SpaceX: Revenue Beat Can't Offset the AI Cash Burn Spiral - Hold

SpaceX: Revenue Beat Can't Offset the AI Cash Burn Spiral - Hold

SpaceX beat Q2 revenue and loss estimates on its first-ever public earnings report today, and the stock initially fell 8% in extended trading, according to CNBC, but has since recovered to $125.33, up 9.43% from the prior close per Yahoo Finance. That disconnect is the entire story. The beat doesn't change the fact that the company is spending $15.8 billion on AI infrastructure in a single quarter to generate $2.56 billion of AI revenue. At that rate, valuation has to do a lot of heavy lifting before this math works.

I am staying at Hold on SPCX. The $1.65 trillion market cap at $125 per share reflects a company executing at the top end of its guidance across every segment. The AI capex trajectory, the August 6 lockup expiration, and the free-cash-flow deficit leave too little margin for error.

What changed today

The earnings print was better than expected. Q2 revenue came in at $7.81 billion versus a Bloomberg consensus of $6.81 billion - a $1 billion beat. Revenue grew 92% year over year, up from $4.1 billion a year ago and 66% sequentially from Q1's $4.7 billion. The loss per share narrowed to 9 cents versus an expected loss of 26 cents. Net loss shrank to $541 million from $1 billion a year ago.

Every segment beat. Starlink connectivity posted $4.29 billion in revenue versus $3.83 billion expected, with operating income of $1.66 billion. The space/launch unit brought in $962 million versus $835 million expected, though it remains deeply unprofitable with a $542 million operating loss. The AI segment - which includes xAI, Grok, the X social platform, and compute leasing - generated $2.56 billion versus $2.18 billion expected, with an operating loss of $1.26 billion versus a $2.39 billion loss estimate.

The results are solid on the surface. But the number that matters most for the stock's direction was the one that missed on the wrong side: Q2 capital expenditures of $18.37 billion, well above consensus estimates of about $14.05 billion. Total capex for the trailing twelve months now sits at $42.4 billion.

The AI burn rate is the problem

SpaceX is spending roughly $6.18 in AI infrastructure for every $1 of AI revenue it generates this quarter. That ratio makes it far more capital-intensive than any hyperscaler doing similar compute business. Alphabet, Meta, and Amazon are facing their own AI spending scrutiny, but they're generating hundreds of billions in revenue elsewhere to absorb the investment. SpaceX's total quarterly revenue is $7.8 billion. The AI buildout ate 617% of AI revenue and 202% of total company revenue.

The company's free cash flow for the trailing twelve months is negative $33.37 billion. Operating cash flow came in at $9.9 billion TTM, which looks fine until you factor in the $42.4 billion of capex. The FCF margin is -144.8%.

SpaceX executives told investors on the call that the company is "making its money back within a year" on its AI investments. That claim is worth stress-testing. It requires AI revenue to grow from $2.56 billion in Q2 to well over $15 billion in a single quarter - a six-fold increase in nine months - just to break even on a run-rate basis. The company has signed compute contracts with Anthropic, Alphabet, and Reflection AI that could total $28 billion in annualized revenue once fully ramped. But "could" and "once fully ramped" are not the same thing as near-term cash flow.

Starlink is the engine - for now

Starlink connectivity remains the only profitable business unit, and it's growing. Subscribers topped 12 million at the end of Q2, up from 10.3 million at the end of Q1. Enterprise and government revenue in the segment grew 108% year over year, which is encouraging.

The concern is ARPU compression. Average revenue per user fell nearly 25% as SpaceX enters more international markets and rolls out lower-priced plans. That pattern is expected in a scaling subscription business, but it means revenue growth alone doesn't guarantee margin expansion. Starlink's adjusted EBITDA came in at $2.6 billion versus $2.41 billion expected, a modest beat.

Starlink is also the supposed funder of everything else. CFO Bret Johnsen and CEO Elon Musk have framed Starlink profits as the fuel for AI capex and Starship development. But Starlink's $1.66 billion quarterly operating income is nowhere near enough to sustain a $15.8 billion quarterly AI capex bill. The company needs Starlink to grow several-fold or the AI unit to reach cash-flow break-even quickly. Neither timeline is guaranteed.

Valuation versus reality

At $125 per share and a $1.65 trillion market cap, SpaceX trades at 71.6 times trailing sales and 300 times trailing EV/EBITDA. For context, Nvidia - the actual proven AI infrastructure business with real margins and real cash flow - trades at 20 times sales and 31 times EV/EBITDA. Amazon trades at 3.9 times sales. Alphabet at 10.3 times.

