The $15.8 Billion AI Bet Hiding Inside SpaceX’s Q3 Earnings
(SeaPRwire) – By: Ethan Gallagher
The stock is down 15% in eight sessions. It sits at $146. Most retail traders think this is a space company story. They watch Starship. They count the satellites. They miss the point entirely. The real story is not in the launch pad. It is in the $15.83 billion spent on AI infrastructure last quarter. That figure dwarfs the connectivity revenue. It dwarfs the launch revenue. It is the true driver of this balance sheet right now. I have seen this pattern before in hardware cycles. It usually ends with the market realizing the “second act” is a black hole, not a growth engine. The market is nervous. It is price in the fear of capital depreciation, not just execution risk.
Wall Street consensus expects revenue of $12.89 billion. That is a 65% jump year-over-year. Morgan Stanley’s Adam Jonas sees $13 billion in revenue and $6.9 billion in adjusted EBITDA. He projects 14.3 million Starlink subscribers. At $63.60 a month each, that is a healthy recurring stream. But look at the cost side. Capital expenditure hit $18.4 billion in Q2. Of that, nearly 86% went to AI infrastructure. The official release frames this as “investment.” The industry subtext is desperation. They are buying compute capacity to secure a moat in inference markets. They are betting that orbital data centers or on-board AI processing will outpace ground-based constraints. The gap between the $143 million operating loss and the revenue growth is where the profit is bleeding out.
The “second quarter showed” section in most reports is a trap. Connectivity revenue rose 66% to $4.29 billion. AI revenue jumped 247% to $2.56 billion. That 247% number looks sexy. But it is small. It is tiny compared to the $15.83 billion spent to build the systems that generate it. Jonas points to Starship Flight 15 as a catalyst. He thinks a ship catch in November is key. That is a hardware milestone. It does not fix the software margin problem. The debt issuance projections are stark. $60 billion in 2027. $80 billion in 2028. $90 billion in 2029. This is not a tech growth story. This is a sovereign-scale infrastructure buildout financed by corporate debt. The market is trading the stock like a tech darling. The filings read like a national grid expansion.
Investors are underpriced on the AI segment. Jonas values it at only $19 per share. That implies a multiple of two times estimated 2028 sales. That is cheap if the infrastructure holds. It is worthless if the power supply fails. He cites electricity and construction timelines as risks. Those are not hypothetical. They are physical limits. The direct-to-cell service with 650 satellites is just the tip of the spear. Regulators approved 15,000 more. That is a massive capex tailwind. But the cash generation must keep up. The Global X Space Tech ETF holds 25.44% in SpaceX. A 10% rally moves that fund by 2.54 points. The dependency is high. The risk is concentrated. Do not buy the stock for the rockets. Buy it for the data centers in orbit, if they ever turn a profit. Until then, it is a high-burn lottery ticket.
Author bio: Ethan Gallagher, a Silicon Valley Hardware Architect and Infrastructure Strategist with two decades of experience modeling capital-intensive technology buildouts and physical layer constraints.