SpaceX is racing toward orbital data centers. Here are ways to play this emerging technology

Morgan Stanley argues orbital compute could become a credible “next leg” of the AI boom, citing tighter constraints on Earth data centers, falling launch costs, and advances in optical satellite networking. It identifies 43 companies tied to orbital compute, with 15 U.S.-based firms dominating the supply chain by market value, and highlights near-term potential in “orbital edge-AI” rather than replacing terrestrial hyperscale data centers this cycle. The story frames SpaceX’s reusable rockets and Starlink as proof points for scalable commercial space networks, supporting upside for defense/civil/commercial space compute enablers.

Analysis

The market is likely to misprice this as a near-term space-tech revenue story when it is really a financing-and-optionality story for already-scaled infrastructure names. The cleanest beneficiaries are NVDA, AVGO, TSM and MU: if orbital edge-AI moves beyond pilots, the spend still starts with inference silicon, advanced packaging, memory bandwidth, and interconnect, not a brand-new hardware stack. That means the first-order effect is multiple support, while the second-order effect is a gradual shift in who gets budget within AI capex — away from pure ground power/real-estate bottlenecks and toward higher-value content per deployed watt.

The less obvious winner set is the “boring” enabling layer: IFNNY, STM and MRCY-type names that participate in radiation-tolerant power management, hardened logic, and defense-qualified subsystems. But the probability-weighted revenue impact is small over the next 1-3 quarters; the real catalyst window is 12-24 months and requires repeated launches, customer qualification, and evidence that on-orbit inference cuts mission cost or latency enough to justify procurement. If launch costs stop falling, reliability issues surface, or terrestrial AI buildouts become easier via cheaper power/grid connections, the orbital thesis loses urgency quickly.

Contrarianly, the consensus may be underweighting how much of the upside accrues to current AI incumbents rather than any pure-play space equity — there is no obvious public-market “orbital compute” leader yet. I would fade the temptation to chase thinly traded theme names on narrative alone and instead own quality suppliers with existing AI demand, using the space angle as a free call option. The overdone risk is microcap rerating without revenue; the underdone risk is that defense and sovereign customers fund early pilots faster than commercial buyers do.

Key Decisions for Investors

• Long NVDA / AVGO / TSM on 2-5% pullbacks over the next 1-3 months: best risk/reward as the theme expands valuation support without requiring orbital revenue to show up in numbers; falsify if AI capex guides down or gross-margin mix deteriorates.
• Pair trade: long NVDA or AVGO vs short a speculative basket of AXTI + MRCY for 1-3 months to express 'quality enablers over narrative beta'; this should work if the market overbids smaller space-adjacent names before contracts materialize.
• Add IFNNY and STM as a 6-12 month structural basket for radiation-hardening / power-management optionality; upside is slower but more defensible than pure space plays, and downside is limited if the orbital thesis stalls.
• Avoid initiating fresh longs in the smallest theme proxies until there is visible customer validation or launch cadence data; set an alert for any defense or hyperscaler pilot announcement before paying up for the story.
• If willing to express the theme via options, use out-of-the-money call spreads in NVDA or AVGO rather than microcap equity: cleaner liquidity, better ability to monetize a sentiment-driven rerating, and less binary execution risk.