SpaceX's stock is experiencing a rebound, trading around $140, yet a notable five-star analyst from Phillip Securities, Glenn Thum, has issued a bearish "Sell" rating with a $75 price target. This contradicts the prevailing optimism on Wall Street, which projects an average price of $227, indicating a substantial potential downside. Thum's skepticism stems from SpaceX's significant capital expenditure in AI, primarily driven by short-term cloud service agreements, and the considerable revenue reliance on a single AI customer. The core argument is that the current valuation of SpaceX may be overestimating the long-term stability and profitability of these rapidly growing, yet potentially volatile, AI-related revenues.

Recent developments have shown SpaceX shares regaining ground, boosted by increased institutional interest. Filings reveal that tech giants like Nvidia and Alphabet hold substantial stakes in SpaceX, signaling strong backing from major players. For instance, Nvidia disclosed owning over 122 million SpaceX shares worth approximately $21 billion as of June 30, while Alphabet reported holdings of 551.2 million shares valued at nearly $94.2 billion. This institutional support has contributed to the stock's resilience, even in the face of significant lock-up releases of restricted shares. A previous release of 911.5 million shares on August 6 did not trigger the anticipated sell-off, and the stock recovered above its initial public offering price of $135.

Despite this broader market confidence, Glenn Thum's assessment stands out due to his impressive track record, boasting an 89% success rate and an average return of 20.8% across his 106 tracked calls. Thum's primary concern centers on the nature of SpaceX's burgeoning AI revenue. While Elon Musk envisions AI as potentially surpassing other SpaceX revenue streams, reporting a 247% year-over-year increase in AI revenue in Q2, largely from cloud service agreements, Thum views these contracts as inherently unstable. These agreements typically involve monthly fees and can be terminated with just 90 days' notice, lacking the long-term commitment necessary to justify a premium valuation.

Furthermore, Thum highlights a significant concentration risk within SpaceX's AI segment, where a single AI customer accounted for 19.5% of Q2 sales, up from less than 10% a year prior. This dependency makes SpaceX vulnerable if this major client reduces spending or withdraws. The substantial capital expenditure, reaching $18.4 billion in Q2—nearly 2.4 times its quarterly sales—to expand compute capacity (projected to exceed 2 gigawatts by December) further exacerbates the risk. Thum argues that SpaceX is investing heavily as if AI demand is perpetual, without sufficient long-term contractual assurances. For the stock to achieve a higher re-rating, these short-term compute deals would need to evolve into multi-year commitments.

SpaceX's valuation is also complex due to its diverse business segments. In Q2, the company generated $7.81 billion in revenue, with connectivity products (primarily Starlink) contributing $4.29 billion (approximately 55%), AI generating $2.56 billion (about 33%), and space products adding $962 million. While AI is rapidly becoming the second-largest revenue driver, Thum believes investors are overvaluing sales underpinned by easily cancellable contracts. The current stock price of $140 implies a forward non-GAAP earnings multiple of 373 times, significantly higher than the sector average of 13.4 times. Although long-term projections by GuruFocus show substantial earnings and revenue growth through 2030, this valuation hinges on the rapid conversion of sales growth into strong profitability. If AI contracts falter or capital expenditures remain elevated with narrow margins, Thum's cautionary stance on the stock's future performance could prove prescient, transforming it into a high-risk, duration-dependent investment.