The SpaceX IPO: Is It a Space or an AI Company? (Part 1: Valuation)
TL;DR -- A company whose origin is in space launch and satellite connectivity has done four AI deals in the four months before its IPO. Here’s a framework for thinking about the largest IPO ever.
SpaceX prices its IPO on June 11 and begins trading under the SPCX ticker on June 12 at a targeted $1.75 to $2 trillion valuation, the largest initial public offering in history. Hot on the heels of the Cerebras IPO in May and possibly ahead of other marquee IPOs that have been teased by OpenAI and Anthropic later this year, there is ample investor interest. But what is a prospective investor actually buying?
SpaceX is a complex of three lines of business that are at different stages of maturity and integration with each other. Its existing operating businesses, valued generously against the most flattering operational peers we can identify, suggest an enterprise value of approximately $1 trillion. The $1T premium on top of that valuation prices lines of business that SpaceX has yet to prove at commercial scale, and some that are simply aspirational.
It’s a given that SPCX won’t trade like a regular stock because of the Musk-driven narrative. But it’s still necessary to have a framework to understanding the company. This piece is the first in a four-part series that takes a data-driven approach to SpaceX’s S-1:
Part 1 (this piece): A valuation of SPCX’s operating businesses
Part 2: What an investor must believe to support the IPO price
Part 3: A close reading of the S-1, and what’s NOT disclosed
Part 1: What the Operating Businesses Are Worth
Note: Throughout, unless otherwise stated or linked, all SpaceX numbers are from the S-1.
Space Segment
The Space segment launches cargo into Low Earth Orbit (LEO). It looks like a slow-growth business on the surface, with revenue growing from $3.56B (FY2023) to $4.09B (FY2025), roughly 7% per year. Q1 2026 revenue was $619M, down 28% YoY. Read in isolation, the segment appears to be stalling.
Yet, underneath is a more nuanced story. Total Falcon launches actually grew from 96 (FY2023) to 165 (FY2025), with launch infrastructure growing every year. What’s absorbing the additional capacity is Starlink; internal Starlink launches grew from 63 to 122 across the same period, while customer-paying launches are down slightly from 45 (FY2024) to 43 (FY2025). In other words, Space segment revenue is capped by whatever capacity is left over after Starlink takes its share. Management makes it explicit that this will continue: “Space revenue growth to continue to be lower than total company revenue growth”.
Starship (the successor platform to Falcon) development sits inside this segment and is the main reason segment EBITDA inflected to a negative $351M in Q1 2026 from positive $224M a year earlier. Starship R&D ran at $930M in Q1 alone and $3.0B across FY2025, meaning the segment is spending most of what it earns on a vehicle that has yet to deliver a commercial payload.
Connectivity Segment (Starlink)
Starlink is a global satellite internet service that operates a constellation of approximately 9,600 LEO satellites, especially designed to reach remote and rural locations with low fiber-optic or cable internet penetration. It is the only SPCX segment that has significant scale, rapid growth, and positive segment economics. As of March 31, 2026, it served 10.3M subscribers across 164 countries. Segment revenue grew from $3.87B (FY2023) to $11.4B (FY2025), roughly tripling in two years. Segment Adjusted EBITDA grew faster, from $1.6B to $7.2B during the same period, with the margin climbing from 41% to 63%. Q1 2026 ran at a 64% EBITDA margin.
This growth has come with deliberate ARPU compression: from $99/mo in FY2023 to $66/mo in Q1 2026, a one-third drop. Per management, ARPU is expected to continue its decline, as international expansion brings lower-priced tiers into the subscriber mix. EBITDA margin growth in Q1 2026 tells us that incremental international subscribers are contributing positive marginal cash flow, even at lower prices. However, Q1 EBITDA margin YoY is negative, suggesting that the rate of growth is likely slowing. So, while Starlink remains the only SpaceX segment with at-scale unit economics today, it is unlikely to be the segment that drives near-term forward growth, as our FY2026 projections below show.
AI Segment
The AI segment did not exist in its current form three months ago and now accounts for most of the FY2026 revenue growth and a meaningful share of the IPO premium. Revenue was $2.96B in FY2023, declined 11.5% to $2.62B in FY2024 on X advertising weakness following Musk’s 2022 acquisition, and recovered 22.2% to $3.20B in FY2025. The recovery was driven by X and Grok subscription growth of $0.37B and data licensing of $0.09B, not by advertising rebound.
Segment Adjusted EBITDA followed a steeper decline. It was $1.2B in FY2023, $0.35B in FY2024, and -$1.24B in FY2025. Q1 2026 ran at -$0.61B of EBITDA against $0.82B of revenue. The deterioration reflects the build-out of AI compute infrastructure to support xAI training. Capex grew from $0.46B (FY2023) to $12.7B (FY2025), and Q1 2026 alone consumed another $7.7B (annualized, the run rate is above $30B).
The current $0.82B of quarterly revenue represents the pre-Anthropic base of X advertising plus Grok subscriptions. The size of the AI segment going forward depends almost entirely on how the deal with Anthropic ramps.
