SpaceX Q2 Earnings: The Beat That Cost $18 Billion

A week ago we published a piece arguing that two dates would decide August for this stock, and that neither of them had anything to do with rockets. The first of those dates has now passed.

SpaceX beat on revenue. It beat on EBITDA. It beat on the loss line. Every segment came in ahead of what analysts had modelled. And the stock, which had rallied more than 9% into the close in anticipation, promptly fell in after-hours trading.

If that sequence confuses you, you are reading the wrong line of the release. The story is not in the revenue. It is in the line that says the company spent $18.37 billion in ninety days to generate $7.81 billion of sales.

What SpaceX Actually Reported

The headline numbers, and how they compared to what the street expected:

Put the first and last of those side by side and you have the entire trading story. Every dollar of revenue this quarter was accompanied by roughly $2.35 of capital expenditure. In the AI segment specifically, the company spent about six dollars for every dollar it billed.

For scale: SpaceX spent more on capex in this single quarter than it did in the whole of 2025, when the figure was around $20.7 billion. Half of one year’s revenue, spent in three months, on things that do not yet exist.

One Profitable Segment, Two That Are Not

The segment breakdown is where the beat gets complicated. All three units exceeded expectations on the top line. Only one of them made money.

Add the three operating lines together and the segments net out to a loss of roughly $142 million. That is close enough to breakeven that, at the operating level, this quarter was genuinely a turning point.

The problem is what sits underneath. Starlink is now funding a launch division that loses half a billion a quarter and an AI division that is consuming capital at a rate no satellite broadband business can indefinitely sustain. The bull case is that this is exactly what building looks like. The bear case is that the building never stops.

Why the Stock Rose 9% and Then Fell

Most coverage published in the last few hours says SpaceX stock sank on earnings. That is not quite what happened, and the distinction matters if you trade around events.

SPCX spent Tuesday’s regular session rallying, closing near $125.33, up more than 9% on the day, extending a recovery off a recent low around $105. That is a market pricing in a good print. When the good print arrived, the shares fell.

This is the classic event-driven sequence and it catches new traders every quarter. Anticipation is a position. By the time the number is public, the people who wanted to own it into the print already do, and the only remaining decision is whether to keep holding. On Tuesday evening, enough of them decided not to.

What they were looking at was not the revenue line. It was $18.37 billion of capex, a launch business still losing money, and a supply event forty-eight hours away.

Thursday: 911.5 Million Shares

This is the part we flagged a week ago in our breakdown of the SPCX lock-up calendar, and nothing about Tuesday’s numbers changes it.

On 6 August 2026, roughly 911.5 million previously restricted insider and employee shares become eligible for sale. Cory Johnson of Epistrophy Capital characterised that release as roughly triple the current tradable float, freeing up to 20% of the company’s shares.

At Tuesday’s close near $125, that block is worth approximately $114 billion of stock. Nobody knows how much of it will actually be sold. Employees who have held illiquid paper for a decade have reasons to take some off. Long-term believers have reasons to hold. Both will be true simultaneously.

What you can say with confidence is this: a stock trading below its $135 offer price, in a drawdown from its June high, meets the largest single supply event in its short public life two days after its first earnings report. That is not a prediction of direction. It is a statement about volatility, and it should be reflected in your position size rather than your opinion.

One quieter detail worth noting. A separate conditional tranche only releases early if the stock trades roughly 30% above the IPO price, near $175.50. At $125 that trigger remains dormant. Weakness in the shares is, perversely, keeping some supply locked away.

The Nvidia Deal and the Tesla Question

Two items landed alongside the numbers, and both feed the same debate.

First, SpaceX announced a partnership with Nvidia to design its Starmind AI-1 payload, intended to put what the company calls datacenter-class compute into orbit. The design uses Nvidia’s Rubin GPUs and Vera CPUs and would lift SpaceX’s peak satellite computing capacity to 250 kW.

That is the orbital data centre thesis moving from slide deck to silicon partner. It is genuine progress. It is also a long way from revenue, and it explains a meaningful share of that $15.8 billion AI capex number.

Second, the merger overhang. Elon Musk is reportedly still pushing for a combination of SpaceX and Tesla. The Wall Street Journal reported on Friday that executives are working through how to dispose of Tesla’s China business should that happen, on the basis that SpaceX’s government and national defence contracts could trouble Beijing.

For a trader, the practical effect of a live merger rumour is that it adds a second source of unpredictable headline risk to an instrument whose beta already sits near 5.9. You are not just trading SpaceX. You are trading a corporate structure that may not exist in its current form.

Separately, Musk has indicated that absent problems surfacing in mission data review, the next Starship flight will attempt to catch the upper stage with the tower at Starbase, using the same Mechazilla arms that already catch the larger Super Heavy booster. It would be a first, and it is the single milestone most tied to the long-term bull case, because cheap reuse is what makes orbital compute economically plausible at all.

What the Headlines Get Wrong

Three errors are circulating tonight, and being able to spot them is worth more to you than any price target.

The drawdown figure depends on the baseline. You will see “down more than 50%” and “down 16%” in coverage of the same stock on the same day. Both are arithmetically defensible. The first measures from the $225.64 June high. The second measures from the June listing. At Tuesday’s close of $125.33, SPCX is roughly 44% below the high and about 7% below the $135 offer price. Anyone quoting a percentage without naming the baseline is telling you nothing.

