Starship, Starlink, or xAI: Which Part of SpaceX Could Move the Stock Most in 2027?

Starlink generated 61% of SpaceX revenue last quarter and produced $2.6 billion of Q2 EBITDA, up 64% year over year, making it the company’s only profitable business. However, 1H 2026 AI capital expenditures surged 610% to $23.5 billion, representing 83% of total capex, while xAI generated only $3.4 billion in first-half sales. Starlink’s 12 million customers and a projected $129 billion satellite-internet TAM support upside, but investors will focus on whether AI revenue can scale quickly enough to justify escalating data-center spending.

Analysis

The central equity issue is capital allocation, not connectivity growth. If AI infrastructure outlays remain materially ahead of monetization, SPCX will increasingly be valued as a capital-intensive compute developer rather than a high-recurring-revenue communications platform; that implies lower FCF conversion and a multiple closer to hyperscale infrastructure than software-like broadband. The key 1-3 month catalyst is credible segment disclosure on AI backlog, utilization, power commitments, and external customer mix—not another model launch or capex announcement.

Second-order beneficiaries of sustained AI buildout are NVDA, AVGO, VRT and power suppliers, but only where orders are independently corroborated: a captive AI program can pull forward hardware demand while destroying the buyer's returns. Starlink's strategic value is its ability to fund launch cadence and network expansion internally, yet its economics are exposed to terminal subsidies, satellite replacement costs, spectrum constraints, and direct-to-device competition from ASTS and terrestrial partners T and VZ. A more mature satellite network can also pressure legacy maritime, aviation and remote-connectivity providers before it disrupts mass-market mobile service.

Contrarian view: the market may be over-focusing on reported AI capex and underweighting financing risk. If spending is supported by equity issuance, related-party arrangements, or vendor financing rather than operating cash flow, the apparent growth optionality can mask dilution and leverage-like obligations; this is especially important before assigning a premium sum-of-the-parts valuation. Conversely, a demonstrated AI revenue run-rate that rises faster than capex for two consecutive quarters would rapidly invalidate the bearish cash-flow thesis.

Key Decisions for Investors

• Do not initiate a directional SPCX position until segment-level cash burn, financing sources, and AI backlog are independently verified; set an alert for AI capex/revenue falling below 3x on an annualized basis or management guiding to positive incremental AI gross margin. Those metrics would support rerating; continued capex escalation without contracted revenue is the falsifier for any long case.
• Use a 3-6 month tactical long basket of NVDA/VRT versus SPCX only after disclosed supplier commitments or capex deployment confirm that the compute build is real. Target modest sizing: suppliers capture early revenue while SPCX bears construction, power, and utilization risk; exit if AI capex is deferred or hardware lead times normalize.
• Monitor ASTS against SPCX as a competitive-risk hedge rather than a standalone short: initiate long ASTS only on evidence of commercial direct-to-device activation or incremental carrier funding. The thesis is falsified by Starlink securing broad terrestrial-spectrum partnerships that materially reduce ASTS's differentiated coverage advantage.
• For an eventual SPCX long, require a valuation entry point tied to normalized connectivity EBITDA/FCF rather than a revenue multiple incorporating unproven AI sales. Avoid buying immediately following AI narrative-driven rallies; the next two reporting periods should determine whether the business earns a compute-platform multiple or suffers FCF-driven multiple compression.