Valuing SpaceX — a real, current ESG sum-of-the-parts

Prologue

The biggest IPO in history — last month

On 12 June 2026 SpaceX priced at $135, popped 19%, and now trades around $153 — about $2.0 trillion of equity, on $18.7bn of revenue and a $4.9bn net loss.

$2T

Predictcommit before you scroll

You are an analyst the morning after the IPO. The S-1 is on your desk; the tape says $153.

Is SpaceX at $153 cheap, about right, or expensive on the fundamentals — and if it's expensive, what would you have to believe?

107× sales · a $4.9bn loss · one founder, 82.4% of the vote

We build the valuation from the filing, segment by segment, and test your answer.

Part One

One filing, three businesses

After xAI merged in, SPCX reports three segments — exactly the cut a valuation wants.

Connectivity · Starlink

$11.4bn revenue, +$4.4bn operating income, 63% EBITDA margin. The cash engine.

Space · launch + Dragon

$4.1bn revenue, near break-even. Starship is the option. The franchise.

AI · xAI / Grok

$3.2bn revenue, −$6.4bn operating loss, $12.7bn capex. The swing factor.

Three businesses, three risk profiles, three discount rates. We value each, then add them up.

Evidencethe shape of the firm

Starlink is most of the revenue — and almost all of the profit

Fig 1FY2025 revenue (filled) vs Adjusted EBITDA (marker) by segment. AI earns less and burns more: a −$1.2bn EBITDA on $12.7bn of capex. SEC S-1/A #2, CIK 0001181412

Starlink prints cash

2025 operating income +$4.4bn; 10.3m subscribers (Mar 2026), +105% YoY.

AI consumes it

2025 capex $12.7bn on $3.2bn of revenue; the IPO's first use of proceeds is "AI compute."

Net, the firm burns

OCF +$6.8bn, capex −$20.7bn → free cash flow −$14.0bn. Starlink funds the burn.

Part Two

How to value it

A sum-of-the-parts DCF: each segment at its own discount rate, summed, bridged to equity, tested against the price.

FCF = EBIT(1−τ) + D&A − capex − ΔNWC → EV = Σ FCF_t/(1+r)^t + TV/(1+r)¹⁰

NOPAT carries an NOL shield · terminal value floored at zero · ESG enters through r

Starlink 10-yr DCF, the anchor Space base DCF + Starship as a real option AI scenario-driven — the swing Bridge ΣEV → equity ÷ 13.08bn shares

Three businesses, three discount rates

WACC(1−w_d)·[r_f + β·ERP] + w_d·r_d(1−T_c) + ESG nudge. r_f=4.3%, ERP=5%. Illustrative teaching marks.
Segment	β	ESG nudge	WACC	why
Connectivity	1.00	−20 bp	8.98%	stable subs; digital-access (S) discount
Space	1.30	+30 bp	10.99%	lumpy; launch emissions, debris (E)
AI	1.55	+20 bp	12.18%	speculative; governance (G) premium

The theory's two channels, in one table: Starlink's −20 bp is a taste discount (Pástor–Stambaugh–Taylor, JFE 2021); Space's +30 bp and AI's +20 bp are risk premia — carbon/externality (à la Bolton–Kacperczyk, JFE 2021) and governance. Small by design — hold that thought.

Part Three

Valuing the parts

Starlink the anchor; Space plus a Starship option; AI as the swing.

Evidencethe anchor

Starlink — the cash engine carries the firm

Subscribers grow into the tens of millions; ARPU keeps falling ($99 → $66 → …); margins expand toward a utility-like 70%. Discount at 8.98%, terminal growth 3%.

≈ $408bn

enterprise value of Connectivity — about 7× a much larger FY2035 EBITDA. Rich, but it is the crown jewel: a global broadband utility.

Evidencefranchise + option

Space — a near-break-even franchise, plus an option

The world's dominant launch provider — but most launches are internal (Starlink), at cost, so reported margins are thin. The upside isn't in today's Falcon.

Base launch business (DCF)

≈ $8bn — yes, below its $28bn order backlog. Thin margins.

Starship — a real option

point-to-point, heavy lift, lunar/Mars cargo. Valued by scenario: $0 / $80bn / $250bn.

Predictthe swing factor

AI — tiny today, enormous if it works

$3.2bn revenue, losing $6.4bn at the operating line, $12.7bn of capex, against an issuer-claimed $26.5 trillion TAM. So we don't pretend to a point estimate — we run three scenarios.

AI segmententerprise value by scenario (10-yr DCF, WACC 12.18%).
Scenario	what you believe	AI EV
Bear	never earns its capital; burn continues	−$24bn
Base	scales to solid profit by ~2031	−$2bn
Bull	captures real TAM share	+$83bn

Even the Bull 10-year DCF is under $100bn. Hold that against the $26.5tn TAM story.