FY2026 Revenue Estimates
Below is our estimate of SPCX FY2026 revenue, built from the segment-level disclosures and operating metrics in the S-1.123
The total consolidated 2026 revenue range of $30B to $34B implies growth of 60% to 82% from FY2025. Almost all of this would come from Anthropic in the AI segment, whereas Connectivity grows meaningfully but at a decelerating pace from FY2025 and Space is essentially flat.
Peer Multiples
We restrict the peer set to public companies whose operations are directly comparable to a SpaceX segment.
A few notes about this peer set:
The mature satellite communications cluster (Iridium, Viasat, EchoStar) fall into a narrow band of 3.2x to 7.4x. These are slow-growing or declining businesses, and the band represents the operational floor for satellite-based subscription connectivity. Iridium has 2.5 million subscribers in satellite voice and IoT. EchoStar’s Hughes consumer broadband business has 0.7 million subscribers and is shrinking. Viasat is a more apt comparison because it operates across consumer broadband, government satcom, maritime, and aviation, but every one of these sub-segments is mature: government satcom grew 5.6% year-over-year for the nine months ended December 2025, maritime declined 3.8% over the same period, and consumer broadband subscribers fell 44% in 18 months.4
The direct-to-cell scarcity peers trade dramatically higher. Globalstar at 35.4x reflects the Apple Emergency SOS partnership and the recently announced acquisition by Amazon. AST SpaceMobile at 120.8x is pre-revenue option pricing on 50 MNO partnerships and a constellation that is still being deployed.
Rocket Lab at 73.2x and CoreWeave at 5.9x are each the only public pure-play in their respective segments, but the multiples are very different because the competitive set around each is very different. Rocket Lab has no comparable public launch peer and earns the full scarcity premium. In contrast, CoreWeave competes with hyperscaler (AWS, GCP, Azure) compute subsegments, private peers including Nebius and Lambda, and a growing list of AI infrastructure providers. Its multiple is further pressured by Microsoft customer concentration of 67%.
The general pattern is that the multiples scale inversely with the number of comparable companies. This is why our Connectivity scarcity multiple ends up much higher than our AI scarcity multiple even though Connectivity is not a better business than AI infrastructure, based on the fundamentals.
Valuation: Peer Ceiling and Peer Floor Cases
Applying these multiples to our 2026 revenue projection produces two scenarios that bound what disciplined peer-multiple analysis can support.
The Peer Ceiling Case uses the highest defensible operational-peer multiple for each segment. Connectivity gets Globalstar’s 35.4x, anchored on the direct-to-cell scarcity premium.5 Space gets Rocket Lab’s 73.2x, anchored on the pure-play launch scarcity. AI gets 12.0x, a 2x premium to CoreWeave that reflects SpaceX’s vertical integration across rockets, satellites, compute, and a frontier AI model.6 The total enterprise value implied is approximately $1.02T.
The Peer Floor Case uses more conservative multiples. Connectivity is valued at 5.0x, the average of Iridium, Viasat, and EchoStar. Space is valued at 5.0x, a defense-prime-style multiple. The absence of a mid-range peer between mature satcom and Rocket Lab at 73.2x makes this a judgment call. AI is valued at 6.0x, CoreWeave’s multiple rounded up. The total enterprise value implied is approximately $177B.
Both cases fall well short of the $1.75T to $2.0T IPO target.
Coming up in Part 2…
Sophisticated private secondary buyers in late 2025 were paying a partial scarcity premium ($800B) but not the full operational-peer scarcity premium ($1T). The February 2026 xAI merger marked SpaceX up to the full Peer Ceiling level. Now, the IPO asks public investors to pay another 75% or more on top of that for businesses that SpaceX has not yet built at commercial scale. We’ll examine those next in Part 2.
We put together this four-part analysis in just a day using Revelata’s deepKPI. You can do this kind of work yourself by signing up for our free tier. No credit card required.
The Connectivity projection assumes quarterly subscriber additions decelerate from the 1.4M in Q1 2026 to around 1.0M by Q4, reflecting natural subscriber S-curve flattening and the late-2026 entry of Amazon’s Project Kuiper into the broadband market. ARPU is assumed to keep declining but at a slower pace than the 23% year-over-year decline in Q1.
The AI projection assumes the Anthropic contract begins at a reduced fee in May and June 2026 and runs at the full $1.25B per month from July onward, and that Anthropic does not exercise its 90-day cancellation right within the year.
In our build, compute infrastructure accounts for approximately 70% of segment revenue, which informs our peer selection below.
We’re only looking at the last nine months ending December 2025 to factor out Viasat’s merger with Inmarsat.
We don’t use ASTS’s 120.82x multiple here because applying it to Starlink’s $14.6B of revenue would imply $1.76T for Connectivity alone, which is a category mismatch between option pricing and operating-revenue pricing.
The AI multiple of 12x is judgment-based since no pure-play public peer trades at 12x. CoreWeave at 5.85x is the only direct AI compute comp, and the 2x premium for vertical integration sets the upper bound that operational peer data can support.
 | A guest post by
|