SpaceX (SPCX.US) shares rebound nearly 16% over three days; Deutsche Bank bets the rally isn't over! US$250 target points to AI computing power leasing taking the lead

Deutsche Bank AG's latest research report shows that the growth focus of SpaceX (SPCX.US), the "AI + space exploration" super-giant founded and led by Musk, is accelerating its shift from the Starlink satellite internet platform to ground-based AI computing infrastructure. Deutsche Bank expects AI computing power leasing to contribute more than half of the tech giant's fourth-quarter revenue; Edison Yu, a senior analyst at Deutsche Bank, raised his quarterly revenue forecast to US$24.5 billion, about 8.4% above the Wall Street consensus of US$22.6 billion, and expects gross margin to approach 60%. Supporting Deutsche Bank's judgment is the delivery capability and relatively high rental rates of the ready-made Colossus 2 super AI computing cluster. In Musk's view, orbital AI data centers are SpaceX's core business of the future, but in the view of Deutsche Bank analysts they are a longer-term uncertain incremental opportunityDeutsche Bank expects them to reach meaningful scale only by 2029-2030. In addition, Musk himself said at the Barron's investment conference on November 6 that SpaceX's annualized revenue run rate could exceed US$500 billion by December 2027, and made clear that this is his personal latest judgment rather than a specific growth commitment; the US$193.2 billion revenue figure forecast in Deutsche Bank's research report is full-year cumulative revenue for 2027, so the two are on different bases. As for the share price trend, SpaceX rebounded from US$123.62 on November 5 to US$143.09 on November 10, a cumulative gain of about 15.75% over three trading days; based on the latest closing price, the US$250 target price firmly maintained in Deutsche Bank's latest bullish research report implies massive potential upside of about 74.7%. Deutsche Bank is particularly optimistic about the pricing advantage of computing capacity already in production, believing that if new data center construction slows, existing facilities such as Colossus 2 may be able to maintain high rental rates for longer. Whether this window can be converted into shareholder returns depends on whether new GPUs can be powered up and delivered on time, whether customers renew contracts, and whether collections can cover construction and financing expenditures. Deutsche Bank expects SpaceX's 2027 capital expenditure to be about US$299.5 billion, equivalent to 1.55 times its full-year revenue forecast. Musk's envisioned US$500 billion is the annualized revenue run rate at the end of 2027, while Starlink ground 5G and orbital computing provide longer-term growth opportunities. The US$300 target price put forward by Morgan Stanley incorporates the synergy value of launch, communications and computing, while "Big Short" Burry's valuation challenge focuses on how much premium the market is paying for these yet-to-be-realized earnings. If a tech stock selloff pushes up single-stock correlations, unwinding crowded dispersion trades could also amplify short-term volatility. The operating indicators that will test the US$250 target price remain computing capacity delivery, renewal rental rates and cash flow after deducting capital expenditure. From rockets to GPU leasing, SpaceX's largest business this quarter may not be rockets When investors think of SpaceX, what comes to mind may be a rocket. But this quarter, more revenue may come from a far less eye-catching business: AI computing chips, or large-scale AI computing cluster leasing for the likes of Alphabet Inc. Class C and Anthropic. Through its Colossus 2 data center, SpaceX rents AI computing power to companies such as Alphabet Inc. Class C and Reflection AI. Wall Street calls this type of business a neocloud business. A Deutsche Bank AG forecast report shows that this AI computing power leasing business will contribute more than half of SpaceX's fourth-quarter revenue. The report was published after Elon Musk attended the Barron's investment conference on November 6. The report said Musk hinted that fourth-quarter revenue could grow about 70% from the third quarter. Analyst Edison Yu raised his revenue forecast from US$23.6 billion to US$24.5 billion, above the Bloomberg-compiled consensus of US$22.6 billion. The report said Musk also stated that a "similar situation" "maybe" would occur in the first quarter of 2027. If revenue grows another 70%, quarterly revenue would exceed US$40 billion, far above SpaceX's own guidance of US$33 billion to US$37 billion. Yu estimates that SpaceX's annualized revenue run rate will reach about US$104 billion by the end of 2026. He also expects fourth-quarter gross margin to approach 60%, above management's earlier indication of about 55%. The stock market responded positively to these developments. After falling nearly 20% in the previous month, SpaceX shares rose 8.6% on November 6 to close at US$134.27. The most striking detail in the report is a table listing SpaceX's AI computing contracts. Deutsche Bank analyst Yu expects that SpaceX's first large customer pays an average price of US$7.78 per GPU hour for different chips. According to the bank's estimates, Alphabet Inc. Class C uses NVIDIA Corporation's newer GB300 chips and pays about US$11.46 per GPU hour; another undisclosed customer using the same hardware pays about