SpaceX's Third Quarter Just Ended. Here Are the 3 Numbers That Matter Most for the Stock.

SpaceX's upcoming third-quarter report will test whether its AI infrastructure spending is converting into revenue, with Anthropic's contract reaching a $1.25 billion monthly run rate and cloud-services revenue targeted at roughly $3.75 billion or more. Starlink ARPU fell to $66 per month in Q2 from $85 a year earlier despite subscriber growth to 12.0 million, while Falcon launches declined to 37 from 45 a year earlier. Investors will also watch whether Google’s $920 million monthly agreement begins at full rate in October, but the company’s $541 million Q2 loss and roughly $2 trillion valuation leave limited room for execution misses.

Analysis

The central equity risk is not demand but conversion: SPCX has effectively exchanged a large, front-loaded infrastructure build for a small number of concentrated counterparties. If recognized AI revenue lags contracted run-rate, investors will re-rate the spend as speculative capacity rather than contracted infrastructure, pressuring both EBITDA expectations and the post-IPO valuation premium. The relevant diligence item is cash collection, minimum-volume commitments, termination rights, and whether revenue is reported gross or net—not management's reported contract value.

Near term, the earnings release creates an unusual liquidity overhang because lockup supply arrives immediately after the principal information event. Even a numerical beat may fail to sustain if incremental cloud revenue is accompanied by weaker gross margin, receivables growth, or a higher 2027 capex outlook; that is the most likely "good news, bad stock" setup over days to weeks. Conversely, verified on-time hardware acceptance by the second large customer would reduce execution risk and could force a sharp squeeze given the stock remains below its prior high.

Over 6-18 months, lower blended connectivity monetization makes launch efficiency more strategically important: more internally dedicated launches support network growth but defer third-party launch revenue and can mask the true capital intensity of subscriber additions. The contrarian view is that the market may be underweight the value of scarce power, cooling, and interconnect capacity if AI customers are paying for a bundled solution rather than commodity compute; that upside requires evidence of durable gross margins and additional customers, not merely signed commitments.

Key Decisions for Investors

• Remain underweight SPCX into earnings; do not add on headline AI-revenue strength unless filings show collections tracking revenue, no material acceptance contingency, and cloud gross margin stabilization. Reassess 1-3 days after the lockup shares become tradable rather than buying the initial print.
• Tactical pair for the event window: short SPCX / long GOOG in equal beta-adjusted dollars for 2-6 weeks. SPCX bears delivery, capex, and lockup risk, while GOOG retains optionality to source capacity elsewhere; cover if SPCX confirms customer acceptance and raises forward contracted-revenue guidance without a capex increase.
• Set a positive SPCX alert—not a buy order—if cloud revenue materially exceeds the implied full-rate customer contribution while receivables remain proportionate and management identifies a third external customer. That combination would validate utilization beyond a bilateral contract and support multiple expansion over 6-12 months.
• Avoid using NVDA as a direct hedge: the available information does not establish that SPCX capacity is NVIDIA-based or that its build changes GPU procurement. Monitor disclosed accelerator mix and supplier commitments before expressing a semiconductor read-through.