At the Baron Investment Conference in New York on November 6, Elon Musk put numbers on SpaceX's trajectory: the company is tracking toward an annualized revenue rate above $100 billion in December, could reach $500 billion by the end of 2027, and has roughly a 2% chance of hitting $1 trillion.

He framed the long-range figures as "a best guess, not a promise."

How steep is the curve

By his account, revenue grew more than 80% quarter over quarter from the second to the third quarter, with the fourth quarter expected to add another 70%. The company's third-quarter results showed revenue up 161% year over year, with AI-related revenue up about ninefold.

Context matters here. SpaceX booked roughly $15-16 billion in revenue last year. An annualized rate above $100 billion in December would mean more than a sixfold expansion in under two years.

Who is paying: from government to Anthropic

The mix has shifted more than the growth rate suggests.

Government work used to dominate. It is now below 10% of quarterly revenue, down from 16.6%. Commercial customers have filled the gap, and the largest single one is Anthropic, which reportedly accounts for more than a third of quarterly revenue under an agreement worth up to $120.5 billion through 2029.

In other words, SpaceX increasingly sells compute, bundled with launch and connectivity. That also explains why it is building its own power plant and its own fabs: customers sign multi-year compute contracts, and any weak link in supply becomes a delivery risk.

Two readings of the valuation

The bull case is orbital compute. Starlink provides global connectivity and edge nodes, Starship lifts payloads, and space-based data centers can use solar power and radiative cooling, potentially at a lower long-run cost per unit of compute. SpaceX listed in June at about $1.25 trillion and now trades above $1.9 trillion.

The bear case has three planks: customer concentration (Anthropic alone is a third of the quarter), guidance that rests on guesses, and enormous capital spending with long payback periods. Two sessions after the midterms, SpaceX and Tesla together shed more than $450 billion in market value, and SpaceX's five-year credit default swaps widened 24.5 basis points to 286.56, moving opposite to hyperscaler CDS. Credit markets registered doubt before equity markets did.

Three metrics to watch

First, actual drawdown on the Anthropic contract. $120.5 billion is a ceiling, not a commitment; quarterly consumption is the real revenue.

Second, Starlink ARPU. Subscriber growth is easy to show; holding the average revenue per user is what decides whether this is a utility or a growth business.

Third, the gap between capex and free cash flow. A $2.4 billion power plant, two Terafab fabs and continuing Starship iteration all consume cash at once, which makes financing cadence the key variable for next year.

The valuation story has moved from launch services and satellite broadband to AI compute supplier, and the move has worked. But concentration and capital intensity came with it, and those are the two questions the next earnings call will press hardest.