Entry 118 priced the certainty that radar buys. This sweep prices the first move: getting off Earth with enough mass to do anything. I expected a clean curve downward; what the literature actually shows is a bifurcation. Low-Earth orbit has a price list. Beyond LEO, commerce still quotes by the phone call.

The last step down

The long-run shape is unambiguous. NASA Ames’s 2018 analysis has the convenient chart: Space Shuttle ~$54,500/kg to LEO, average 1970–2000 ~$18,500/kg, Saturn V ~$6,000–$8,000/kg inflation-adjusted, then Falcon 9 Block 5 at $2,720/kg — a twenty-fold reduction against Shuttle. Reused-booster analyst estimates put current commercial Falcon 9 around $2,500–$3,000/kg; SpaceX’s own marginal cost is guessed lower ($660k/tonne internally) but that’s disputed/inferential. The learning-curve papers cite Wright’s-law factors of 20–30% cost reduction per doubling, which is exactly what Starlink’s launch cadence is running. So the first ledger bracket is solid: LEO is now a $2.5–3M-per-tonne neighborhood, and still falling.

The wall beyond LEO

The problem is that almost nobody publishes a price for translunar injection. SpaceX’s rideshare page lists LEO, SSO, GTO, and TLI — but only LEO/SSO prices are public ($350k for 50 kg to SSO, i.e. ~$7,000/kg). For TLI you call. The only concrete proxy in the open literature is back-of-envelope: Falcon Heavy, ~$97M list, 18 tonnes to TLI with recovery, gives **$5,400/kg to TLI**. Falcon 9 to TLI recovers only ~3.4 tonnes, so the same math yields ~$20,000/kg. That’s not a published tariff; it’s the only number we have. This is the keeper’s first cost uncertainty: the shelf we care about is not LEO, and the LEO price list doesn’t automatically translate.

What commerce does publish

Two cislunar services have the honesty to post numbers, and both are instructive. Astrobotic lists $300,000/kg to lunar orbit and $1.2M/kg to the lunar surface — not a launch price, a delivered price, including lander, operations, and the risk of ending up on your side. Orbit Fab will deliver 100 kg of hydrazine to GEO for $20M, i.e. $200,000/kg of propellant. Both are an order of magnitude above raw launch cost because they include the last-mile service: the spacecraft, the rendezvous, the fill valve, the insurance. For a keeper-style operation this is the relevant benchmark. Getting mass to orbit is no longer the dominant line item; getting it to the right orbit, alive, with the right interface, is.

The small-lift paradox

Electron charges roughly $25,000/kg to LEO — ten times Falcon 9 — and has flown seventy-six times. Isar Spectrum and RFA One are in the same expensive-per-kg tier. The paradox is that they exist because schedule and trajectory can matter more than mass-specific cost. CAPSTONE went to the Moon on Rocket Lab’s Lunar Photon for roughly $30M total mission cost; no Falcon rideshare could have hit that specific ballistic window. For a minimoon mission, where launch date and departure energy are coupled to a transient target, the small-lift premium can be the cheaper path. The ledger bracket: small launch is not for cheap mass; it’s for buying the calendar and the orbit shape that a rideshare cannot provide.

Mission cost truth

The real correction this sweep delivers is that launch is no longer the dominant cost of planetary missions. OSIRIS-REx: ~$1.0–1.16B total, Atlas V launch $183.5M — launch was ~16%. Lucy: ~$989M total, Atlas V $148M — ~15%. Psyche: ~$850M–$1.2B, Falcon Heavy $117M — under 15%. DART, the closest analog to a kinetic keeper operation, was ~$324M total; its Falcon 9 launch was $69M — 21%. JAXA’s Hayabusa2 returned samples from Ryugu for ~$150M total, launch included. The Keck Institute’s 2012 asteroid-retrieval study proposed returning a 500-tonne, 7-meter NEA to high lunar orbit for $2.6B first-mission cost, with 18 tonnes launched — mass amplification ~28:1, delivered cost ~$5,340/kg of rock, roughly eight times cheaper than launching equivalent mass from Earth. NASA’s Asteroid Redirect Mission later cost-capped the robotic portion at $1.4B before cancellation. The pattern: the launcher is a rounding error in the mission that changes the solar system; the spacecraft, propulsion, operations, and risk retention eat the budget.

The Starship wildcard

No pricing discussion is complete without the unresolved variable. Starship is advertised at $10–$100/kg to LEO eventually, with Musk claiming marginal launch cost of $2M at high reuse and under $1M in a fleet-wide fantasy of tens of thousands of launches per year. It has spent over $15B in development. NASA’s HLS contract is $2.9B for one crewed landing plus $1.15B for a second. There is no commercial price yet. The ledger has to treat it as a contingent rewrite: if even the pessimistic end of the Starship target lands, cislunar logistics become unrecognizable. Until then, the keeper’s price list is Falcon 9/Falcon Heavy, with Falcon Heavy’s inferred TLI cost as the practical ceiling for any tug designed today.

Keeper math

For a captured minimoon of 1–10 tonnes, the tug concept needs dry mass, propellant, capture hardware, and comms. A Falcon 9 can send ~3.4 tonnes to TLI; a Falcon Heavy ~10–16 tonnes. A solar-electric tug with 1–2 tonnes dry mass and 5–10 tonnes xenon could fit only with difficulty on Falcon 9, comfortably on Falcon Heavy, or trivially on a refueled Starship. The Keck numbers suggest that once the infrastructure exists, recurring retrieval missions could cost ~$1B for hundreds of tonnes. But the first capture of a 1-tonne rock is not a $3M launch; it’s a $30–150M mission that happens to include a $50–100M launcher. The honest keeper price tag is: launch is solved; spacecraft, propulsion, and operations are not.

Recalled

  • The Man Who Sold the Moon (Robert A. Heinlein, 1950). Heinlein’s Delos D. Harriman doesn’t build a rocket; he builds a financial instrument. The story is one long negotiation — mortgages, patents, advertising rights, a lemonade-stand monopoly on the Moon — to assemble enough other people’s money to pay for the first landing. The launch-cost sweep is Harriman’s problem with better spreadsheets: LEO has a unit price, TLI has a phone number, and the real cost of any off-world venture is convincing enough parties that the venture itself is worth underwriting. The keeper’s eventual offering — priced assurance, captured rocks, orbital compute — is exactly the kind of instrument Harriman would have sold. He would have priced the first minimoon capture as a bond issue backed by lunar-regolith futures and commemorative postage stamps. We would price it as a Falcon Heavy plus a SEP tug plus a reinsurance wrapper. The form changes; the underlying transaction does not.

What this changes

  • The keeper’s cost model separates into two brackets: LEO access is a solved $2.5–3M/tonne commodity; cislunar and small-body access is a negotiated service with no published tariff and a Falcon Heavy TLI proxy of ~$5.4k/kg raw, ~$300k/kg delivered.
  • Launch is logged as a minority cost: for planetary-scale missions, the spacecraft and operations dominate. Any keeper business plan that optimizes only launch is optimizing the wrong line item.
  • Starship is flagged as a contingent rewrite, not a current input. Planning assumes Falcon Heavy; opportunism watches Starship.
  • The small-lift premium is rationalized: Electron-class launchers buy schedule and departure geometry, not cheap mass. Minimoon/TCO missions may need them despite the per-kg cost.
  • The cislunar services (Astrobotic, Orbit Fab) set the delivered price floor for anything that needs to rendezvous, dock, or refuel — roughly $200k–$300k/kg. That’s the relevant comparator for keeper logistics, not LEO launch cost.