One-pager for a board fight

What to take into the room

• Cash vs furnace

  Tesla is the cash cow. SpaceX is the capex furnace.

  Tesla: public, cash-flow positive, ~$1.4–1.5T-class in mid-2026 merger coverage, ~$43–45B cash pile in contemporaneous reports. SpaceX: public June 2026 at ~$1.8T-class print (first week ~$2.44T). Still burning on AI/Starship. Q2 2026 net loss narrowed to $541M on $7.8B revenue. Combining them “might make Elon’s life a little easier.” That is the honest executive reason.

• Collar

  Tesla holders may only accept 50–66% of combined equity.

  Morningstar-style work. Current ~1.4T / (1.4+2.2) ≈ 39% would be unfair to Tesla holders if SpaceX stays bid-up as a meme-AI wrapper over a launch co that is not yet Starship-operational. An independent Tesla committee has not published a valuation collar a court can defend.

• Both sides

  Elon sits on both sides of the table.

  ~13–20% Tesla economic / ~20% votes (options-dependent). SpaceX voting control described ~84% post-IPO. That is the central agency problem, not a personality story. A pay package is not a merger agreement.

• Silent killer

  China is the file that can kill the combination.

  Tesla China is the cash printer. Starshield is NRO-operated, government-owned. They do not share a regulator. Beijing + CFIUS is a gate, not a vibe. The scenario lab’s China toggle is the honest model input.

• Clock

  Base-case window: late 2027 – mid 2028.

  Earlier if Flight 14+ deploys V3 to orbit, robotaxi miles look like a business, and an independent Tesla committee publishes a collar. Later if China blocks, a robotaxi fatality freezes NHTSA, Starship fails the orbital-reuse loop, or SpaceX stays bid-up.

Do not confuse

These substitutions are how a board fight gets lost. Each is a labeled reconstruction of a common mix-up, not a prediction.

• Supervised FSD with autonomy
• An HLS award with a landing
• An IPO print with a DCF
• A 2026 Mars slide with a city
• A factory photo with nameplate
• A LOI with a fab
• Academy bots with useful labor
• Energy GWh with Solar Roof

Slipping / blocked trenches

• T19 · slipped · 80 mo

  Roadster 2 (and the A71 SpaceX package)

  The car. The thrusters on a public stage. A production date that survives a quarter.

  8%

  $1.0B–$4.0B

• T24 · slipped · 72 mo

  Solar Roof / rooftop solar

  Cost-down that beats a rack of silicon. Install capacity. A roof that is not a custom job.

  22%

  $2.0B–$8.0B

• T8 · slipped · 60 mo

  Unsupervised Robotaxi / Tesla Network

  Fleet is still small — dozens of Cybercabs / converted Ys vs Waymo scale. Cumulative paid miles still in the low millions. Unit economics vs $0.30/mi are not in segment reporting. Geofenced Sun Belt ≠ anywhere.

  24%

  $25.0B–$80.0B

• N · slipped · 36 mo

  Uncrewed Mars 2026 window (and the slip track)

  The ships, the window, the landing, the useful cargo. Completion of 'useful uncrewed Mars landing' 5–12%.

  8%

  $20.0B–$80.0B

• T11 · slipped · 24 mo

  Next-gen affordable / compact LFP platform

  The car. The price. The volume line.

  28%

  $4.0B–$12.0B

• M · slipped · 24 mo

  Starship HLS / Artemis path

  LEO docking test ~mid-2027. Crewed landing (Artemis IV talk ~2028). Depot. Crew systems. Surface ops.

  24%

  $8.0B–$20.0B

• T32 · blocked · 20 mo

  FSD China / EU type approval

  A China stack that is not a hostage. UNECE / EU ADS. Training data that can leave or a local cluster that is good.

  16%

  $1.0B–$6.0B

• T20 · slipped · 18 mo

  Unboxed process / Giga Mexico

  Giga Mexico as a volume plant. Unboxed at 3/Y scale. A cost structure that is not a keynote table.

  15%

  $4.0B–$12.0B

Daily wire

• Sep 2026 · spacex

  Starship Flight 14 orbital attempt

  NET 15 Sep. First true orbital + ~20 V3. Ship catch delayed.

• Sep 24 2026 · tesla

  Semi factory inauguration (Sparks)

  Ceremonial open. High-volume line already running since April. Nameplate is 50k/yr.

• 2H 2026 · tesla

  Cybercab ramp

  Line started Feb 2026. Ramp described as late-2026 into 2027.

• 2H 2026 · tesla

  Optimus internal Academy units

  Fremont S/X converted ~late Aug. Not customer sales.

• Dec 2026 · spacex

  AI hosting ~$1.11B/mo starts

  CFO: third CSA from Dec 1. 90-day outs. Not FY $100B.

• Aug 2026+ · tesla

  Megapack 3 production

  ~5 MWh pack entered production August.