Elon Musk's salary last year was $54,080. That's not a typo. The guy running one of the most consequential companies on the planet took home less than a first-year teacher in most U.S. states — at least on paper. That detail, buried inside SpaceX's freshly filed S-1 document, is just one of dozens of revelations from a company that spent 24 years keeping its books tightly sealed.
SpaceX filed for an IPO this week, targeting a public market debut as early as June 12. The nearly 400-page filing is the first time the company has ever disclosed detailed financials, and the picture it paints is both impressive and complicated.
On the revenue side, SpaceX pulled in $18.67 billion in 2025, a meaningful jump from $14 billion the year before. But here's the catch: the company swung to a $4.94 billion net loss after posting a small profit in 2024. The culprit is an aggressive — some might say reckless — bet on artificial intelligence. After merging with Musk's xAI earlier this year, SpaceX plowed billions into AI infrastructure, dragging down earnings even as its core rocket and Starlink businesses hummed along.
The filing's most eyebrow-raising claim is about market size. SpaceX says it has identified what it calls the largest total addressable market in human history — $28.5 trillion across space, data, and AI services. To put that in perspective, the entire U.S. GDP is roughly $29 trillion. Most of that projected opportunity, about $26.5 trillion worth, comes from AI, not rockets. Space and Starlink together account for only around $2 trillion of the estimate. Whether that math holds up to scrutiny is a different question, but it tells you a lot about where Musk thinks the company is headed.
For investors eyeing the IPO, the governance structure deserves attention. Post-listing, Musk will control 85.1 percent of combined voting power, effectively making him immovable as CEO and chairman. Public shareholders will have money in the game but almost no say in how the company is run. That arrangement is increasingly common among founder-led tech companies, but at this scale it carries real risk — especially given the filing's own acknowledgment that Musk's close ties to the Trump administration could materially affect the business depending on how the political winds shift.
Gwynne Shotwell, who has quietly kept SpaceX operational through every improbable milestone, received total compensation valued at $85.8 million in 2025, including stock awards. The contrast with Musk's nominal salary is stark and probably intentional — Musk has long structured his compensation around equity rather than cash, a strategy that legally minimizes his tax exposure while keeping him aligned with long-term performance.
The bottom line is that SpaceX is no longer just a rocket company. It is a sprawling conglomerate spanning launch services, satellite internet, AI compute, and now social media through xAI. Whether public markets will reward that complexity or punish it is the central question heading into June.
AI
Anthropic Commits $15 Billion to Elon Musk's SpaceX Data Centers
Anthropic is paying SpaceX $1.25 billion per month. Every month. Through May 2029. Let that sink in for a second. That single contract — for access to SpaceX's Colossus data centers in Memphis — is worth roughly $15 billion annually, which is nearly double SpaceX's entire 2025 revenue. It is, by almost any measure, one of the largest compute agreements ever disclosed publicly.
The deal came to light through SpaceX's IPO filing this week. Anthropic and SpaceX had announced the partnership earlier this month, but neither company disclosed the price tag until regulators required SpaceX to put it in writing. The filing confirms Anthropic gets access to Colossus I and Colossus II, SpaceX's AI training facilities, with fees slightly reduced during the ramp-up period covering this month and next.
There is a notable escape hatch built into the agreement: either party can walk away with 90 days' notice. That clause probably felt necessary given how fast things move in AI right now. Models that are state-of-the-art today can become obsolete in under a year, and locking yourself into a multi-billion-dollar infrastructure commitment with no exit would be a serious risk for a company like Anthropic. Still, the fact that they signed at all is a signal of just how desperate AI labs have become for raw compute capacity.
The relationship is also ideologically awkward in ways worth noting. Anthropic's Claude competes directly with Grok, the AI product from Musk's xAI — the same xAI that merged with SpaceX earlier this year. Anthropic is essentially paying its competitor's parent company to survive. That kind of arrangement only happens when there are no good alternatives, which tells you everything about the current state of AI infrastructure. Data center buildouts are running into community opposition, energy grid constraints, and long equipment lead times. Demand is massively outpacing supply.
For SpaceX, the deal is a financial lifeline for its AI division, which lost $6.3 billion on just $3.2 billion in revenue in 2025. The company spent $12.7 billion on AI capital expenditures last year alone — about 61 cents of every dollar it invested. In the first quarter of 2026, SpaceX spent $7.7 billion on AI infrastructure versus $1 billion on its actual space business. The rocket company has, for now, become a data center company that also launches rockets.
Musk posted on X that SpaceX is open to cutting similar compute deals with other AI companies, positioning Colossus as an open marketplace for training capacity. If additional labs sign on, the revenue potential is enormous — though it also deepens the irony of AI competitors funding each other's infrastructure.
AnthropicFor its part, Anthropic appears to be on solid footing despite writing these enormous checks. The company is reportedly on track for its first quarterly operating profit, with annual revenue expected to exceed $10.9 billion — more than double its March quarter haul. Apparently, betting $15 billion a year on compute is something you can afford when your own growth is moving that fast.
Source: The Verge
The deal came to light through SpaceX's IPO filing this week. Anthropic and SpaceX had announced the partnership earlier this month, but neither company disclosed the price tag until regulators required SpaceX to put it in writing. The filing confirms Anthropic gets access to Colossus I and Colossus II, SpaceX's AI training facilities, with fees slightly reduced during the ramp-up period covering this month and next.
There is a notable escape hatch built into the agreement: either party can walk away with 90 days' notice. That clause probably felt necessary given how fast things move in AI right now. Models that are state-of-the-art today can become obsolete in under a year, and locking yourself into a multi-billion-dollar infrastructure commitment with no exit would be a serious risk for a company like Anthropic. Still, the fact that they signed at all is a signal of just how desperate AI labs have become for raw compute capacity.
The relationship is also ideologically awkward in ways worth noting. Anthropic's Claude competes directly with Grok, the AI product from Musk's xAI — the same xAI that merged with SpaceX earlier this year. Anthropic is essentially paying its competitor's parent company to survive. That kind of arrangement only happens when there are no good alternatives, which tells you everything about the current state of AI infrastructure. Data center buildouts are running into community opposition, energy grid constraints, and long equipment lead times. Demand is massively outpacing supply.
For SpaceX, the deal is a financial lifeline for its AI division, which lost $6.3 billion on just $3.2 billion in revenue in 2025. The company spent $12.7 billion on AI capital expenditures last year alone — about 61 cents of every dollar it invested. In the first quarter of 2026, SpaceX spent $7.7 billion on AI infrastructure versus $1 billion on its actual space business. The rocket company has, for now, become a data center company that also launches rockets.
Musk posted on X that SpaceX is open to cutting similar compute deals with other AI companies, positioning Colossus as an open marketplace for training capacity. If additional labs sign on, the revenue potential is enormous — though it also deepens the irony of AI competitors funding each other's infrastructure.
AnthropicFor its part, Anthropic appears to be on solid footing despite writing these enormous checks. The company is reportedly on track for its first quarterly operating profit, with annual revenue expected to exceed $10.9 billion — more than double its March quarter haul. Apparently, betting $15 billion a year on compute is something you can afford when your own growth is moving that fast.
Enjoyed this?
Get stories like this delivered every Tuesday — free.