$1 Trillion in Market Cap Erased, but This AI Stock Could Soon Be the World's Most Valuable Company (According to Elon Musk)

SpaceX shares remain roughly 5% below their IPO price after its valuation fell from a peak of $2.7 trillion to about $1.4 trillion in mid-July, despite a modest rebound. Elon Musk projects $3.5 trillion in revenue by 2033, but the company generated only $12.5 billion in first-half revenue and faces intense AI competition. Capital expenditures jumped 308% year over year to $28.5 billion in the first half, with 86% directed to AI; Evercore ISI estimates capex could reach $360 billion by 2030, heightening investor concerns over negative free cash flow and execution risk.

Analysis

SPCX is becoming a clean test of whether public markets will fund a frontier-AI buildout before utilization and pricing are visible. The relevant valuation risk is not the headline TAM but the gap between incremental invested capital and eventual data-center returns: if depreciation, power, networking and financing costs outrun contracted AI revenue, the market will re-rate SPCX from platform optionality toward a capital-intensive infrastructure multiple. That creates a 1-3 month vulnerability around any disclosure of rising cash burn, debt issuance, customer concentration, or delayed capacity monetization.

The second-order read-through is more nuanced for the AI supply chain. Heavy deployment remains incrementally supportive for MU and SNDK through memory/storage content, and modestly positive for NVDA, but only while spending converts to firm orders rather than speculative internal capacity. MSFT and GOOG are relative beneficiaries if SPCX's funding costs rise: their existing cloud distribution, enterprise sales channels and balance sheets let them absorb lower near-term AI returns while a less proven entrant must defend liquidity.

Consensus may be too focused on whether ambitious end-state projections are credible and not enough on the financing reflexivity. A weak SPCX equity tape can raise its cost of capital, forcing lower capex or dilutive funding; that would hurt suppliers after a lag even if reported capex initially stays elevated. Conversely, a disclosed multi-year external-capacity contract with a credible hyperscaler or enterprise customer would invalidate the near-term short thesis by converting an open-ended build into backlog.

Key Decisions for Investors

• Maintain a 1-3 month underweight/short bias in SPCX only on rallies; use a stop on a verified contracted-revenue or backlog disclosure sufficient to demonstrate data-center utilization, rather than on promotional targets. Risk/reward is favorable if the next filing shows further free-cash-flow deterioration or financing needs, but position size should reflect IPO volatility and uncertain float.
• Express relative quality via long MSFT / short SPCX over the next 3 months. MSFT has balance-sheet capacity and enterprise distribution to monetize AI through an installed cloud base; close the spread if SPCX reports credible third-party capacity contracts or MSFT materially cuts Azure AI capex guidance.
• Keep MU and SNDK as tactical beneficiaries of confirmed AI infrastructure orders, but do not add solely on aggregate capex commentary. Set an alert for supplier guidance that separates committed customer demand from inventory build; a reduction in AI-related bit-growth or storage demand would be the signal to cut exposure.
• For NVDA, treat SPCX as a marginal demand datapoint rather than a thesis driver. Hedge a long NVDA position with defined-risk downside around the next SPCX funding/capex update if broader evidence emerges that AI projects are being financed ahead of customer commitments.