Is SpaceX a Top Artificial Intelligence (AI) Stock Pick in September?

SpaceX spent approximately $15.8 billion on AI compute infrastructure in Q2 after acquiring xAI and its Grok model, but the article argues this may be insufficient to compete with OpenAI and Anthropic. OpenAI has indicated $750 billion of AI-infrastructure spending through 2030, or roughly $38 billion per quarter if spread evenly over five years. While SpaceX reportedly raised about $100 billion through its IPO and subsequent debt sale, investors are urged to wait for more disclosure from prospective OpenAI and Anthropic IPOs before assessing xAI's competitive position.

Analysis

The relevant valuation question for SPCX is not whether Grok can grow, but whether AI can earn a return above the company’s blended cost of capital without diverting capital from its higher-conviction launch, connectivity, and satellite businesses. A subscale frontier-model spend can create the worst of both outcomes: enough fixed compute and talent expense to pressure consolidated free cash flow, but insufficient training capacity and distribution to close model-quality gaps. That would shift SPCX from being valued as a scarce aerospace/platform asset toward a capital-intensive AI conglomerate, warranting multiple compression over the next 6-18 months.

The more investable read-through is for AI infrastructure suppliers rather than a directional view on SPCX. If frontier labs’ public filings validate materially higher compute commitments, NVDA benefits from an industry capex floor even if individual model developers fail to monetize quickly; networking, memory, power, and data-center cooling suppliers would also gain. Conversely, if IPO disclosures reveal rapidly rising inference costs, weak enterprise retention, or aggressive depreciation assumptions, the market may re-rate AI capex as demand-pull rather than supply-constrained—creating a 1-3 month de-risking catalyst across high-multiple AI infrastructure.

Consensus may be too focused on model rankings and too little on financing duration. A well-capitalized but levered AI entrant can still face a funding mismatch if training obligations are front-loaded while revenue is uncertain; debt spreads, quarterly capex guidance, and free-cash-flow conversion matter more than headline liquidity. The thesis is falsified if SPCX demonstrates sustained AI revenue growth with stable gross margin and AI capex falling as a share of revenue for two consecutive quarters.

Key Decisions for Investors

• Remain neutral on SPCX until comparable IPO registration statements provide audited revenue, compute-lease obligations, depreciation policy, and customer concentration; this is an event-driven watch item, not a pre-disclosure long.
• Maintain an AI-infrastructure bias via NVDA rather than SPCX over the next 1-3 months, with risk controlled by reducing exposure if disclosed frontier-lab capex commitments fall below current market expectations or NVDA data-center guidance decelerates materially.
• For a relative-value expression after the IPO filings, consider long NVDA / short SPCX only if filings show materially larger competitor capex budgets and SPCX’s AI investment remains cash-flow negative; target a 10-15% relative move, with a stop if SPCX reports AI monetization or margin metrics above disclosed peers.
• Monitor SPCX debt spreads and consolidated free-cash-flow guidance quarterly. A widening of credit spreads alongside increased AI capex would be an early warning of balance-sheet-driven equity multiple compression; avoid adding long exposure into that setup.