AivexaNewsSearch
AI news for builders and product teamsChecked every hour

SpaceX’s AI business is beating expectations

Collected Oct 20, 2026

SpaceX reported third-quarter revenue of $14.6 billion and net income of $4.4 billion, according to a regulatory filing disclosed on Tuesday, reversing a $1 billion loss in the same quarter a year earlier. The company credited growth across all segments, but the standout was its renamed Super Intelligence business line, which brought in $7.5 billion, or 52% of income. That unit comprises the Grok bot, the recently acquired Cursor coding tool, and a neocloud compute business now reported at 1.7 GW of capacity.

The AI segment produced $431 million in profit. That figure would have been higher without $1.2 billion in one-time stock-based compensation tied to the Cursor acquisition, which is a non-cash charge that depresses reported operating results in the quarter the deal closes. Investors typically look past such charges when judging underlying performance, though the size of the number here is large relative to the segment's profit.

Revenue also got a lift of roughly $1 billion from the recovery of Bitcoin prices and unspecified related-party income that the report suggests may be tied to a SpaceX investment in another company. Those are not operating gains from the core businesses, so the quarter's top line is somewhat flattered by items that may not recur. For the first nine months of the year, SpaceX still shows a net loss of $413 million, meaning the third quarter alone was not enough to put the full year into the black.

Capital spending is running well ahead of profit. Total capital expenditures came in at $19.6 billion as SpaceX builds out a second data center project, Colossus II, and keeps funding Starship, the heavy-lift rocket it views as central to deploying a new generation of Starlink satellites and eventually to orbital data centers. CFO Bret Johnsen said new capital investments in the fourth quarter would run between $35 billion and $40 billion, a step up that signals the company expects to keep spending heavily on infrastructure.

On a livestreamed call with investors, CEO Elon Musk said he is spending four days a week in Memphis, Tennessee, where Colossus II is sited, to bring the data center online. He argued that the binding constraint on deploying advanced AI is now power, cooling and infrastructure rather than chips. This is a view increasingly voiced across the industry: GPU supply has loosened in relative terms, while grid connections, cooling capacity and turbine lead times have become the harder bottlenecks. Musk said he had assigned a team of SpaceX engineers to the problem and claimed the company can deploy power and cooling infrastructure faster than others.

He offered a concrete example: makers of the turbines that power data centers face 12- to 18-month delays in obtaining the metals used for blades and vanes. SpaceX is now applying metal casting capabilities developed for its rocket work to produce those components and get turbines sooner. The logic is that propulsion hardware and turbine hot sections share metallurgy and precision casting demands, so tooling built for one can serve the other. Whether that transfers at commercial scale is an open question, and the claim comes from the company rather than an independent assessment.

On the launch side, Musk said the next Starship test flight, expected in November, will attempt to recover both the first-stage booster and the second stage. Recovering both halves of the vehicle would be a significant engineering milestone, he said, though he framed reflying both stages as the milestone that actually matters. Reuse is the economic core of the Starship program; catching hardware is only useful if it can be flown again cheaply. Starlink, meanwhile, reached 13.6 million consumer subscribers, and Musk predicted that growth in AI agents would drive additional demand for connectivity.

Musk was aggressive in his forward projections, saying SpaceX's annual recurring revenue by the end of 2027 would land somewhere between $500 billion and $1 trillion. Johnsen gave a narrower figure, forecasting total revenue of $33 billion to $37 billion in the first quarter of 2027. The gap between the two numbers is enormous, and the CFO's guidance is the more grounded of the pair; readers should treat the CEO's recurring-revenue target as aspiration rather than forecast.

Why it matters: The quarter shows AI compute and agent products now carry a company long identified with rockets, and that the neocloud model can be profitable at scale even while absorbing acquisition charges. For developers and product teams, a large new supplier of compute with 1.7 GW online and aggressive expansion plans could affect pricing and availability of inference capacity, though the heavy fourth-quarter spend suggests the buildout depends on continued capital access. The November Starship attempt to recover both stages, if successful, is worth watching as the next marker for launch economics.

Read at TechCrunch · AI

Based on reporting from the original publisher. Visit the source for full context and later updates.

Publisher excerpt

SpaceX hit profitability in the third quarter, with net income of $4.4 billion. AI was the engine.