Cathie Wood Keeps Buying SpaceX Stock. Here's Why I Wouldn't Buy It Above a $1.8 Trillion Valuation.

SpaceX/Starlink reported $4.3B Starlink revenue in Q2 2026 with 38.6% operating margins, but the company posted -$25B free cash flow for the first half of 2026, driven largely by ~ $16B in AI capex. The article argues that while AI revenue rose from $818M to $2.6B in one quarter, the cash burn and uncertain long-term ROI make SpaceX stock unattractive at a valuation above $1.8T. It also notes Ark Invest/Cathie Wood continues buying, though the author says that doesn’t change the valuation/cash concerns.

Analysis

The market is likely overpaying for a “high-margin satellite software” story while underweighting the consolidated capital intensity. A business with strong unit economics can still be a poor equity if one adjacent segment behaves like a permanent build cycle; that shifts the stock from compounder to funding vehicle and makes the multiple highly sensitive to any slowdown in external capital appetite. In that setup, the real beneficiaries are upstream compute and networking vendors, especially NVDA, because they get paid on the spend even if the end-market economics remain unresolved.

Near term, sentiment can stay hot for days to weeks because narrative dominates cash burn in the first post-IPO phase. The bigger inflection is 1-3 months: investors will look for evidence that AI capex is decelerating faster than revenue growth, or the stock will start trading on funding risk rather than optionality. If management is forced to defend the spend with more debt, equity, or repeated capital raises, that would likely compress the entire high-duration growth complex.

The contrarian point is that the consensus may be treating the connectivity business as if it can ring-fence the rest of the enterprise. In reality, conglomerate valuation breaks when the best asset cannot reliably fund the weakest one. The thesis is falsified if capex falls sharply, AI gross profit inflects toward true cash conversion, or Starship materially reduces launch costs and capex intensity within the next several quarters.

Key Decisions for Investors

• If SPCX is liquid enough to short, fade strength on any post-IPO multiple expansion with a 3-6 month horizon; use puts or put spreads rather than outright short if borrow is tight. Falsify if consolidated FCF burn narrows by more than ~50% over the next two reported quarters.
• Go long NVDA as a pick-and-shovels beneficiary of the same capex cycle; thesis holds for 1-3 months if AI infrastructure spending stays elevated, but trim if major customers start signaling capex deferral or order digestion.
• Do not chase TSLA on any Musk-ecosystem sympathy bid; if speculative-duration sentiment rolls over, TSLA can underperform on multiple compression even without a fundamental reset. Treat as a watch item unless the stock breaks out on volume with improving deliveries.
• Set an alert for any disclosure that shifts SPCX from growth investment to explicit funding need; that is the catalyst that would convert the story from expensive to structurally vulnerable.