US$13.90. For future contracts, Deutsche Bank uses a price of US$14 per GPU hour in its model. Based on consensus forecasts, this is about 80% higher than the blended average price of SpaceX's first contract. Reflection AI's agreement shows how such contracts work. According to Data Center Dynamics, the startup pays about US$150 million per month to use Colossus 2's GB300 computing power; after the initial three months, either party may terminate the contract with 90 days' notice. This exit clause applies to both parties. Yu suggested that SpaceX could use the clause to renegotiate contract prices or reclaim computing capacity for its own AI business. Yu wrote: "SpaceX could trigger the 90-day termination clause in existing contracts, effectively charging customers higher prices." He also cautiously noted that Deutsche Bank has not incorporated these "more extreme outcomes" into its model. After the U.S. midterm elections, Yu also expects AI computing power prices to remain elevated for longer, because he believes data center construction will slow, with the impact mainly seen in 2028. Reduced supply means computing capacity already built can command higher prices. As a result, he raised his 2027 revenue forecast from US$184.4 billion to US$193.2 billion. The variable brought by Tesla, Inc.: could the merger price be lower? According to media reports, Tesla, Inc. and SpaceX have each set up committees to study a potential merger, and Musk has also called the idea "a good idea." Deutsche Bank analyst Yu previously assumed that SpaceX would need to pay a relatively high premium to Tesla, Inc. shareholders to secure shareholder vote support and avoid legal disputes. That expectation may have helped support Tesla, Inc.'s share price. Now, he is less certain. Yu wrote: "Tesla, Inc. shareholders may be willing to accept a lower price or premium to ensure the merger is completed faster." Yu maintains a "Buy" rating on SpaceX and a US$250 target price. Based on consensus forecasts, that implies about 86% potential upside from the November 6 closing price; however, the stock is still about one-third below its 52-week high of US$201.80. However, the scale of spending is also enormous. Deutsche Bank expects 2027 capital expenditure to be about US$299.5 billion, equivalent to about 1.55 times its revenue forecast for that year based on consensus forecasts. On orbital computing, Deutsche Bank is also more cautious than Musk. Musk expects AI satellites to begin launching in the third quarter of 2027 and to scale up in 2028, but Yu believes meaningful-scale computing will not emerge until 2029 or 2030. For now, the business story is more direct than space data centers: SpaceX is renting out AI chips at rising prices, and this leasing business is growing faster than Deutsche Bank expected just two weeks ago. Connecting the world, renting out computing power: the market's debate over SpaceX's valuation comes down to rental sustainability and capital returns SpaceX is trying to build both a nationwide mobile communications network and a large AI computing platform at the same time. Qualcomm has confirmed that Starlink Mobile will adopt the Dragonwing platform and plans to begin deploying a U.S. ground-based 5G network starting in 2027, building tens of thousands of sites over the multi-year agreement period. Satellites are responsible for expanding coverage in remote areas, while ground base stations handle high-capacity communications demand in densely populated areas, giving Starlink an opportunity to move from a satellite supplementary service to an integrated mobile communications business competing with the three major U.S. carriers; however, nationwide network construction and user acquisition still need to be completed step by step. The key to the computing business is converting equipment already in production into enough billable GPU hours. Deutsche Bank believes construction constraints may slow new supply, thereby extending the high-rental window for ready-made computing capacity. However, the US$14/GPU hour used for future contracts is a model assumption and is not a like-for-like comparison with the US$7.78 cross-chip blended average price of the first contract. The two-way exit clause in the Reflection contract also means revenue sustainability must be tested through customer renewals. High rental rates can increase the revenue potential of assets already in production, but Deutsche Bank forecasts 2027 capital expenditure of as much as US$299.5 billion, and ultimate shareholder returns still depend on delivery, collections and financing costs. This is also where the divergent outlooks of Morgan Stanley and Michael Burry, the "Big Short" who once accurately predicted the 2007 U.S. subprime crisis, convergeMorgan Stanley's US$300 target price incorporates the synergy value of launch capability, global connectivity and AI computing; Burry questions whether the market is paying too high a price for growth yet to be realized. *Morgan Stanley's own US$75 bear case also includes assumptions of Starship delays and AI deployment and monetization falling short of expectations. From this perspective, Deutsche Bank emphasizes revenue that ground-based computing is currently realizing, Morgan Stanley places more emphasis on long-term platform synergy, and Burry's challenge reminds investors to test whether these returns can cover the capital required for expansion.