Morgan Stanley: SpaceX (SPCX.US) Q3 earnings are just an "appetizer"; Starship's 15th flight and the "physical intelligence economy" are the real catalysts for the stock price.

Ahead of SpaceX's (SPCX.US) third-quarter earnings report, due after the U.S. market close on Tuesday Eastern Time, Morgan Stanley published a research note maintaining its "Overweight" rating and $300 price target on SpaceX. Analyst Adam Jonas's team believes the core value of SpaceX that investors underestimate lies not in finely beating expectations in any single quarter, but in the central role it is building in the "Physical Intelligence Economy" (PIE). The bank believes the eventual scale of PIE could be several times global GDP and carry major implications for national sovereignty; therefore, SpaceX's share price direction will depend less on the fine details of the Q3 report and more on whether the PIE still fermenting in Musk's plans can deliver some substantive progress. Q3 expectations: Revenue and compute metrics are the focus Morgan Stanley expects SpaceX's total revenue for the third quarter of 2026 to be $13.005 billion, slightly above the consensus of $12.810 billion; EBIT is expected at $2.363 billion versus consensus of $2.396 billion; adjusted EBITDA is expected at $6.931 billion versus consensus of $6.837 billion; adjusted diluted EPS is expected at $0.19 versus consensus of $0.16. By segment, Morgan Stanley expects Space revenue of $1.570 billion versus consensus of $1.458 billion; Connectivity revenue of $4.857 billion versus consensus of $4.917 billion; AI revenue of $6.578 billion versus consensus of $6.435 billion. The AI segment includes $699 million in advertising revenue and $5.879 billion in AI solutions and infrastructure revenue. On profitability, the report expects the Space segment to remain at an EBIT loss of $507 million, Connectivity EBIT of $1.866 billion, and AI EBIT of $1.005 billion. Gross margin is expected at 56.3%, with an operating margin of 18.2%. On key operating metrics, the market expects 14.293 million consumer Starlink subscribers, while Morgan Stanley expects 14.288 million; consumer monthly ARPU consensus is $66.74, while Morgan Stanley expects $63.64; period-end compute capacity consensus is 1.7GW, while Morgan Stanley expects 1.62GW; revenue per watt (Rev/Watt, annualized on average compute) consensus is $16.61, while Morgan Stanley expects $17.42. The bank stresses that revenue per watt may be one of the most important yet hardest-to-model variables in SpaceX's earnings, because new cloud contracts have short terms (about six months) and the inference economics in internal models remain unclear. Investor sentiment: Leaning positive, but off-quarter overhangs weigh on valuation The report notes that investor positioning sentiment ahead of the Q3 report has been like a "roller coaster" and is hard to gauge precisely. Overall, however, investors are slightly positive on the numbers: compute pricing is only rising, and market expectations may not yet fully reflect the value from Cursor, Grok Bot, and announced new cloud orders. That said, overhangs beyond quarterly results are still keeping a sizable portion of investors from turning more positive. The bank specifically notes that SpaceX's share price once surged from below $150 to $172 in about two weeks before pulling back, with little news flow during the period, and the total volume of investor questions received during that time was comparable to the volume received last Friday ahead of the earnings release, indicating sensitive sentiment and wide disagreement. On strategy, Morgan Stanley says outright that "market complacency is the opportunity." The report argues that the rest of the year will enter a catalyst-rich, upward-skewed path, starting with Starship's 15th test flight and a possible ship catch (using the launch tower's mechanical arms to catch the Starship upper stage), expected in the first half of November. This could be SpaceX's biggest catalyst since its IPO and is expected to de-risk the total addressable market (TAM) of its major businesses. On valuation, based on the current share price of $146.03, Morgan Stanley believes SpaceX's stock already reflects $127 per share of the Space and Connectivity opportunity in its model, covering most or even all of that value, leaving only $19 per share for the AI business. That equates to about 2x 2028 EV/Sales for the new cloud business. Core debate: 5GW or 10GW of compute by end-2027? The report shows investors lack confidence that SpaceX's compute capacity by end-2027 will be "closer to 10GW than 5GW," mainly because of uncertainty over future data center siting, power access, and whether buildout pace can outrun the industry. The sell-side consensus is 6GW, Morgan Stanley's model is 4.9GW, and buy-side expectations are roughly in the 4-6GW range. Although Musk said on the last earnings call that it is "closer to 10 than 5," the report says it has not heard customers firmly betting on close to 10GW. The report also notes that many investors still tend to view one SpaceX business in isolation based on their own backgrounds (aerospace and defense, internet, telecom) and are only just beginning to understand the interconnectedness of the businesses. Apart from a handful of investors close to Musk's companies, few seriously discuss "physical AI" as a future opportunity, even though Morgan Stanley sees it as the most obvious TAM expansion direction for AI and satellite communications. Questions about Terafab also remain confined to sizing the company's near-term capex burden. On the new cloud business, the bank says recent transaction prices are about $30-50 per watt, and the industry pricing index is still rising before year-end; although hyperscalers and new cloud players are expected to bring more than 30GW of incremental compute capacity in 2027, helping balance supply and demand. The company may lease up to half of its 2GW of compute capacity at end-2026 to new cloud customers, after announcing two new deals on the Q2 earnings call and at industry conferences; as capacity rises above 5GW by end-2027, further deal announcements are worth watching. The report also notes the company has repeatedly said the payback period on new cloud compute investment is about one year, indicating compute supply and demand will remain quite tight through next year. Earnings call and subsequent catalysts: Starship, Starlink V3, Grok, and more new cloud business The report expects SpaceX management to update the verbal guidance from the Q2 call: Q4 capex expected similar to Q3; year-end compute "above 2GW"; initial guidance for end-2027 compute of 5-10GW, with Musk previously saying "closer to 10 than 5." The company may also provide details on the date and objectives of Starship's 15th flight and the cadence of Starlink V3 satellite deployment via Starship. The report estimates each Starship launch equals more than 1-2 months of Falcon 9 Starlink launch capacity, depending on the load factor. On the call, the AI business is a key area to watch. Several AI milestones occurred in Q3, including completion of the Cursor acquisition, the Grok Bot launch, and marked improvements in Grok 4.6 and 4.7, moving closer to the model frontier. This quarter may be the first real window for investors to observe inference economics. The bank suggests watching Grok and Cursor user growth, token volumes, potential new enterprise deals, and penetration into existing Cursor customersabout two-thirds of Cursor's customers are Fortune 500 companies. On Starlink, the report expects the company to provide more V3 planning details after deploying the first 26 V3 satellites on the 14th test flight, including deployment cadence and balance with V2 mobile satellites. SpaceX recently announced the purchase of 800MHz spectrum from Grain Management, pending FCC approval. Morgan Stanley believes the direct-to-cell satellite business model remains relatively uncertain and expects management to provide incremental information, but may deliberately keep options open. On Starship, the 15th flight is expected in early November, with the market watching whether a ship catch will be attempted. If successful, it could be SpaceX's biggest share price catalyst since its IPO and a major validation point for reusability, paving the way for a higher launch cadence in 2027. The company may also give general expectations for future test flights without a firm timetable, including the first simultaneous catch of both ship and booster in the same flight, the first launch from Florida, and more Starlink satellite deployment plans. The 16th flight is expected before year-end; if no catch is attempted on the 15th, the 16th could be the first catch opportunity; if the 15th succeeds, the first same-flight catch of ship and booster could be seen. In addition, Grok 4.8, 4.9, and 5.0 are expected to be released before year-end. Musk said on X on September 14 that Grok 4.8 will be markedly improved over 4.7, 4.9 "may reach the level of current frontier models," and 5.0 "may be better than anything, let's see." More new cloud deals are also seen as potential catalysts: Morgan Stanley's model expects SpaceX compute at about 2.1GW at year-end, rising to about 5GW in 2027, opening room for more deals. Financing, lock-up, and other points of focus The report views additional financing needs as a key overhang on sentiment, especially after recent new cloud player debt events. Morgan Stanley's model expects SpaceX to issue $60 billion, $80 billion, and $90 billion of bonds in 2027, 2028, and 2029, respectively, currently assumed to be plain investment-grade issuance, though the company has multiple financing options. The report calls this the biggest risk to its model and does not want to downplay it; however, a hypothetical scenario in which SpaceX cannot raise financing could have broader implications for the economy and investors' other holdings. On the lock-up, investors no longer treat the staggered expiration of restricted shares as a core concern, as the impact of expirations so far has been relatively limited. After the Q3 report, 1.3 billion shares will unlock, with a total of 2.4 billion shares unlocking before year-end; however, more than 20% of these may be absorbed by passive index buying, possibly more than the shares actually sold. The report also mentions that on the new Roadster jointly developed by Tesla, Inc. and SpaceX, there is no more new information beyond acknowledging that the demonstration was impressive. It is the first major product jointly designed by the two companies and a major showcase of their engineering capabilities and collaboration potential.