Here's the thing about SpaceX's first-ever earnings call: it was less a financial briefing and more a live demonstration of what happens when the founder has no PR handler and a microphone.
The pattern was consistent enough to be its own product feature. Chief financial officer Bret Johnsen would carefully build a hedged, legally defensible claim — the kind that takes a team of lawyers and a sleepless IR director to construct — and then Elon Musk would walk in twenty minutes later and detonate it.
Johnsen's prepared remarks landed a specific, qualified projection: SpaceX's cloud services business, which rents compute capacity to AI companies, had already contracted '$6.7 billion of cloud services revenue over a six-month period' in the first weeks of Q3 alone. He said the company believed it was 'on a trajectory' — including contribution from Cursor — 'to reach $100 billion of ARR... by the end of this year, based on our expected revenue in the month of December.'
Musk's version: 'The $100 billion ARR in December is not a question mark. That's what we would achieve if we basically did nothing. I think it may be higher than that. It probably will be higher than that.'
The compute business is real. The less-than-one-year payback on new capital deployments Johnsen cited is a genuinely striking number. But Musk's habit of treating a CFO's careful forward guidance as a floor to be immediately raised is, at minimum, the kind of thing that makes securities lawyers age in real time.
Musk also updated SpaceX's internal revenue projections on the fly, announcing that the company's target of reaching $1 trillion in revenue — not ARR, he clarified — had moved from 2031 to 2030, with 'a non-zero chance' of hitting it in 2029. The trillion-dollar figure had appeared in IPO documents filed two months prior.
On Starlink, Musk predicted the service would 'deliver a majority of the world's internet' in 'less than 10 years.' Chief operating officer Gwynne Shotwell, speaking minutes later, said Starlink would represent 'a significant portion of global internet traffic in the years ahead.' Same direction, much smaller blast radius.
Shotwell also dropped the call's most consequential competitive signal: SpaceX intends to build a terrestrial mobile network using the EchoStar spectrum it acquired, potentially deploying femtocells mounted on existing Starlink dishes. 'I anticipate us to be able to acquire quite a few of their customers,' she said of T-Mobile, AT&T, and Verizon subscribers. Shotwell pegged the US wireless market at around $600 billion annually.
On Starship, Musk declared the heat shield 'problem solved' before the most recent test vehicle had even been recovered from the Indian Ocean. Shotwell's stated goal: 'boots on the moon in 2028.'
The demo is not the product. But the compute revenue numbers Johnsen cited are not vaporware, the EchoStar spectrum is real, and a newly public SpaceX with genuine cash flows and a founder willing to say anything on a recorded call is exactly the kind of disruptive variable that makes incumbent telcos and hyperscalers nervous. The benchmarks Musk picked are generous. The underlying business, for once, might not need them.
