2 Space Stocks to Buy in September

The article frames the space sector as potentially growing to $1.8T by 2035, with SpaceX positioned as the standout due to AI-driven revenue. For SpaceX, the piece cites compute-rental deals that could generate $26B annually (39% above 2025 sales of $18.7B) plus a Reflection AI arrangement potentially worth $6.3B through 2029, alongside heavy AI capex of $12.7B in 2025. Despite recent volatility since SpaceX’s June 12 IPO, analysts tracked by CNN reportedly rate 75% of coverage as buy, with a median one-year target of $217 (≈51% above the Aug. 31 close of $143.69).

Analysis

MDA is the cleaner public-market expression because it monetizes procurement, not narrative. In a space cycle, the durable winners are the firms selling mission-critical hardware and serviceability, since their pricing power comes from backlog visibility and switching costs, not TAM slides; that usually shows up as margin mix expansion before the broader sector rerates. If sovereign/defense buyers keep funding resilience and on-orbit servicing, MDA should benefit more consistently than the headline-grabbing launch/platform names.

SPCX is a different animal: the upside case is dominated by very long-dated AI infrastructure optionality, while the near-term P&L is likely to be defined by capex intensity, launch cadence, and regulatory execution. The market risk is that investors are paying today for a 2028+ architecture that still needs power, spectrum, orbital-debris, and customer adoption to line up; that creates a classic multiple-over-earnings mismatch if growth enthusiasm cools over the next 1-3 quarters. The broader beneficiaries in the next 12 months are more likely to be NVDA, networking, power, and launch-adjacent suppliers than the platform owner itself.

Contrarian view: the consensus is probably overestimating how much of the AI-space story is incremental economic value versus a re-labeling of existing compute demand. If the space compute buildout mostly displaces terrestrial cloud spend, the addressable profit pool is smaller than the TAM rhetoric implies, and that should matter more than the headline revenue figure. For public equities, the cleanest setup is still relative value: proven industrial monetization versus speculative duration; if MDA keeps converting backlog, it can compound for months, but if SPCX fails to show customer traction or capex discipline, the story can de-rate quickly.

Key Decisions for Investors

• Long MDA on pullbacks or into confirming backlog/margin print; target a 3-6 month hold for backlog conversion and margin expansion, with thesis invalidated if adjusted margin or order intake decelerates.
• If borrow/liquidity are workable, pair long MDA / short SPCX as a 3-6 month relative-value trade: own the cash-flowing picks-and-shovels winner, fade the duration-heavy AI optionality. Falsify if SPCX shows credible near-term revenue conversion or capex-to-revenue efficiency improves sharply.
• Use NVDA as the higher-quality second-order beneficiary rather than SPCX itself; any pullback tied to skepticism around space AI is a buy-the-dip setup only if it confirms that GPU demand from alternative data-center builds is real. Invalidated by evidence that SpaceX-like projects remain too small to move supplier demand.
• Avoid initiating a new long in SPCX until the market gets proof points on customer retention, capex payback, and regulatory path for orbital infrastructure. Current risk/reward looks asymmetric to the downside if execution slips over the next 1-2 quarters.
• Watch GOOGL only as a sentiment barometer, not a direct beneficiary; if the market starts pricing alternative-compute optionality into hyperscalers, fade any overreaction unless it shows up in tangible cloud pricing or capex guidance.