SpaceX Beat Earnings and the Stock Still Crashed - Because the Real Problem Wasn't the Quarter

SpaceX reported its first earnings as a public company on Tuesday and topped analysts' expectations across every major line. Revenue jumped 92% year over year to $7.8 billion, adjusted EBITDA - a rough proxy for cash earnings before the distortions of interest, taxes, and depreciation - tripled to $3.5 billion, and the net loss narrowed to $541 million, roughly half what the market feared.

And yet the stock fell over 10% from Tuesday's close through Wednesday's pre-market, sliding to around $114 from a $225.64 all-time high earlier in June. That's a drawdown of roughly 50% from peak, with the company's market cap having shed close to $600 billion.

I must admit that I've been baffled by this reaction. Not because the results were flawless, but because the market is punishing SpaceX for a problem that was already baked into the stock since day one of its public life.

What the earnings actually showed

Revenue beat was broad-based. Starlink, the satellite-internet crown jewel, brought in $4.3 billion, up 66% year over year, with 12 million subscribers globally and 1.7 million net new sign-ups in the quarter alone. Average revenue per user (ARPU, or the monthly revenue earned per subscriber) held steady at $66, easing concerns that international expansion and lower-priced plans would erode the unit's economics.

The AI segment - which includes xAI, Grok, and X's advertising platform - more than tripled year over year to $2.6 billion. Management said the company now has $14.1 billion in contracted cloud services sales and expects to reach $100 billion in annualized revenue run rate by December. That's a $100B ARR figure that includes the pending $60 billion Cursor acquisition, which Musk said is expected to close in the third quarter pending regulatory approvals.

The space launch segment, which includes Falcon 9 and Starship development, generated $962 million, up 29% year over year. SpaceX completed 78 launches year to date and delivered 1,041 tonnes to orbit.

On every revenue line, SpaceX beat. On every profit line, the company also beat. The AI unit's operating loss came in at $1.26 billion versus consensus expectations of a $2.39 billion loss. That matters because the AI segment is the one dragging the company's overall profitability down.

So why is the stock falling?

Capital expenditures. SpaceX spent $18.4 billion in the second quarter on building out its infrastructure, with $15.8 billion going directly to AI compute capacity. Analysts had expected roughly $13.1 billion. The gap between what the market modeled and what SpaceX actually spent was nearly $3 billion in a single quarter.

That's not the whole story, though. The TTM (trailing twelve-month) picture is even starker: $43.3 billion in capex over the past year, against $9.9 billion in operating cash flow. Free cash flow - operating cash flow minus capital spending, the money that actually flows back to investors after the company reinvests - stands at negative $33.4 billion. The FCF margin is negative 145%, meaning the company's free cash flow is negative by an amount equal to 145% of its revenue.

That's the number that has investors spooked. SpaceX is burning cash at a rate that would make a venture-capital-backed startup nervous, only this one happens to carry a $1.5 trillion market cap.

But here's what the market reaction ignores

The capex problem wasn't revealed on Tuesday. It was the premise the stock traded on from the moment it opened at $150 in June. The IPO prospectus itself disclosed an operating loss of $4.2 billion in 2025, driven primarily by AI infrastructure investment. The take-it-or-leave-it IPO price of $135 was set by Elon Musk himself, with no traditional roadshow price range, and the stock's subsequent climb to $225 was driven by AI fervor, not an expectation that the AI business would be profitable tomorrow.

JPMorgan analyst Doug Anmuth projected capex of nearly $200 billion in both 2027 and 2028. That's the consensus view already. The market has been pricing in a multi-year cash burn and a path to profitability that doesn't arrive until at least 2028 or 2029. Tuesday's capex number was higher than the Q2 estimate, but it doesn't change the multi-year framework.

What does change the picture is whether the spending generates returns. Management said the company is seeing less than a one-year payback period on new AI compute deployments. If that holds, the capex intensity is aggressive but not necessarily wasteful. Starlink remains the cash engine, with operating income of $1.7 billion in the quarter, and the company ended with $100 billion in cash and marketable securities and a $47.5 billion backlog.

The real overhang isn't capex - it's the lockup

The market is using capex as the narrative excuse to sell, but the mechanical reality is more prosaic. On Thursday, August 6th, the post-IPO lockup period expires, freeing up to 20% of shares for sale. As Epistrophy Capital's Cory Johnson noted in a client memo, the unlock releases "hundreds of millions of insider shares - roughly triple the current tradable float".

That's the supply shock that the stock has been working itself down to since June. The earnings report simply provided the catalyst. The stock had already shed approximately 30% from its $150 IPO debut before Tuesday's results, closing at $108.37 on Friday July 31st, well before the capex number was even known.

Is this a falling knife or a falling opportunity?

The fundamental gates are mixed. Revenue growth at 92% is exceptional. Starlink's subscriber momentum is real. The AI compute business is scaling - 1.4 gigawatts of compute capacity at quarter end, up from 400 megawatts a year ago - and it already generated positive adjusted EBITDA of $1.1 billion despite the massive operating loss.

But the valuation is still extraordinary even at today's depressed levels. At $114, the stock trades at roughly 65 times trailing sales, with a negative P/E and negative free cash flow. There is no traditional earnings multiple to anchor the stock. The $1 trillion revenue target Musk pulled forward to 2030 is ambitious - S&P Global's Melissa Otto called it "possible" but "a show-me story". The TAM that SpaceX laid out in its prospectus - $28.5 trillion across broadband, mobile, advertising, AI infrastructure, and enterprise - is eye-popping but not yet reflected in the $18.7 billion of revenue the company actually generated last year.

The moat question is worth examining. Starlink's orbiting satellite constellation, government launch monopoly status, and exclusive Nvidia hardware partnership create real barriers to entry. No competitor can replicate the vertical integration of launch, satellite deployment, and compute infrastructure. Starship V3 completed two successful flights with the heat shield problem reportedly solved, which could enable rapid reusability and a step-change in launch economics.

The posture

I'm not in a hurry to buy here. The stock has fallen roughly 50% from its peak, and the RSI of 39.6 suggests the selling hasn't quite reached exhaustion territory. The lockup expiration on Thursday adds near-term supply risk that price action needs to absorb first. The MACD remains deeply negative at -10.2, confirming the downtrend is still active.

What I am watching is whether the stock can stabilize near the $105-to-$110 zone after the lockup flushes through. If the supply overhang clears and the stock holds, the risk/reward becomes more compelling. A company growing revenue at 92%, with Starlink as a profitable engine, $100 billion in cash, and a path to $100 billion ARR by year end could reasonably trade above where it sits now.

I would reassess if Starlink ARPU declines below $60 per month, if AI capex exceeds $20 billion in a quarter without corresponding revenue growth, or if the lockup sell-off pushes the stock below $100 with no sign of demand. Those would be thesis-breaking signals, not just price noise.

For now, the setup isn't a double-down moment. It's a watch zone. The fundamental story hasn't broken, but the mechanical overhang hasn't cleared either. Patience may be the better play than conviction at these levels.