Is SpaceX worth $1.75 trillion?

SpaceX is expected to IPO at a $135, and roughly $1.75 trillion in equity value. This is close to 90x trailing revenue — a multiple with absolutely no precedent at this scale. Truth be told though, SpaceX is a company with few precedents of any kind.

The “it’s expensive but makes sense” case rests on assets no SpaceX competitor can replicate: 83% of all mass launched to orbit in 2025, a broadband business growing revenue 50% a year at 63% segment EBITDA margins, and a cost curve that has fallen more than 95% per kilogram since the Falcon program began.

The “it’s overvalued and will tank the market” case? Nearly $5 billion in consolidated losses, an AI segment burning $2.5 billion a quarter, and a price that analysts value $450 billion to nearly $1 trillion below the print.

• SpaceX prices its IPO at $135 per share for a ~$1.77 trillion valuation and a ~$75 billion raise (555.6 million Class A shares), the largest offering in history. Independent sum-of-the-parts estimates range from $780 billion (Morningstar) to $1.25–1.35 trillion (Damodaran) — a pretty wide gap, but both below the offer, and both acknowledging upside scenarios near or above $2 trillion if Starship and orbital compute deliver.
• Starlink is the underwritable core: the Connectivity segment generated $11.4 billion of 2025 revenue (+50%), $4.4 billion of segment EBIT, and $7.2 billion of segment adjusted EBITDA at a 63% margin, with enterprise and government revenue growing 51% at virtually zero churn. The AI segment (xAI/X) lost $6.4 billion at the operating line but holds ~1.0 GW of compute and a contracted compute deal of $1.25 billion per month from Anthropic through May 2029.
• Governance versus alignment: Elon Musk controls ~82–85% of the vote and the company opts out of independent-board rules, while ~78% of proceeds are pre-committed to debt, spectrum, and related-party obligations. This can be either read this as positive founder control (Elon built this), or a concentration risk priced at a premium rather than a discount.

SpaceX comes to market as three in one. The consolidated entity reported $18.7 billion of 2025 revenue, up 33% from $14 billion in 2024 and $10.4 billion in 2023. Those prior-year figures were retrospectively recast to fold in xAI (acquired in February 2026) and X (acquired by xAI in March 2025) under common-control accounting, presenting the three companies as if they had always been one. Standalone SpaceX did closer to $15–16 billion in 2025, so the gap is xAI and X, which arrive carrying both most of the losses and most of the speculative upside.

The segment disclosures separate the engines cleanly. Connectivity (Starlink) produced $11.4 billion of revenue, $4.4 billion EBIT, and $7.2 billion of adjusted EBITDA, with year-over-year growth of 50%, 120%, and 86% respectively. EBIT growing more than twice as fast as revenue shows the signature of a network business entering economies of scale.

Space (3rd party launch) produced $4.1 billion of revenue, up 8%, and an operating loss of $657 million after absorbing $3.0 billion of Starship development R&D — a loss that is a choice about investment, not a statement about the launch business's earning power.

AI produced $3.2 billion of revenue and a $6.4 billion operating loss. The consolidated company lost $5 billion in 2025 and $4.3 billion in the first quarter of 2026, against an accumulated deficit of $41.3 billion and $24.7 billion of year-end cash.

How anyone weights those facts depends almost entirely on how they treat the AI segment: is it a value sink diluting two (excellent) businesses, or a 3rd growth engine that is being funded by the first two?

We believe the headline multiple is the wrong starting point in either direction. At $1.77 trillion on $18.7 billion of 2025 revenue, SpaceX prices at roughly 90x trailing sales.

It’s the highest P/S ratio in IPO history, but also a blended number applied to three segments with pretty nothing in common. And while a $1.75 trillion valuation would put SpaceX on 67x sales (three times as much as Nvidia based on its past FY), single-multiple comparisons miss the crucial point: SpaceX is being priced as an infrastructure monopoly plus two call options, and the honest approach is to value each piece against its own comparable set.

Starlink. This is where the underwritable value lives. The Connectivity segment's economics resemble a network utility compounding like a software company: 63% segment adjusted EBITDA margin, near-zero enterprise churn, a revenue mix shifting toward higher-margin enterprise and government contracts. Public comps frame the range - T-Mobile trades around 10x EV/EBITDA on ~39% EBITDA margins — Starlink's margin is 24 points higher and its growth roughly ten times faster. Iridium, the only profitable pure-play satellite operator of scale, trades at roughly 9-10x EV/EBITDA and ~4.3x EV/revenue, with low-single-digit growth. A growth-adjusted 18–22x against Starlink's $7.2 billion of 2025 segment EBITDA yields roughly $130–160 billion, a forward multiple on 2026 EBITDA that could approach $10–12 billion at the current trajectory reaches $200–300 billion. Bulls argue even that is conservative: no listed comp combines Starlink's growth rate, margin, and a vertically integrated launch cost advantage that competitors must buy at market rates. A reasonable range would be $250–450 billion, with the high-end requiring continued subscriber compounding — and the subscriber base did grow from zero to 10.3 million in six years.

