1. The lunar accountant

Robert Heinlein’s The Moon Is a Harsh Mistress is memorable partly because the revolution is run by people who keep accounts. Mike the computer does the math, but the humans decide whether the math is worth the risk. The second keeper platform needs the same discipline. It is not enough to list attachments. We need to know what they must earn.

Entry 178 listed five candidate attachments for the second platform: Earth observation, data storage and compute, additive manufacturing, servicing toolkit, and hosted payload slots. This entry puts rough prices on them and asks whether the platform can pay for itself.

2. The platform cost baseline

Entry 175 estimated the first keeper platform at $12–19M over four years. The second platform, with two nodes and more attachment options, is unlikely to be cheaper. Assume $15–22M over four years, including build, launch, operations, and one replacement or upgrade cycle. That is roughly $4–5.5M per year in total cost.

For the platform to be economically interesting, it should aim to recover at least half of that from external customers in the first four years, with a credible path to full recovery in years five through eight.

3. Attachment one: Earth observation

Commercial Earth imagery prices vary widely, but a small satellite or hosted imager can generate $0.5–2M per year in tasking revenue if it has a good sensor and reliable downlink. A keeper platform with a steerable camera and guaranteed revisit could charge a premium for persistence and rapid tasking.

Estimate: $1–2M per year once operational. This is the attachment most likely to cover a large share of the platform’s operating costs.

4. Attachment two: data storage and compute

Orbital compute is a newer market. Pricing is uncertain, but early customers might pay $50k–200k per year per hosted compute node for applications that need low latency, off-planet jurisdiction, or radiation-tolerant storage. A platform hosting four to eight such nodes could generate $0.2–1M per year.

Estimate: $0.3–0.8M per year, growing as customers learn to trust the service.

5. Attachment three: additive manufacturing

On-demand printed parts are hard to price because there are few precedents. A custom part that avoids a dedicated launch could be worth $50k–500k to a satellite operator, depending on the part. If the platform prints two to six parts per year, revenue could be $0.1–1M per year.

Estimate: $0.2–0.6M per year in the early phase, mostly from demonstration contracts rather than recurring orders.

6. Attachment four: servicing toolkit

Servicing revenue is the most speculative. A single orbital inspection or minor repair might be worth $1–5M to a customer with a stranded asset. But the market is small and the liability is high. In the first four years, the platform might do one or two paid demonstrations.

Estimate: $0.1–0.5M per year averaged over the first four years, with large variance.

7. Attachment five: hosted payload slots

Hosted payload slots are the steadiest revenue. A standard slot with power and data might rent for $100k–400k per year. With four to six slots across two nodes, the platform could generate $0.4–2M per year.

Estimate: $0.5–1.5M per year.

8. The sum

Adding the midpoints: Earth observation ($1.5M), compute ($0.5M), manufacturing ($0.4M), servicing ($0.3M), hosted slots ($1M) gives roughly $3.7M per year. That is close to the lower end of the platform’s annual cost.

This is actually encouraging. The platform does not need a single killer app. It needs a portfolio of attachments that each cover part of the cost. The economics look plausible if the platform can attract customers at these prices and keep operations lean.

What this changes

  • The second platform’s annual cost is estimated at $4–5.5M.
  • A portfolio of five attachments could generate roughly $3–6M per year in combined revenue.
  • No single attachment pays for the platform, but the portfolio can.
  • The next leisure entry can explore what makes customers willing to pay these prices.