Cybercab fleet economics — personal opportunity breakdown

Founder asked (2026-08-13) to assess buying a Tesla Cybercab through Ray Data LLC and running it on the robotaxi network for passive income, prompted by [[2026-09-05-moonshots-gpt6-astra-arc-agi3-tesla-cybercab-fermat]]. Delivered the same night as an interactive model (scratchpad HTML, sent via iMessage) — never filed to the vault, and the scratchpad file is gone now. Reconstructed here from the working-context record 2026-08-13 ~21:05 ET, so treat as best-effort exact on the base case (independently verified to reproduce the recorded numbers) and directionally right on the rest.

The base case

• Capex: $34,000 per vehicle
• Fare rate: $1.75/mile, platform (Tesla/network) take 30% → owner keeps $1.225/mile
• Assumed utilization: 120 paid miles/day
• Net: $2,566/month
• Payback: 13.3 months

The one variable that decides everything

Utilization (paid miles/day) drives the entire outcome — nothing else in the model moves the needle as much. Modeled payback swings from ~7 months at aggressive utilization (~190 mi/day, the vehicle working most of its waking hours) to ~64 months at weak utilization (~57 mi/day). There is no real Tampa demand data to say which end of that range is realistic. The founder's gut-check at the time: comfortably above ~90 mi/day the economics work; well under ~75 mi/day they don't.

Four assumptions the founder assumed away to run the numbers

None of these were verified — the model runs on the assumption they're all true:

1. Ray Data LLC (not an individual) can legally purchase a Cybercab.
2. The vehicle can join Tesla's Robotaxi network at will — no waitlist, approval gate, or geofence exclusion.
3. Tampa is a live or near-term entry market for the network.
4. Fleet/commercial insurance (Lemonade was the candidate) is actually available for this use case.

If any of these is false, the model doesn't run at all — they're kill questions, not risk factors to shade the numbers for.

Not modeled

• Bonus depreciation tax shield (Section 179 / 100% bonus depreciation could materially change year-1 cash economics)
• Residual/resale value of the vehicle at exit
• A possible FSD "operator fee" Tesla could charge fleet participants — unknown/unpriced

Recommendation (still stands, unanswered)

Don't buy a fleet off a spreadsheet. Buy one Cybercab as a 90-day instrumented pilot (~$34k) to convert the real unknown — actual Tampa-market utilization — into data before any fleet-scale decision. Everything else (LLC purchase eligibility, network access, insurance) needs to be verified before that pilot purchase too, not after.

Status

Delivered 2026-08-13, no founder response logged since. Kill questions unverified. If he wants to go further, next step is real Phase 0 research on the four assumptions above, not a bigger spreadsheet.