Launch. The Space segment's reported numbers understate it in both directions. Its $4.1 billion of 3rd party revenue excludes the internal launches that make Starlink's economics possible, and its operating loss is an artifact of Starship R&D. What about public comps? Rocket Lab trades at 92x EV/revenue on ~$680 million of trailing revenue. This is a multiple reflecting launch scarcity and growth, which SpaceX possesses in far greater measure. The big defense comps (Lockheed at ~2.3x revenue and ~17.6x EV/EBITDA, Northrop at ~15x EV/EBITDA) anchor the mature end. SpaceX's position sits in neither category though: 165 orbital flights in 2025, a 99%+ (!!) Falcon success rate, and a launch cost per kilogram reduced by more than 95%. NASA Ames analyst Harry W. Jones documented the longer arc: average launch cost of $18.5k/kg from 1970 to 2000 against Falcon 9's advertised $2,720/kg. A standalone launch business might be reasonably worth $150–350 billion, with the spread driven by how much Starship option value an investor assigns.

xAI/X. The fun part. The AI segment generated $3.2 billion of 2025 revenue against a $6.4 billion operating loss, with X advertising shrinking post-Elon revolution (down roughly $100 million year-over-year in Q1 2026) and capex of $12.7 billion in 2025. On surface this reads as a frontier lab competing against OpenAI, Anthropic, and Google but with weaker unit economics, funded by guarantees the parent had to provide. On the other hand though, there is a compute play. The COLOSSUS and COLOSSUS II clusters provide about 1.0 GW of operating compute in a market where compute is the binding constraint. The Anthropic agreement contracts $1.25 billion per month (!) through May 2029 — a $15 billion annualized run-rate from a single customer. We assume a defensible valuation mark of $150–350 billion.

Let’s add the parts: $250–450 billion of Starlink, $150–350 billion of launch, $150–350 billion of AI = $0.9–1.3 trillion of equity value on demonstrated economics. The offer price sits $450 billion to nearly $900 billion above that midpoint. Whether that premium is excess or rational depends on what an investor believes scarcity, control, optionality, and Elon-factor are worth.

Pricing at $1.77 trillion is an embedded-expectations exercise. To support the multiple at a normal long-run infrastructure rating, SpaceX needs to become a $70–175 billion revenue company depending on the exit multiple assumed — 4-9x its current top line. The key question is whether the demonstrated trajectory makes that plausible.

The case that it does: Starlink went from zero revenue to $11.4 billion in six years and is still growing 50%. Reaching 30 million subscribers at $50 monthly ARPU with roughly flat constellation cost would put segment EBITDA near $18 billion, and 50 million at $40 ARPU above $25 billion. The EchoStar spectrum acquisition converts the direct-to-cell product from messaging to full broadband on unmodified handsets (a market measured in billions of subscribers). V3 satellites, at one terabit per second each and 60 per Starship launch, multiply constellation capacity roughly 20x per flight. And the company's track record on seemingly impossible engineering — propulsive booster landing, full reusability, a 9,600-satellite constellation — is the strongest qualitative argument the SpaceX lovers have.

The case that it doesn't, or at least not at this price: ARPU is falling, and the bull math requires ground-operations costs to scale far more slowly than subscribers. Starship has flown 11 test flights (and no paying customer), and commercial payload delivery is guided to the second half of 2026. Plus, the program absorbs roughly $3 billion a year in R&D, and SpaceX lists its execution as the top risk factor. Every dependent revenue line — V3 deployment, direct-to-cell at scale, lunar logistics, orbital data centers — waits on its cadence. Orbital AI compute and Mars remain moonshot optionality, and the S-1 prices Elon's compensation against them precisely because they sit outside any underwritable forecast.

The repeatable engine is reusability driving cost down, which makes new businesses viable — and the flywheel is very self-reinforcing. SpaceX launches its own satellites, so constellation capex is internal cost rather than a 3rd party bill that competitors like Amazon's Leo must pay at market rates.

The enterprise and government side of Starlink is the high-quality revenue story. Maritime ARPU runs around $34k annually and aviation around $300k per airline customer, against residential ARPU in tens of dollars per month. Enterprise and government revenue grew 51% to $4.2 billion in 2025, faster than consumer, at higher margins and near-zero churn. This is a mix shift that partially offsets the consumer ARPU decline, and is the strongest counterargument to the margin-compression thesis.

