A US$75bn raise funds Starship, Starlink V3 and a 2028 orbital compute bet that reframes valuation
Space Exploration Technologies Corp. (NSDQ:SPCX) has set its IPO at US$135 per share for 555.6 million primary shares, with pricing expected on 11 June 2026. Goldman Sachs, Morgan Stanley, BofA, Citi and J.P. Morgan are running the book. At the offer price, the raise alone clears US$75 billion before any greenshoe.
The headline number is enormous, but the more interesting question is what investors are actually buying. The roadshow deck pitches three businesses welded together. Launch (Falcon and Starship), Connectivity (Starlink) and AI (xAI, Grok, X and the new Colossus compute build).
We think the most underappreciated piece sits in the AI section. The deck flags potential deployment of orbital AI compute satellites as early as 2028, leveraging Starship payload capacity and Starlink’s mesh network. That is the line that quietly does the heaviest lifting in justifying the valuation.
For ASX investors, this matters because SpaceX listing in the US reprices every domestic space, satellite and defence-adjacent name overnight.
The numbers behind the raise are bigger than the headline suggests
Group revenue grew from US$10.4 billion in 2023 to US$18.7 billion in 2025, a 33% step up year on year. Adjusted EBITDA hit US$6.6 billion, up from US$3.8 billion two years earlier. Connectivity is now the profit engine, throwing off US$7.2 billion of segment Adjusted EBITDA in 2025.
Capex tells the other half of the story. Spend went from US$4.4 billion in 2023 to US$20.7 billion in 2025, with AI infrastructure now the largest single bucket.
GAAP net income remains negative at a 26% margin in 2025, but management targets a roughly 45% net income margin longer term. That is the gap investors are being asked to underwrite at IPO.
Connectivity is doing the heavy lifting the AI story leans on
Starlink ended Q1 2026 with 10.3 million subscribers, up 105% year on year. Connectivity revenue hit US$11.4 billion in 2025, growing 50%. Coverage now spans 164 countries with more than 9,600 active satellites in orbit.
The V3 satellite generation, expected to begin deployment on Starship in the second half of 2026, lifts bandwidth per satellite from 96 Gbps to 1,024 Gbps. That is more than a 20 times step up in bandwidth per launch.
Our concern is timing. Starship has flown 12 test missions, but ship reusability has not yet been demonstrated end to end. Every quarter of slippage compresses the window for the orbital compute narrative to land.
The orbital AI pitch is where the valuation has to be defended
AI revenue was US$3.2 billion in 2025, up 22%, but segment Adjusted EBITDA was negative US$1.2 billion. Cloud compute agreements with Anthropic (US$1.25 billion monthly) and Google (US$920 million monthly from October 2026) are anchoring near-term monetisation.
The bigger swing is orbital. Management frames terrestrial compute as supply constrained, citing a 2025 US shortfall of roughly 13 gigawatts. Orbital compute is positioned as the alternative, with solar power, radiative cooling and Starship-enabled deployment.
The skeptical read is that orbital AI compute is a 2028 line item being used to justify a 2026 valuation. The bullish read is that nobody else can credibly attempt it, and optionality on a new compute substrate is worth a lot.
The Investors Takeaway for SpaceX
At US$135 per share, investors are paying for three businesses moving at different speeds. Connectivity is already a high-margin compounder. Space is the foundational moat, and AI is the optionality that closes the valuation gap.
We think the next 18 months will be defined by Starship cadence, V3 satellite deployment, and whether the Anthropic and Google compute contracts ramp on schedule. Slippage on any of the three weakens the orbital AI argument the IPO price is leaning on.
For broader context on global space and AI infrastructure plays accessible to Australian investors, see our coverage at stocksdownunder.
