// experiment 004 :: valuation
SpaceX, taken apart. Twenty years, four pieces.
An independent sum-of-the-parts model for SpaceX (SPCX). It values Starlink, launch, Starship and the AI compute arm on their own, marks year-end 2026 with a $30B annualized compute run-rate, then projects 2030, 2035, 2040 and 2045. Every assumption is editable and the full Excel is free to download.
By Joe Halstead · Assumptions as of July 21, 2026 · Model v1.1
SpaceX (SPCX) sum-of-the-parts valuation, 2026 to 2045
In this independent sum-of-the-parts model, the base case puts SpaceX near $99 per share at year-end 2026 with Colossus AI compute at a $30B annualized run-rate, then about $332 in 2030, $816 in 2035, $1,361 in 2040 and $1,737 in 2045, versus about $120 today. Compute is the largest slice early; Starlink is the steady second engine; Starship grows from a sliver toward a large option by 2045. Bear and bull cases are shown alongside each number.
This is a scenario model, not a prediction, and not investment advice or a recommendation to buy or sell any security. The figures are independent, chosen inside ranges from public analysts (from a sober intrinsic floor to the long-horizon bull case) and current 2026 fundamentals. No single source is treated as gospel. Models can be wrong. Do your own research.
// the method
How the number is built
Instead of one guess at a price, this model values SpaceX as the sum of four separate businesses: Starlink connectivity, launch and space, Starship, and the AI compute arm known as Colossus. Each segment is valued on its own forward revenue times a sales multiple that reflects its margin and growth quality. I add the segments up and divide by shares outstanding for a price per share. SpaceX went public in June 2026 and trades near $120, with a market value around $1.6 trillion.
The numbers are independent. They were set inside ranges from live 2026 research and public analysts, from a sober intrinsic floor to the long-horizon bull case, and none of them is copied from any one analyst. This is a scenario model with its work shown, not a prediction, and not investment advice.
// price targets
Base, bull and bear
Modeled price per share at year-end 2026 and each five-year mark after. Base is the center I would defend today; bull and bear are the plausible edges.
| Scenario | 2026 | 2030 | 2035 | 2040 | 2045 |
|---|---|---|---|---|---|
| Bear | $31 | $53 | $99 | $144 | $180 |
| Base | $99 | $332 | $816 | $1,361 | $1,737 |
| Bull | $213 | $1,110 | $3,218 | $5,182 | $6,405 |
// where the value comes from
Sum of the parts
Each slice is one business segment's share of the base-case enterprise value in that year.
// the segments
Each piece, over time
Percent of base-case value by segment. Watch how the mix shifts across twenty years.
| Segment | 2026 | 2030 | 2035 | 2040 | 2045 |
|---|---|---|---|---|---|
| Starlink (connectivity) | 19% | 25% | 23% | 21% | 21% |
| Launch / Space (Falcon + Starship cargo) | 5% | 4% | 3% | 3% | 3% |
| Starship P2P + Lunar/Mars | 2% | 5% | 11% | 16% | 20% |
| AI Compute / EWS (Colossus) | 73% | 66% | 63% | 60% | 57% |
// the workbook
Download the model
The full Excel workbook has a cover card, price targets, base/bear/bull segment tables, segment-mix percentages, every assumption as (bear, base, bull), and a method tab. It is the same engine that generates the charts and tables on this page.
// questions
Frequently asked
-
Q
Is this a SpaceX stock price prediction?
No. It is a scenario model that shows what SpaceX is worth if a specific set of assumptions comes true, in three cases: bear, base and bull. It is not a prediction and not investment advice.
-
Q
What is SpaceX's biggest value driver in this model?
Year-end 2026 anchors AI compute at a $30B annualized run-rate. Colossus remains the largest early slice, Starlink is the steady second engine, and Starship grows into a large option by 2045.
-
Q
Can the assumptions be changed?
Yes. The whole model is driven by a single assumptions file. Any input, like Starlink subscribers by 2035, can be changed and the price targets, charts and Excel all regenerate together.
// honesty
A range, with its work shown
Twenty-year valuations of a frontier company carry very large uncertainty, which is why this is a range with every driver visible. The base case is the center I would defend today. The AI compute arm is both the reason the number is large and the reason it could be wrong.