SpaceX: The Cash-Flow Hole the Market Isn't Pricing

The old story is simple: SpaceX is the biggest IPO in history, the gateway to an orbital AI boom, and the stock will reclaim its highs once earnings clear. Deutsche Bank's Edison Yu called the run into this first public report a "bumpy ride" and kept his $255 buy rating intact. The broader Street consensus - 35 analysts polled by S&P Global - sits at a similar Buy, with an average target of $223, implying the shares should nearly double from here.

The newer evidence makes that old story harder to sustain.

The Cash-Flow Bridge Doesn't Exist Yet

SpaceX will report its first public earnings tonight. Analysts expect Q2 revenue of roughly $6.9 billion, up sharply from $4.7 billion in Q1. The adjusted loss per share should narrow to about 23 cents from the 119 cents posted last quarter. Revenue growth is impressive. The loss trajectory is moving in the right direction.

That is not the story that matters for the next 12 months.

The last twelve months of cash flow tell a different one. Operating cash flow came in at $7.1 billion. That sounds positive until you see what's sitting on the other side: capital expenditures of $26.9 billion. Free cash flow - the cash a business generates after paying for the assets it needs to operate and grow - is minus $19.8 billion. The FCF margin sits at minus 102.5 percent.

That burn is heavily concentrated in AI infrastructure. Q1 alone saw the AI segment consume $7.72 billion in capex - roughly three-quarters of total spending. Analysts expect that figure to climb above $10 billion in Q2. S&P Global's Visible Alpha data projects full-year capex rising from nearly $46 billion in 2026 to $87 billion in 2027, with debt growing fivefold from roughly $42 billion to over $218 billion by FY2028.

The market is pricing a trillion-dollar AI infrastructure company that hasn't proven the economics of its AI bet. It is paying 83 times trailing revenue for a business that will not be free-cash-flow positive until at least 2029, according to published analyst estimates. Morgan Stanley's analysts recently wrote that most investors are assigning little or even negative value to the AI unit "given the high capital spending requirements" and "largely uncertain economics."

What Starlink Can and Cannot Fund

The counterargument is that Starlink - SpaceX's satellite broadband service - is the cash engine that subsidizes everything else. The connectivity segment generated roughly $3.8 billion in revenue last quarter with an expected operating profit of around $1.4 billion. Subscribers topped 10.3 million by the end of March, roughly double a year earlier.

The problem is scale. Starlink's quarterly operating profit would need to grow dramatically to offset an AI capex program that analysts now annualize at $30 billion or more. Will Rhind of GraniteShares put it bluntly: Starlink "cannot single-handedly fund a $30 billion annualized AI capex program." The satellite business is executing well, but it is not a bridge to the AI ambitions the valuation assumes.

Average revenue per subscriber has fallen nearly 25 percent year over year. That margin pressure has to come from volume growth, which keeps the math tight. And the AI revenue side - currently projected at roughly $2.2 billion for Q2, up from negligible levels a year ago - is built on compute-lease deals with Anthropic, Alphabet, and Reflection AI. Those contracts are real. Whether they generate enough return on the capital deployed to justify the spending cadence is the question the earnings call will not fully answer.

The Tape Pain Is Worse Than the Business Pain. For Now.

The stock closed Monday near $115, up 6 percent today on pre-earnings positioning. The intraday low of $104.83 set in recent weeks represents a roughly 50 percent pullback from the June 16 intraday peak of $225.64. The market cap has shed nearly $500 billion since the IPO.

That kind of tape action looks like a broken thesis. It isn't - yet. The revenue bridge from $18.7 billion in 2025 to a consensus $39.1 billion in 2026 is real and accelerating. Gross margins are expanding toward the mid-50s percent range. The launch business, Starlink, and AI compute leases are all growing. The company's balance sheet still carries $15.9 billion in cash against $60.5 billion in total debt, leaving net debt of roughly $6.6 billion.

What the tape is also pricing in - and what hasn't fully hit the public mind - is the lockup expiration on August 6. About 912 million shares, representing roughly 7 percent of the company and valued at over $100 billion at current prices, become eligible for trading in two days. That is roughly triple the current tradable float. Supply elasticity of this magnitude will keep the stock volatile regardless of tonight's numbers.

The Inflection Setup Is Still There - But the Proof Path Is Longer

Here is what gets better over the next 12 months: revenue nearly doubles, AI lease revenue moves from a rounding error to a multi-billion-dollar annual run rate, and the loss per share continues to narrow as Starlink scales and capex eventually peaks relative to top-line growth. Consensus EPS moves from $0.20 this year to $0.90 next, implying the operating model starts printing adjusted earnings by FY2027.

Here is what makes the setup harder to trust: the valuation assumes that entire trajectory hits without meaningful stumble, and the free-cash-flow hole doesn't close until roughly three years out. A stock trading at 83 times revenue with negative ROIC (-14.7 percent) and negative ROE (-50.3 percent) requires extraordinary execution to justify its price. Even a modest capex overshoot or AI lease slowdown could trigger another leg down.

The market is still pricing the old story - the orbital AI megacap that will inevitably validate a $1.6 trillion valuation. But the cash-flow path says the inflection is farther away than the IPO tape made it feel. The earnings tonight will not change that timeline. They will only show whether the revenue acceleration and margin trajectory are on track to eventually get there.

The actionable frame: if you already own the position, the lockup-driven volatility and the fundamental gap between revenue growth and cash-flow generation suggest sitting tight is a rational stance, provided the revenue and AI-lease trajectory stay intact. For new money, waiting for the post-lockup dust to settle and the first full year of public guidance gives a cleaner entry. A simple forward-revenue multiple approach would put a defensible long-term target in the $200 to $220 range - near Street consensus - only if the $39 billion revenue estimate for 2026 holds and the AI segment shows credible unit economics. That implies 60-70 percent upside from current levels, but it also requires the entire consensus growth path to materialize.

The tripwire: if Q2 capex runs materially above the $14 billion consensus, or if Musk raises full-year spending guidance above $50 billion, the cash-flow hole deepens and the valuation story loses its anchor. At that point, the business may be growing beautifully but the math stops supporting the multiple. Discipline over ego.

I can be wrong again. But the setup requires the revenue bridge to hold and the capex burn to eventually bend toward self-funding. Tonight's numbers are the first checkpoint.