Even annualizing today's Q2 revenue run rate of $7.81 billion to $31.2 billion gives a forward P/S of roughly 53x. That is not a valuation that can sustain execution missteps. It requires sustained triple-digit growth, margin expansion on AI, and flawless Starship progress for the multiple to hold.

The stock has fallen 45% from its all-time high of $225.64, reached after its June 12 IPO. It is now trading below its opening price of $150 for the first time. That sell-off represents a partial multiple compression, from an implied enterprise value above $2 trillion down to $1.59 trillion. But even at the reset, the stock demands flawless execution.

The lockup overhang

Two days from now, on August 6, the first post-IPO lockup expires. About 912 million shares - roughly 7% of outstanding shares - become eligible to trade. The current float is below 5% of total shares outstanding. After the unlock, the float expands to over 12%.

That is not a directional call. But in a stock already pressured by capital intensity and AI spending skepticism, more than doubling the tradable float without a corresponding demand catalyst is supply risk. Pre-IPO investors and insiders who bought shares at deep discounts now have the right to sell into a market that is questioning the valuation thesis. The stock has the technical profile of a name that will see headwind, not tailwind, from this event.

Catalysts on the clock

The next quarter and the next six months still offer potential catalysts. The $60 billion Cursor AI acquisition is expected to close in Q3, pending regulatory approval - Musk says the company is "pretty close" to that. The Nvidia partnership for the Starmind AI-1 orbital compute payload could eventually differentiate SpaceX's AI offering with datacenter-class compute in orbit, though that is years out. Shotwell expects a standalone Starlink Mobile service by end of 2027.

Starship's commercial debut, the most important operational milestone in the long-term story, was previously targeted for year-end 2026 but most analysts now expect it to slide into 2027. The company is attempting its first tower-catch of a Starship upper stage on the next test flight. Success would be a signal; failure would not move the stock much at this juncture.

The company is also targeting $100 billion in annualized revenue by end of 2026. Historically, according to one analyst's review, SpaceX has hit only 17% of its stated targets on time, with an average delay of roughly 20 months. That track record is worth factoring into how much weight to give forward guidance.

Risks and what would change the rating

The case for upgrading to Buy requires one of two things to happen:

• AI unit turns a corner faster than the market expects. If Q3 capex moderates or AI revenue grows enough to narrow the $6.18-to-$1 spend-to-revenue ratio materially, the multiple could re-rate. Compute contract ramp with Anthropic, Alphabet, and Reflection AI would need to accelerate beyond current market assumptions.
• Starlink scales profitability faster than subscriber growth implies. A sustained ARPU recovery or a major enterprise/government contract that resets the margin profile would make Starlink a more credible funder of the broader enterprise.

The case for downgrading is simpler and already partly priced in:

• Capex stays elevated while revenue decelerates. If Q3 capex again exceeds $15 billion and AI revenue growth slows from this quarter's pace, the valuation gap to peers could widen further.
• Lockup selling overwhelms demand. The August 6 unlock is a mechanical supply shock. If insider selling is aggressive and there's no offsetting institutional demand, the stock could revisit the $105 level near its 52-week low.
• Starship delays compound. Each quarter that Starship doesn't hit its milestones erodes the long-term growth narrative that justifies the current multiple.

The takeaway

SpaceX reported a strong first quarter as a public company. The beat was real. Revenue growth is real. Starlink is a genuine business engine with 12 million subscribers and growing enterprise demand.

But the stock fell after the beat because the AI capex number was a problem, not a proof point. Spending $18.4 billion in a quarter to run a company with $31 billion in annualized revenue is a model that requires relentless growth to justify. At 71 times sales, there's no room for the growth curve to flatten, even temporarily. And with 7% of shares about to unlock, supply pressure adds to the near-term headwind.

The rating is Hold. The business is ambitious, the Starlink engine is strong, and the AI upside is real if execution holds. But the valuation has already priced in perfection, the capex burn is unsustainable at current revenue levels, and the next two weeks bring more supply than there's evidence of demand to absorb.

I'll upgrade to Buy if Q3 shows a meaningful improvement in the AI spend-to-revenue ratio and capex guidance resets lower. Until then, the risk/reward at $125 doesn't support buying in.

Key metrics to monitor: Q3 capex and AI revenue run rate, Starlink ARPU trajectory, lockup selling volume post-August 6, and Starship test flight 14 results.