Author: Alessandro Radice · M.Sc. Economics and Business Law (Finance), Università Cattolica del Sacro Cuore, Milan
SpaceX listed at a fixed $135 per share, about $1.77 trillion. Was that price supported by the fundamentals? An Equity Capital Markets case study on the largest IPO in history. It rebuilds the valuation from the S-1 with a sum-of-the-parts of Starlink, Space and xAI, analyses the offering structure and back-tests the model against the IPO price and three months of trading. It comes with a Colab notebook, an Excel model with live formulas, a pricing memo and a presentation deck.
Before an IPO, the ECM team builds the valuation that sets the price range and defends it in front of the issuer and investors. SpaceX skipped that step: the price was fixed at $135 before the roadshow, with no marketed range.
This project has three goals:
- Value a company with no market price. A sum-of-the-parts across the three reporting segments, each with the method that fits it: a DCF for Starlink, peer EV/Sales for Space, reference points for xAI.
- Analyse the offering. Proceeds, fees, greenshoe, free float, retail allocation, share classes and money left on the table, compared with a typical US IPO.
- Test the price. Compare the model with the IPO price and the aftermarket, and reverse the valuation to show what investors had to believe.
| Bear | Base | Bull | |
|---|---|---|---|
| Starlink (ten-year DCF) | $158bn | $323bn | $666bn |
| Space (EV/Sales) | $9bn | $43bn | $213bn |
| xAI (reference points) | $39bn | $250bn | $351bn |
| Enterprise value | $206bn | $617bn | $1,229bn |
| Value per share | $15.38 | $48.16 | $97.07 |
| vs $135 IPO price | –89% | –64% | –28% |
- No case reaches the IPO price. A conventional marketing range off the Base case, after a 10–15% IPO discount, would have been $41–43.
- What the price implies: with Starlink and Space at their Base values, $135 implies an xAI worth $1.34tn: 5.4× its February 2026 merger value and 46× its run-rate plus contracted revenue.
- The offering: $75.0bn raised ($86.3bn with the greenshoe), a 0.67% gross spread against ~5% on a typical US IPO, a 4.9% free float, 30% retail allocation, and $16.6bn left on the table on a +19.3% first day.
| Step | Module | What it produces |
|---|---|---|
| 1 | Data | FY2025 and Q1 2026 segment figures from the 424B4 prospectus, balance sheet, share count, offering terms, market prices |
| 2 | Peer comps | EV/Sales and EV/EBITDA for connectivity (T-Mobile, Iridium, Globalstar, AST SpaceMobile), space & defense (Lockheed Martin, Northrop Grumman, Rocket Lab) and AI infrastructure (CoreWeave, Nebius) |
| 3 | Starlink DCF | Ten-year unlevered DCF in Bear / Base / Bull cases, with terminal value share and implied exit multiple |
| 4 | Sum-of-the-parts | Segment values, equity bridge, value per share and the indicative marketing range |
| 5 | Offering structure | Proceeds, fees and fee saving, greenshoe, shares outstanding, free float, retail allocation, money left on the table |
| 6 | Back-test | At each market price (IPO, first day, peak, July low, September): market cap, implied xAI value and multiples |
| 7 | Sensitivity | Value per share across Starlink and xAI enterprise values |
| 8 | Exports | Excel model with live formulas and a chart |
Cover · Inputs · Comps · Starlink DCF · SOTP · Offering · Back-test · Sensitivity
- Banker colour code: blue = hard-coded input, black = formula, green = link to another sheet.
- 568 formulas: change a DCF driver, a peer multiple or a market price and every tab recalculates, including the 36-cell sensitivity grid.
- Every value reconciles with the Python engine, with zero formula errors.
- Starlink: revenue growth fades linearly from the FY2026 rate (30% / 40% / 50%) to the FY2035 rate (4% / 5% / 6%). The EBITDA margin moves from 63% to 60% / 65% / 68% by 2030. Capex falls from 37% to 15% of revenue and D&A from 24% to 13% by 2030. Cash tax at 21% of EBIT. WACC 10.5% / 9.5% / 8.5%, terminal growth 2.5% / 3.0% / 3.5%, mid-year convention.
- Space: FY2026E revenue (+10%) at the defense primes' median EV/Sales (Bear), Rocket Lab's EV/Sales (Bull) and their geometric mean (Base).
- xAI: Q1 2026 run-rate revenue at CoreWeave's EV/Sales (Bear); the $250bn value in the February 2026 SpaceX–xAI merger (Base); run-rate plus the contracted compute from Anthropic and Google ($26.0bn a year) at CoreWeave's EV/Sales (Bull).
- Equity bridge: net debt of $13.2bn at 31 March 2026 and 12.53bn pre-IPO shares. Primary shares sold at fair value do not change value per share, so the pre-money basis is used.
- Back-test: implied xAI = price × pre-IPO shares + net debt − Starlink (Base) − Space (Base).
- Peer multiples are a snapshot from 18 September 2026, not the June pricing date. Set
REFRESH_PEERS = Truein the notebook to pull current multiples from Yahoo Finance. - The pre-IPO share count comes from S-1 coverage; unvested CEO performance awards, options and RSUs are excluded.
- Adjusted EBITDA excludes stock-based compensation; the DCF ignores working capital and deferred revenue.
- The compute contracts carry 90-day termination clauses; the Bull xAI case treats them as recurring revenue.
- Starship, Mars and space-based data centres carry option value that multiples do not capture.
This project is for educational purposes and is not investment advice.
| Requirement | Details |
|---|---|
| Environment | A Google account to run the notebook in Google Colab, free tier is enough. It also runs in any local Jupyter with Python 3.10+. |
| Python libraries | pandas, numpy, matplotlib, xlsxwriter, optionally yfinance. The first cell installs them. |
| To open the outputs | Microsoft Excel or Google Sheets; any PDF reader. |
| Background knowledge | DCF, trading multiples, enterprise vs equity value, IPO mechanics (greenshoe, gross spread, free float). |
- Open
SpaceX_IPO_Valuation.ipynbin Google Colab. - Change the assumptions in the Configuration cell if you want: DCF drivers by case, peer multiples, IPO discount.
Runtime → Run all. The Excel model and the chart download automatically.- In Excel, change any blue cell on
InputsorCompsand the valuation updates.
├── SpaceX_IPO_Valuation.ipynb # the notebook (run this)
├── spacex_ipo.py # same code as a plain Python script
├── SpaceX_IPO_Valuation.xlsx # Excel model with live formulas
├── SpaceX_IPO_Memo.pdf # three-page pricing memo
├── SpaceX_IPO_Deck.pdf # seven-slide presentation
├── sotp_vs_price.png # chart used in this README
├── implied_xai.png # slide used in this README
└── README.md
- SpaceX, 424B4 prospectus and S-1, SEC EDGAR
- SpaceX S-1 breakdown, Mostly Metrics (cash, debt, share count)
- SpaceX IPO at $135: what the S-1 financials say, TradingKey
- Initial public offering of SpaceX, Wikipedia (offering terms, first day, peak, July low)
- SpaceX IPO oversubscribed as order books close, The Next Web (fixed price, retail allocation)
- SpaceX absorbed xAI at a combined $1.25 trillion valuation, The Motley Fool
- SPCX historical prices, Investing.com
- Peer multiples: stockanalysis.com, 18 September 2026
Python · pandas · numpy · matplotlib · xlsxwriter · yfinance · Google Colab · Excel