Famously secret about its finances, SpaceX opens its books for the first time

After operating as a private company for nearly a quarter century, keeping its financial accounts closely guarded, SpaceX released a detailed accounting of its business on Wednesday afternoon in a nearly 400-page S-1 filing with the U.S. Securities and Exchange Commission.

SpaceX, founded in 2002 and still led by Elon Musk, submitted the filing in anticipation of an initial public offering of its stock on June 12.

The document didn’t reveal any big surprises about the company’s space operations, but there were a bunch of details about its vast operations, which now include launches, space flight, space-based internet, and thanks to Musk’s recent acquisition of XAI, social media and AI.

The company projects revenue of $18.67 billion in 2025, up significantly from $14.02 billion a year earlier. However, after making a small profit in 2024, the company is set to lose $4.94 billion in 2025, largely due to spending on artificial intelligence development.

That’s a big market you got there

SpaceX estimates a “total addressable market” or TAM of $28.5 trillion for its current and future offerings in space, data and AI services. However, only about $2 trillion of this amount is directly related to space or the company’s Starlink network. The remaining $26.5 trillion is believed to come from AI, primarily from enterprise applications.

“We believe we have identified the largest TAM in human history,” the company says on page 171 of the filing. “We believe our next trillion-dollar market is AI compute, which we think will be leveraged across our rockets and satellites for large-scale orbital deployments.”

The company said its estimate for this large market was based on multiple sources.

“Our AI market projections are based in part on estimates of global data center compute demand from third-party sources, including estimates published by the RAND Corporation, as well as internal assumptions about the share of global compute capacity that can be used for AI workloads and other operational assumptions such as power usage, usage rates, and pricing,” the filing said.