Elon Musk delivers ‘totally nuts’ plans for moon robots and insists $1 trillion revenue target will hit but capex tanks SpaceX on debut earnings

SpaceX’s first public earnings show $7.8B Q2 revenue (nearly doubling YoY, above the $6.9B consensus) and losses cut to about half of a year ago, but shares fell >5% after results. The market focused on capex: total capex hit $18.4B in Q2, with roughly $16B on AI compute infrastructure vs $13.2B expected, raising investor concerns about capital intensity and slower-than-expected payback. Starlink revenue reached $4.3B (+66% YoY) and the AI segment grew 247% to $2.6B, yet an additional overhang is expected as nearly 1B insider shares become sellable on Thursday.

Analysis

The market is signaling a shift from rewarding narrative growth to demanding visible marginal returns on AI capex. That is a headwind for capex-heavy, story-driven names like GOOGL and TSLA, where investors now punish spending before they trust monetization, and a relative tailwind for NVDA, which monetizes the buildout regardless of who wins the end-user application layer.

The more important second-order effect is that private-market enthusiasm can still contaminate public comps: if a marquee name with strong growth gets de-rated on lockup supply and spend intensity, public peers will trade with a higher discount rate until they show either backlog conversion or margin inflection. Over the next 1-4 weeks, the supply overhang matters more than fundamentals; over 1-3 months, the key catalyst is whether AI revenue scales faster than depreciation and interest-like operating expense. If it does not, multiple compression can persist even on good top-line prints.

Contrarian view: consensus is probably overemphasizing "capex bad" and underestimating the bargaining power of scarce compute. NVDA still sits at the center of the ecosystem, and the names that can prove demand translation fastest should outperform. The thesis is falsified if the next quarter shows capex moderating without a slowdown in growth, or if backlog/margin commentary turns sharply better than expected, which would force the market to re-rate these assets as infrastructure rather than moonshot optionality.

Key Decisions for Investors

• Go long NVDA vs short GOOGL for 1-3 months: pair the beneficiary of compute demand with the name most exposed to capex skepticism; target 10-15% relative outperformance if AI spend remains front-loaded. Falsify if GOOGL shows clear margin reacceleration and capex discipline on the next update.
• Buy TSLA 1-2 month downside put spreads or short into strength: the market is re-pricing Musk-linked capex stories with less patience, and TSLA remains vulnerable to multiple compression when the AI/robotics narrative weakens. Exit if the stock reclaims post-event levels on improving delivery or margin data.
• Add on pullbacks to MSFT long exposure over the next 1-3 months: it remains the clearest beneficiary of the market rewarding monetization visibility and large backlog, while its valuation can absorb modest capex intensity better than peers. Falsify if cloud growth decelerates or AI margin contribution stalls.
• If looking for a tactical hedge, use QQQ or a megacap tech basket short against NVDA longs for the next earnings cycle: it isolates compute beneficiaries from names where investors are now demanding faster payback. Cover if the market begins rewarding capex expansion again.