Part Four

Adding it up

Sum the segments, bridge to equity, divide by 13.08bn shares — and compare to $153.

Evidencethe bridge · Base case

Sum-of-the-partsstandard post-money bridge; pre-money cross-check lands ~$4 lower.
Connectivity / Starlink	$408bn
Space (base launch)	$8bn
Starship real option	$80bn
AI / xAI	−$2bn
Σ Enterprise value	$494bn
+ cash & IPO proceeds − debt	+$61bn
Equity value ÷ 13.08bn shares	$555bn
Value per share	$42
Market price	$153

The football field

Fig 2Connectivity + Space are the fixed anchor; the scenario swings only the moonshot (AI + Starship). Every scenario — even the Bull — sits far below the $153 market price.

The disciplined DCF says $35–62. The tape says $153.

So what is the market seeing that the cash flows are not?

Part Five

What must you believe?

We can't reach $153 from the cash flows. So invert the question.

Predictturn the DCF around

Hold Starlink, Space and Starship at our marks. What must the market be assigning to the AI / moonshot bucket to justify $153?

Guess the number before you scroll — it lands larger than you'd think.

Evidencethe implied premium

Fig 3The value the price implies for the AI / moonshot bucket, given disciplined marks elsewhere — a residual, not a standalone AI valuation. vs ≈$0 DCF · ~$200bn private xAI mark (press)

To pay $153, you must believe the moonshot is worth ≈ $1.45 trillion.

Versus roughly zero on a disciplined DCF — and ~$200bn at xAI's last private mark (press-reported).

Idearead the residual honestly

What it is

the value the price assigns to everything a DCF can't capture: Starship at scale, Mars, 100 TW of orbital compute, AI TAM capture.

What it is not

a precise AI valuation. It is a residual — it absorbs every conservative choice we made elsewhere.

A narrative stock is priced on the option, not the cash flows.

Part Six

Where ESG actually bites

We put real E, S and G factors into the discount rate. Predict: dollars, or cents?

S · taste (PST)

Starlink connects 3bn unconnected → lower cost of capital (−20 bp).

E · risk (à la BK)

launch emissions, NEPA, orbital debris / Kessler → premium (+30 bp).

G · governance risk

dual-class control, key-person, related parties → premium (+20 bp).

Verdictdollars, or cents?

ESG moves the value by a few dollars. The moonshot moves everything.

So is ESG irrelevant? No — and it's the morning's two inputs in one stock. ESG entered through the discount rate — the taste and risk channels of the theory — but for a narrative stock a basis-point repricing is second-order. The moonshot lives in the other input, the cash-flow scenarios: the market's Mars / AI beliefs, not its discount rate, carry the ~$110 gap. ESG bites here too — debris rules, Starshield geopolitics — and, above all, in governance.

Know where each force lives: ESG in the discount rate (cents), the moonshot in the cash flows (everything).

Part Seven

The governance overlay

One founder, 82.4% of the vote, and a pay award that vests on Mars.

Dual-classa Nasdaq "controlled company" — exempt from the independent-board-majority rule.
Class	shares	votes/share	voting power
A (public)	7.38bn	1	11.5%
B (Musk & insiders)	5.70bn	10	88.5%
Elon Musk — total voting power			82.4%

Musk is CEO, CTO and Chairman, with ≈46% of the economics but 82.4% of the vote.

The pay award: vest on Mars

1,000m Class B shares

vest on 15 market-cap milestones and a permanent Mars colony of ≥1,000,000 people.

302m Class B shares

vest on 12 milestones and non-Earth data centres at 100 TW/yr of compute.

≈ 1.30bn shares (~9% dilution) — but only in states of the world where SpaceX is worth multiple trillions. The dilution is contingent on the very upside that triggers it.

Concentration

82.4% control means minority holders cannot discipline strategy, capital allocation, or the $20.2bn of related-party xAI↔Valor compute leases.

…vs. conviction

the same control let SpaceX make decade-long, capital-destroying bets — reusable rockets, Starlink — a quarterly market would have killed.

The G in ESG is not a checklist here — it is whether you trust the one person who controls the moonshot.

Synthesis

What SpaceX teaches about valuation

Sum-of-the-parts disciplines a story

three segments, three discount rates — cash engine, franchise, swing factor — beats one blended guess.

When DCF and price diverge sharply, invert

the reverse DCF turns "it's expensive" into a falsifiable claim about the moonshot.

ESG lives somewhere specific

basis points in the discount rate; first-order in the scenarios and in governance.

Governance is priced

82.4% control is both the risk and the reason the bets got made.

Ideadiscussion

Are you the analyst, or the market? Value AI by DCF at all — or only as an option? Does 82.4% control deserve a discount, or a premium? Is the rest of SpaceX a giant call option on Mars?

Bring your number. We rebuild the model live and see whose assumptions survive.