SpaceX Has Three AI Customers Paying $27.8 Billion a Year. How Big Can This Revenue Stream Get?

SpaceX has signed three AI compute deals totaling about $2.32 billion in monthly revenue, or $27.8 billion annualized, more than doubling its 2025 companywide revenue of $18.7 billion. The contracts with Anthropic, Google, and Reflection AI monetize underutilized GPU capacity and could become a high-margin business, with gross margins potentially near or above 70% based on comparable GPU cloud providers. The piece is constructive on SpaceX's long-term earnings power, though it notes the stock remains expensive at roughly 40x sales and lacks an established profitability track record.

Analysis

The immediate winner is not just SpaceX; it is the broader GPU ecosystem that gets monetized twice — first through hardware sales, then through cloud-like lease economics on the same installed base. That structurally favors NVDA because these deals validate persistent enterprise demand for frontier compute, but the bigger second-order effect is on capital-light inference/training capacity becoming a new asset class, which should compress the moat of hyperscalers that previously used balance sheet size as a barrier to entry.

The underappreciated takeaway is that this is a supply-demand relief valve for AI capex rather than a pure growth story. If third-party leasing proves sticky, the marginal customer is likely to be a cash-constrained model lab or a hyperscaler optimizing ROIC, which would shift spend away from owned data centers and toward contracted capacity. That is mildly negative for GOOGL/MSFT capex efficiency optics, while structurally positive for CRWV-style intermediaries if pricing remains disciplined — but if SpaceX can truly offer excess capacity at high utilization, it threatens to undercut pure-play GPU clouds on cost.

The key risk is duration mismatch: these contracts look great in the next 12-24 months, but the market will eventually discount whether the revenue is durable after the current wave of AI training demand rolls over into lower-intensity inference workloads. Another risk is pricing power; if this becomes a commodity rack-rental market, headline revenue stays high while margins compress. The consensus may be overestimating the permanence of the margin profile and underestimating the chance that large customers re-insource once financing conditions improve.

Key Decisions for Investors

• Long NVDA on any 3-5% pullback over the next 2-6 weeks; the thesis is validation of multi-year GPU scarcity, with upside if the market re-rates backlog durability, but trim if lead times normalize faster than expected.
• Relative value: long CRWV / short a basket of hyperscaler capex proxies (equal-weight GOOGL/MSFT) for 1-3 months; this expresses the view that outsourced capacity monetization is the cleaner ROIC trade, though stop if hyperscaler capex spend re-accelerates without margin pressure.
• Sell OTM puts on GOOGL and MSFT with 60-90 day tenor; the market is likely to overreact to capex intensity headlines, but downside should be buffered unless there is evidence of lease pricing collapsing or demand decelerating sharply.
• Avoid chasing SpaceX-related AI enthusiasm at the current implied valuation multiple; if listed exposure becomes available, prefer a tactical short via call spreads rather than outright shorting, because narrative momentum can persist for several quarters.
• Watch for contract expansion announcements over the next 1-2 quarters; if additional counterparties sign, add to NVDA and CRWV on confirmation, but take profits quickly if the market starts treating compute leasing like a commoditized utility.