Our analysts just identified a stock with the potential to be the next Nvidia. Tell us how you invest and we'll show you why it's our #1 pick. Tap here.

SpaceX finally had to show its homework, and the homework beat… but not if you keep looking at the numbers with any kind of insight.

Revenue landed at $7.81 billion against the $6.93 billion the Street penciled in per LSEG, up 92%. Loss per share came to 9 cents where analysts modeled the mid-20s. Operating loss shrank to $143 million from $1.94 billion in Q1 and $970 million a year ago, somehow spitting distance from breakeven for a company that lost $4.9 billion last year.

But let's peruse the addbacks before popping any champagne. That $3.54 billion of adjusted EBITDA is the same $143 million operating loss with $2.85 billion of depreciation and $831 million of stock comp piled back on top. Adjusted EBITDA measures earnings before subtracting the cost of what you bought, and SpaceX bought a staggering amount.

The AI segment shows the trick best. Its first profitable quarter, $1.15 billion of adjusted EBITDA, starts as a $1.26 billion operating loss and turns positive only once $1.89 billion of depreciation gets added back. Depreciation spreads a $23.6 billion shopping spree across the years those machines stay useful, so the profit holds up only if you ignore what the machines cost. And with Colossus II, the next data center, still going up, you need to ignore quite a lot.

SpaceX also spent $18.4 billion in 3 months, or $2.35 of capex for every $1 of revenue and better than 5x the adjusted EBITDA it just advertised. AI took $15.8 billion of it. Analysts penciled in $48.7 billion for the year and the company burned $28.5 billion in 6 months, so either the back half slows to a crawl or that estimate gets torn up. Doing that while crawling through a $60 billion acquisition would be a very neat trick.

Starlink clearly funds this whole opera. Connectivity turned $4.29 billion of revenue into $1.66 billion of operating income, the only segment in the black. Subscribers doubled to 12 million while ARPU slid to $66 from $85, the price of finding your next 6 million customers in places that can't pay New York rates.

One stock. Nvidia-level potential. 30M+ investors trust Moby to find it first. Get the pick. Tap here.

The rockets did less. Space revenue was $962 million, and the 6-month figure shrank to $1.58 billion. Launches fell to 38 from 46, mass to orbit to 485 tons from 652. Segment R&D of $1.08 billion exceeded segment revenue.

Related-party debt tripled to $13.3 billion from $4.5 billion at year-end, and SpaceX paid $327 million in related-party interest against zero a year ago. Somebody is collecting a handsome coupon and the release won't say who (we can guess). Six-month other expense hit $1.96 billion with only $86 million landing in Q2, so something unpleasant happened in Q1 that still has no explanation, possibly the $539 million evaporation in digital assets.

That 9-cent loss rests on 5.86 billion weighted average shares against 13.18 billion Class A and B issued, so Q3 per-share math gets worse on identical performance. Six-month loss attributable to shareholders was $5.49 billion against a $4.82 billion net loss, a $671 million gap of preferred accretion IPO buyers swallowed up front.

The Cursor deal is stock-for-stock, so that $60 billion is essentially dilution.

Elon Musk's take is measurable. His roughly 5 billion shares were worth $1.13 trillion at the June 16 peak and $573 billion Monday, a $556 billion paper haircut in 7 weeks. The Mars grant adds 1 billion more against milestones SpaceX calls improbable, so it costs the income statement nothing.

SpaceX's pre-IPo lockup expires Thursday. That should be very very interesting.