SpaceX Is Going Public: Here's the 1 Thing the Company Must Get Right to Justify a $1.5 Trillion Valuation

SpaceX’s S-1 shows 2025 total revenue of $18.674B and operating loss of $2.589B, with only the Connectivity segment profitable at $4.423B of operating income. The article argues that a rumored $1.5T IPO valuation is hard to justify given weak Space and AI profitability, and notes xAI may rent $15B of annual capacity to Anthropic at an unknown margin. Investors are told to watch whether SpaceX’s AI strategy, including space-based data centers and Terafab, can become a real medium-term growth driver.

Analysis

The market is likely anchoring on the headline valuation while underweighting the more important variable: which part of the conglomerate can compound capital without needing repeated capital raises. That makes the connectivity cash engine the key cross-subsidy source, but it also means the equity is being priced more like a pre-revenue frontier platform than a telecom-like cash generator, leaving little margin for execution slippage. The most important second-order effect is that any AI disappointment won’t just compress the AI segment; it could force more of the burden onto Starlink, implicitly turning a maturing asset into the financier of speculative growth.

The AI narrative is more fragile than the top-line aggregate suggests because the disclosed economics imply capacity utilization, not product-market fit, is doing the heavy lifting. Renting compute to a better-capitalized competitor may support near-term revenue, but it also signals that the company is still not extracting enough value from its own model stack, which makes the addressable market assumptions look aspirational rather than bankable. If that persists for 2-4 quarters, investors will likely re-rate the AI optionality down and focus on capital intensity, where every extra dollar of infrastructure has to earn its keep against hyperscaler and chip-leader benchmarks.

The real sleeper variable is supply chain power: if this group truly pushes toward in-house chips and space-based compute, it threatens the economics of both general-purpose cloud and the custom-accelerator ecosystem, but only on a multi-year horizon. In the nearer term, the likely beneficiaries are not the company itself but upstream component makers, launch-adjacent suppliers, and power/thermal management names that can sell into experimental infrastructure without taking model risk. Conversely, the most direct public-market loser is probably the “AI infrastructure at any price” cohort, because a credible path to lower energy and silicon costs would compress the scarcity premium embedded in NVDA/TSM-adjacent multiples.

The contrarian point is that the headline valuation may not be crazy if investors treat the equity as a call option on execution across three stacked moonshots, not as a sum-of-the-parts telecom or aerospace business. The market is likely overpricing near-term monetization but underpricing the strategic value of owning launch, connectivity, and compute under one capital structure. That makes the stock highly sensitive to evidence on utilization, margins, and capex discipline over the next 6-12 months rather than to the IPO price alone.

Key Decisions for Investors

• Avoid chasing the IPO in the first 1-2 weeks; wait for post-listing price discovery and a first-quarter lockup/secondary supply window before considering exposure. Risk/reward is poor if the stock prices the AI option before utilization and capex disclosures stabilize.
• Long NVDA / short TSM as a tactical pair over 3-6 months if the market starts believing in customer-designed silicon and lower-margin internal fab ambitions. The trade works if sentiment shifts toward disintermediation risk for foundry economics while NVDA retains the higher-quality software/system moat.
• Add a small long in power and thermal-management beneficiaries, not the headline AI names, on any pullback: names exposed to data-center power infrastructure and liquid cooling should benefit if the market prices space/compute experimentation as a real capex theme over 12-24 months.
• For TSLA, treat the linkage as optionality rather than core earnings support; only own upside via defined-risk calls if there is evidence the fab joint venture gains real milestones. The risk/reward is asymmetric only if the market starts capitalizing the chip-manufacturing angle, which is still a multi-year execution story.
• Watch for a short setup in AI infrastructure hardware on any rally tied to the IPO narrative. If the group’s AI economics remain reliant on renting capacity to third parties, the market may eventually rotate from 'scarcity premium' to 'utilization scrutiny,' which would compress multiples across the highest-multiple infrastructure names.