SpaceX is going through a quiet but major shift. Its space and satellite businesses still define the brand, but the numbers now tell a different story. The company is using profits from Starlink to fund a fast-growing and very expensive push into artificial intelligence.
This shift matters because it changes how you should think about SpaceX. It is no longer just a space company. It is becoming an AI company that happens to launch rockets.
Right now, Starlink is doing almost all the heavy lifting.
In 2025, it generated about $11.4 billion in revenue. That is roughly 60% of SpaceX’s total. Growth is still strong at around 50% year over year.
More important is profitability. Starlink runs at high margins, with adjusted EBITDA near 63%. After spending on satellites and infrastructure, it still produces about $3 billion in free cash flow.
That makes it the only part of SpaceX that consistently generates real cash.
In simple terms: Starlink is funding everything else. The launch business tells a different story.
SpaceX 2025: The Rocket Engine Fueling an AI Burn
SpaceX brought in about $4.1 billion from launches in 2025. Growth slowed to around 8%. Free cash flow sits near negative $3 billion.
The reason is structural. A large share of launches are for Starlink satellites. These are internal missions priced lower than commercial launches.
So while launch volume is high, margins stay weak. This means rockets support Starlink’s expansion, but they do not yet support themselves.
The most striking numbers come from SpaceX’s AI segment. This includes xAI and X. Together, they generated about $3.2 billion in revenue in 2025. Growth is steady but not explosive.
Losses, however, are massive. The AI unit burned close to $14 billion in one year. That is more than the estimated losses at OpenAI and Anthropic combined.
This places SpaceX among the most aggressive spenders in AI today. The spending goes into data centers, chips, and platform infrastructure. These are long-term bets, not short-term returns.
The shift becomes even clearer when you look at capital spending.
SpaceX’s High-Stakes Pivot to AI-First Infrastructure
In 2025, SpaceX spent about $20.7 billion on assets. Around 60% of that went into AI. That means AI now absorbs more capital than rockets and satellites combined. This is not a side project. It is the main investment focus.
The company is building AI capacity at a pace closer to a startup than a mature infrastructure firm.
SpaceX’s IPO pitch reflects this shift.
The company presents itself as “AI-first”, even though the public still links it to space exploration. This creates tension for investors. On one hand, Starlink generates strong cash flow. On the other hand, AI is consuming that cash at a faster rate.
The key risk is simple: can Starlink keep up? If growth slows or costs rise, the funding model weakens. That could force cuts in AI expansion, acquisitions, or even space programmes.
SpaceX also faces a structural disadvantage. Companies like Alphabet and Microsoft fund AI with large, diversified businesses. They have cloud, ads, and enterprise software.
SpaceX relies heavily on one engine: Starlink. That concentration raises risk. If one pillar weakens, the whole system feels it.
The core idea is straightforward. Starlink generates cash. Rockets consume some of it. AI consumes most of it. Right now, the model works because Starlink grows fast and stays profitable. But the balance is tight.
SpaceX’s High-Stakes AI Gamble
If AI spending stays high and Starlink growth slows, SpaceX may need outside funding, slower expansion, or sharper priorities. At the same time, if the AI bet pays off, SpaceX could move into a new tier of tech companies.
That is the trade-off. It is a high-risk, high-scale strategy built on one strong cash engine. And for now, that engine is still running.