The competitive field is real but trailing. Amazon's Leo had launched roughly 331 production satellites as of May 2026 against Starlink's ~9,600. It also faces a July 30, 2026 FCC milestone to deploy half its constellation, and depends on 3rd party rockets. AST SpaceMobile targets direct-to-cell through carrier partnerships reaching operators serving 2.8 billion subscribers but generated just $14.7 million of Q1 2026 revenue against 2026 guidance of $150–200 million. Chinese constellations and OneWeb round out the field, but clearly none is positioned to contest Starlink's lead this decade — which both supports the premium (a durable moat) and frames the risk (the valuation requires the moat to hold for considerably longer than that).

SpaceX will be a controlled company under Nasdaq rules, with Elon Musk holding roughly 82–85% of voting power through 10-vote Class B shares, and intends to opt out of the majority-independent-board requirement. Class A buyers are purchasing economic exposure with minimal governance leverage. The skeptical reading is pretty clear: concentration risk usually means a discount, but in SpaceX’s case it is being priced at a premium. The counter-reading is that Elon’s control is not incidental — the same concentrated decision-making that public-market governance would have constrained is what funded Starlink through years of losses and bet the company on reusability when the industry called it uneconomic.

The Tesla footprint is large and multidirectional too: xAI bought $506 million of goods and services from Tesla in 2025 and $303 million in the first four months of 2026. SpaceX bought another $144 million including over $130 million of Cybertrucks. Tesla's roughly $2 billion xAI investment converted into approximately 19 million SpaceX shares. The Cursor option signed in April 2026 carries a $10 billion downside too (if SpaceX walks). Key-man concentration is huge — Elon is CEO, CTO, and Chairman across the combined entity — and can be both the largest single risk factor and the largest single asset.

Elon’s award is obviously the filing's most discussed part. Per the filed exhibits, Musk received two performance grants of super-voting Class B stock. First, roughly 1 billion shares, vesting in 15 tranches, each requiring a market-cap milestone and "a permanent human colony on Mars with at least one million inhabitants". Then, roughly 302 million shares, across 12 tranches tied to market cap thresholds and "completion of non-Earth-based data centers capable of delivering 100 terawatts of compute per year."

SpaceX accounts for the Mars milestone as "improbable," recognizing zero compensation expense against it. The structure technically means that Elon’s largest award pays nothing unless shareholders first see the equity value multiply several times over. It’s pay-for-performance at a scale (and difficulty) that no other public company package approaches.

The prospectus frames proceeds around AI compute, launch infrastructure, and satellite constellations. Separate disclosures show a large majority is pre-committed: analysis puts roughly 78% (~$63 billion of a ~$80 billion raise) as already pledged across Valor obligations, repayment of legacy xAI/X creditors, and the EchoStar closing, leaving under $18 billion of discretionary growth capital. A $20 billion bridge loan arranged in March 2026 must be repaid within roughly six months of receiving proceeds.

The bad reading is that the growth story is funding debt service. The neutral reading is that the obligations being retired are themselves growth investments already made — the spectrum that enables direct-to-cell, the compute clusters already operating — and the IPO is the planned refinancing of a deliberate sequencing: buy the assets first on bridge financing, term them out in the public market.

SpaceX skipped the conventional bookbuilding and went straight to a fixed $135. Roughly 30% of the offering is reserved for retail through a directed-share program, and those shares carry no lockup, while pre-IPO institutional holders face the standard 180 days. The float is small (relative to market cap), SpaceX absolutely clears S&P 500 and Nasdaq-100 thresholds by a wide margin. This can be a very volatile, momentum-driven open that may detach from fundamental value in either direction.

Regardless of where the stock settles, this offering is a landmark for late-stage private markets. SpaceX raised more than $10 billion across roughly 30 private rounds over nearly a quarter century. This IPO is the largest liquidity event the VC ecosystem has seen, returning capital to Founders Fund, DFJ, Alphabet, Fidelity, and many more. Thousands of SpaceX employees will become millionaires. It arrives alongside Anthropic's and OpenAI’s filings. This is the largest concentration of pre-IPO capital ever brought to market at once. A strong SpaceX debut pulls the entire late-stage backlog forward into 2026, but a weak one might push the window toward 2027.

Strip away the noise and all of this comes down to one question - what do you pay today for businesses that don't exist yet, sold by the only company that keeps building them?

We believe the math is sound. Demonstrated economics — Starlink being a cash machine, having the launch monopoly, the contracted compute — it all supports $1.0–1.3 trillion. The valuation gap, somewhere between $450 billion and $900 billion, is the price of believing that Starship flies commercially on schedule, that orbital data centers become a line item rather than a slide, and that the next decade of SpaceX will be as good as the last one.

It's also a bet that has previously paid off. Investors who passed at $137 billion in 2023 or $350 billion in late 2024 ran the same math and were right on the numbers, but wrong on the outcome. SpaceX is an incredible success story that no spreadsheet can explain. It’s a bet on Elon's thesis — that humanity becomes a multi-planetary species, and that the company building the rockets, the bandwidth, and the compute to get there captures most of the value of that transition. If he's even half right, the largest IPO in history will eventually look like the cheapest entry into the largest company ever built.