Pre-print previews are not results. Several articles still circulating quote revenue near $6.8 billion and a loss near $1.9 billion. Those are the estimates that were published before the release. The actual numbers were better on both counts. Check the timestamp before you check the number.

At least one widely syndicated report contains an obvious data error in the Starlink subscriber figure, stating a number that exceeds the human population of the planet. It is a typographical error rather than a claim, but it will be copied and recopied for the next fortnight. When a statistic is physically impossible, do not repeat it, and treat everything else in that paragraph with more care.

SpaceX Q2 Earnings Through Mind, Method and Money

Mind. Notice how uncomfortable the honest answer is. The company beat on every reported line, and the stock still fell, and both of those facts are legitimate. Your brain will want to resolve that tension by deciding which one is “real.” Resist it. Markets routinely price two true things at once: the quarter was better than expected, and the capital intensity is a problem. A trader who can only hold one idea at a time will be wrong at every inflection.

Method. This stock is now seven weeks old. There is still no meaningful market structure to read. Every level on the chart was created in the last two months, most of it during a violent one-way move, and the float is about to triple. Support and resistance require a memory of participants at those prices, and the population of participants is about to change composition overnight. If your edge is technical, this instrument does not currently offer you one. Sitting out an event you cannot model is a decision, not an absence of one.

Money. A beta near 5.9 means your standard position size carries roughly six times your standard risk. Layer an earnings reaction and a lock-up expiry into the same 48-hour window and the honest sizing answer for most traders is smaller than usual or nothing at all. If you have not recalculated what your normal risk percentage actually permits on an instrument this volatile, our trading calculators will do it in under a minute. This is the difference between a bad week and a career-ending one.

The Uncomfortable Summary

SpaceX had a good quarter. Revenue nearly doubled year over year. The AI segment lost about half what analysts feared. Starlink threw off $1.66 billion of operating income. At the segment level the business was within touching distance of breaking even.

It also spent $18.37 billion, roughly a full prior year of capital expenditure, in ninety days, and it will meet its largest ever increase in tradeable supply on Thursday.

Both things are true. The market’s job over the next few weeks is to decide which one it cares about more, and it will probably change its mind several times while doing so. Your job is considerably simpler: do not size a position as though you know the answer.

We have written about this pattern at length in Market Mayhem. The specifics rotate. The structure does not. A compelling story, a constrained float, a vertical move, and then the slow mechanical grind as supply catches up with enthusiasm. The people hurt worst are almost never the ones who lacked conviction.

Related Reading

• SPCX Stock: The IPO That Fell Below Its Price, and the Calendar That Explains Why — the full lock-up schedule through June 2027, and why the float, not the story, has been setting the price.
• SpaceX IPO 2026: What Traders Must Know Before the Biggest Listing in History — the pre-listing analysis and the three scenarios we mapped.
• SpaceX: The Company That Bet Everything on Its Last Rocket — the founding story and the 2008 near-bankruptcy that nearly ended all of this.
• The M·M·M Assessment — find out whether your weakest link is Mind, Method or Money before an event week finds out for you.

Frequently Asked Questions

What did SpaceX report in Q2 2026?

SpaceX reported revenue of $7.81 billion against consensus near $6.9 billion, adjusted EBITDA of $3.5 billion against $2.0 billion expected, and a net loss of 9 cents per share, all ahead of estimates. Capital expenditure was $18.37 billion, of which $15.8 billion went to AI infrastructure. It was the company’s first quarterly report as a public company following its June 2026 listing.

Why did SPCX stock fall after beating estimates?

The shares had already rallied more than 9% during Tuesday’s regular session in anticipation of a good print, closing near $125.33. After the release, attention shifted from the revenue beat to the $18.37 billion of quarterly capital expenditure, the continuing operating losses in the launch and AI segments, and the lock-up expiry on 6 August that frees roughly 911.5 million shares.

Is Starlink profitable?

Yes, and it is the only SpaceX segment that consistently is. Connectivity generated $4.29 billion of revenue in Q2 2026 against expectations of $3.83 billion, producing $1.66 billion of operating income and $2.60 billion of adjusted EBITDA. The Space and AI segments both reported operating losses in the same quarter.

When does the SpaceX lock-up expire?

It expires in stages. Roughly 911.5 million shares unlock on 6 August 2026, two days after the first earnings report, described as approximately triple the current tradable float. Further tranches release through late August and September, a larger block is tied to Q3 earnings, and the 180-day lock-up fully expires on 8 December 2026. Elon Musk’s holding of roughly 6.4 billion shares remains locked until 12 June 2027.

How far is SPCX below its IPO price?

At Tuesday’s close of $125.33, SPCX traded about 7% below the $135 IPO price and roughly 44% below its high of $225.64 set on 16 June 2026. Percentage figures quoted in the media vary widely because different outlets measure from different baselines, so always check which starting point a given number uses.

This article is educational and is not financial advice. It does not recommend buying or selling SpaceX, SPCX, or any security. Figures are drawn from the company’s Q2 2026 earnings release and contemporaneous reporting on 4 August 2026, including Yahoo Finance, Invezz and crypto.news, and from analyst estimates published by Bloomberg and StreetAccount. Consensus figures vary between survey providers. Prices and estimates change constantly; verify current figures against primary sources and the company’s filings before making any decision. Trading and investing carry risk, including the loss of capital, and leveraged products such as CFDs can result in losses exceeding your deposit.