You are reading the wrong beat. SpaceX closed its first quarterly report as a public company with $7.8 billion in revenue, a beat against Wall Street’s $6.81 billion forecast, and a share price that first climbed 9.43% to $125.33, then gave back more than 8% in after-hours trading. The market was not reacting to the revenue number. It was reacting to the balance sheet. Digital assets fell from $1.637 billion to $1.098 billion over six months. That is a $539 million mark-to-market markdown, and it lands like a cold block on a quarter otherwise full of launch-flavored warmth. The ledger remembers what the mempool forgets. The after-hours slide is the first honest sentence the market has produced today.
SpaceX is no longer a private company story. The August 4, 2026 print is its first as a listed company, and the operating results are strong. Connectivity revenue reached $4.291 billion, up 66% from a year earlier. Operating income for that unit climbed 79% to $1.656 billion. Starlink subscribers doubled over 12 months to 12 million, with average revenue per user unchanged at $66. The artificial intelligence segment brought in $2.561 billion, a 247% annual increase, supported by $14.1 billion in contracted cloud services agreements. AI operating losses narrowed to $1.257 billion, roughly half of the $2.39 billion analysts had modeled. Loss per share landed at $0.09 against a $0.24 expected loss. Adjusted EBITDA rose 191% to $3.538 billion, while analysts had tracked scenarios near $2 billion.
Those are real numbers. Management did not need permission to celebrate. But the balance sheet contains a second dataset. Digital assets fell 33% over six months. The company did not break out coin counts in the release. Grayscale’s tracking puts the stack at 18,712 BTC, which would make SpaceX the largest diversified public holder of bitcoin. Against that count, the $1.098 billion carrying value implies roughly $58,700 per coin. Bitcoin traded near $64,073 on Tuesday. The accounting gap matters because the crypto line is the only part of this earnings call that cannot be polished by an investor-relations slide. Code is not law, it is merely preference; accounting is the same when disclosure is thin.
Let me isolate the forensic pieces, because a press release does what press releases always do: it points at the stage while ignoring the floor. The floor here is made of two lines. Bitcoin is one of them. Capital expenditure is the other.
The bitcoin line. The $539 million decline can be explained without a single sale. If SpaceX ended December with $1.637 billion and held 18,712 BTC per Grayscale, the implied carrying value was near $87,500 per coin. On June 30, with $1.098 billion on the books, the implied carrying value falls to roughly $58,700. Bitcoin spent the first half of 2026 sliding and bouncing; a mark-to-market loss is mechanically consistent with unchanged holdings. That is the good news. The bad news is that the entire crypto position is now worth less than one month of SpaceX’s own capital expenditures. The $88 bitcoin transfer in July drew all the attention because the industry loves a whale waking up. But $88 is not a strategy. It is dust being tested. It tells me the custody architecture still exists and that the treasury rules engine is alive. It does not tell me whether SpaceX will hold through the next cycle. I spent three weeks in 2017 auditing an ICO contract that was supposed to be immutable until a reentrancy bug surfaced. The lesson stuck: believe the transaction log, not the announcement.
The transaction log here says BTC holdings are falling in dollar terms while the coin count remains undetermined. That ambiguity is a risk. If the company holds 18,712 coins and never sells, the current price matters only as noise. If it sold, the market deserves to know what cost basis triggered the sale. Without that disclosure, the only rational assumption is a range of possible coin counts and a range of possible treasury strategies. Truth is a derivative of transparent data. This release gives us an aggregate number, not the underlying log.
The capex line. The capital expenditure figure is the more brutal piece of arithmetic. Q2 capital expenditure hit $18.369 billion. Of that, $15.828 billion went to the AI segment. Compute capacity expanded to 1.4 gigawatts from 1.0 gigawatt in the first quarter. SpaceX closed June with $100 billion in cash and securities, plus $47.5 billion in backlog. That works out to roughly five quarters of current capex in cash, and no formal guidance on when the burn will accelerate. The $60 billion agreement to buy Cursor, an AI coding tool, adds another layer that cannot be squared with a light touch. You cannot combine a $60 billion acquisition, a $15.8 billion quarterly AI build-out, and a 191% EBITDA beat without asking a simple question: what is the funding plan? The IPO was not enough to answer that. There is no capital return program, no range of earnings expectations, and no statement about the next funding layer. There is just a tweet framed around extreme vertical integration. From my experience auditing blockchain project ledgers, I can name that pattern immediately. A founder who says “we are vertically integrated” usually means “we are spending on everything at once and hoping the revenue curve catches up.” Sometimes it does. But it is a bet, not an operating model.
Space’s legacy rocket business is not helping the discipline. Revenue rose 29% to $962 million, yet the unit widened its operating loss to $542 million on Starship research spending. The company is burning cash in orbit, on the ground, and inside its AI data centers. The only reason the quarter still looks elegant is because the top line is growing quickly enough to obscure the bottom line’s shape. But investors are not blind. After-hours trading does not read tweets; it reads cash flow projections. The drop is the market telling SpaceX that revenue beats no longer justify capex surprises.
What the bulls got right. The bearish parse is incomplete without acknowledging the operational substance. SpaceX is not a token project with a whitepaper and a matching T-shirt drop. It has actual satellites, actual subscribers, actual compute, and actual contracts. Starlink doubled subscribers to 12 million. AI revenue nearly tripled. Adjusted EBITDA nearly tripled. The $14.1 billion in contracted cloud services and the $47.5 billion backlog provide visibility that most public technology companies would envy. If the question is whether SpaceX can sell, the answer is yes. If the question is whether the digital asset markdown is a crisis, the answer is probably no. The crypto line is roughly 1% of the cash balance; a $539 million swing is a footnote for a company with $100 billion in cash. The bull case correctly points out that a falling Bitcoin price is not evidence of a bad treasury decision. Holding through the cycle is a design choice. Immutability is a feature, not a virtue; surviving the drawdown is a feature, not a strategy. The two positions can coexist.
What bulls cannot explain is the funding roadmap. The company closed the quarter with $100 billion in cash, but it is spending $18.4 billion per quarter. At that pace, the war chest covers about five quarters. The Cursor deal adds $60 billion to the obligation side. Maybe Cursor accelerates AI revenue. Maybe it accelerates AI capex. Both can be true. The after-hours decline is not a rejection of the quarter; it is a demand for a clear statement on where the next $100 billion goes. That cannot be solved with a mathematical proof or a crypto surplus. It can only be solved with a capital allocation plan.
The takeaway is uncomfortable because the quarter itself is good. SpaceX beat estimates, grew subscribers, narrowed AI losses, and still lost the market’s attention in after-hours trading. The market is no longer treating Bitcoin as the primary risk. It is treating capital intensity as the primary risk. SpaceX has no formal guidance. No formal guidance is itself guidance. The ledger remembers what the mempool forgets, and the ledger says the next bull run for this stock will be funded with clarity or not at all. The stock now depends on one deliverable: a roadmap for turning $18 billion per quarter of capex into enough revenue to make that number feel planned rather than desperate. If management cannot provide that, after-hours traders will keep making the first correction.