, corrected 2026-09-23 in the pathway files). Standing effect: every correction strengthened the core claim — launch is not the mechanism. Master: knowledge/07; appendable public mirror linked from the hub artifact table. Next refresh ~2026-12.