Two major investment banks have published research reports expressing optimism about the potential returns on SpaceX (SPCX.US) capital investments as the company accelerates its expansion into satellite internet and artificial intelligence infrastructure.
Morgan Stanley estimates that by 2030, a single Starship launch fully loaded with 50 Starlink V3 satellites could generate approximately $400 million in incremental annual revenue, with a cost recovery period of only about 19 weeks based on incremental operating profit. Bernstein believes that if SpaceX can seize the current window of tight AI computing supply and elevated leasing prices, its ground-based data center investment could be recouped in less than one year. Both banks' analyses point to a key characteristic of SpaceX's business model: by accelerating infrastructure deployment, the company can achieve investment returns within a relatively short timeframe, then leverage the established satellite network and computing facilities to generate long-term profits. However, both institutions also emphasize that these returns are highly dependent on factors such as Starship reusability technology, data center construction speed, and AI computing market pricing. Morgan Stanley maintains an "Overweight" rating on SpaceX with a $300 price target, while Bernstein maintains an "Outperform" rating with a $268 price target. Based on the closing price of $143.50 on November 18, the two targets imply potential upside of approximately 109% and 87%, respectively.
Starship Launch Economics Take Center Stage: Cost Recovery in 19 Weeks
In a research report published on November 18, Morgan Stanley noted that Starship's greater payload capacity is expected to significantly improve the economics of the Starlink satellite internet business. The bank estimates that a single fully loaded Starship launch can deploy 50 third-generation Starlink satellites, each capable of generating approximately $8 million in incremental annual revenue, totaling about $400 million. The total communication capacity of the satellites on a single launch is approximately 50 Tbps, of which commercially sellable capacity is estimated at about 2.5 Tbps. Meanwhile, Morgan Stanley estimates that by 2030, the variable cost of a single Starship launch will be approximately $100 million, including $60 million in launch costs and $40 million in satellite payload costs. Based on these assumptions, the cost of a single launch equates to approximately 13 weeks of incremental revenue from the satellites it deploys. Calculating at a 70% incremental EBIT margin, the approximately $280 million in annual incremental EBIT corresponds to a cost recovery period of roughly 19 weeks. This return efficiency is notably superior to the bank's estimates for emerging cloud computing businesses. A comparison chart in the Morgan Stanley report shows that under its assumed computing leasing prices, capital expenditure, and margin assumptions, emerging cloud computing businesses have revenue and EBIT payback periods of approximately 41 weeks and 46 weeks, respectively. Morgan Stanley also noted that each Starlink V3 satellite has approximately 10 times the downlink communication capacity of existing V2 Mini satellites. Because Starship can carry more satellites at once, the additional communication bandwidth from a single fully loaded launch is equivalent to 20 to 25 times that of a single Falcon 9 launch, roughly equal to two to three months of the latter's Starlink satellite deployment volume. As launch frequency increases and rocket reusability technology matures, the bank expects SpaceX's internal variable launch cost per unit payload to decline from approximately $1,000 per kilogram in 2026 to $500 by 2030, and further to approximately $140 by 2040.
Bernstein: Seizing the AI Computing Shortage Window, Ground Data Centers Could Pay Back in One Year
Unlike Morgan Stanley's focus on Starship and Starlink, Bernstein's research report published on November 19 focused on the business logic behind SpaceX's large-scale construction of ground-based AI data centers. Bernstein noted that Musk plans to build at least 20 gigawatts (GW) of power infrastructure by the end of 2027 and significantly expand computing deployment scale. Although this expansion plan has raised market concerns about investment costs and construction speed, the bank believes that the current tight AI computing supply provides SpaceX with an opportunity to sell short-term computing services at elevated prices. The report shows that SpaceX has already secured computing deals at price levels of $30 and $50 per watt this year, and company management believes there are still opportunities to obtain prices above $50 per watt in the future. Bernstein believes that while this price level is difficult to sustain long-term, it is not entirely detached from market reality. Emerging cloud computing service providers such as CoreWeave (CRWV.US), Nebius (NBIS.US), and IREN (IREN.US) all reported strengthening computing demand and pricing in the third quarter, with recently signed contracts priced at approximately $20 to $45 per watt. The bank estimates that building 1 GW of computing facilities requires approximately $35 billion in investment. If SpaceX can rapidly sell newly added computing services at $50 per watt, its investment could be recouped through revenue in less than 12 months. Bernstein emphasizes that the key is not how many years the $50 per watt price can be maintained, but whether SpaceX can complete construction before more competitors' capacity enters the market and lock in high-priced short-term contracts. The bank expects that as large numbers of newly built data centers gradually come online between 2028 and 2029, the tight supply-demand situation for computing will gradually ease, and prices will face downward pressure. By 2030 to 2031, raw computing service prices may fall back to approximately $13 to $15 per watt. Therefore, for SpaceX, construction speed will directly determine the investment return of its ground-based AI business.
Ground and Space Working in Tandem: SpaceX Bets on Next-Generation AI Infrastructure
Bernstein believes that SpaceX's long-term strategy is not simply to earn computing leasing revenue from ground-based data centers, but to use ground facilities to complement future orbital data centers. Under the bank's vision, SpaceX can leverage the high prices brought by the computing shortage between 2027 and 2028 to recoup its ground infrastructure investment as quickly as possible. Afterward, even if computing leasing prices decline, the company can still use these facilities to train Grok AI models or continue selling computing power to third parties at market prices. As orbital data centers are gradually deployed, ground facilities can also handle data storage, AI training, and backup tasks, reducing customer concerns about relying entirely on space-based computing infrastructure. Bernstein estimates that SpaceX's ground computing installed capacity will reach approximately 20 GW by 2029, while orbital computing capacity is expected to reach approximately 27 GW by 2031, surpassing ground computing scale. By comparison, SpaceX's own 2031 orbital computing target is more aggressive, reaching 77 GW. The bank notes that orbital data centers may reduce the land approval, power supply, and cooling cost pressures faced by traditional data centers, but their economics still need to account for factors such as launch costs and equipment residual value. Unlike ground data centers, once orbital facilities reach the end of their service life, their equipment residual value may be close to zero. Morgan Stanley's forecast of continuously declining Starship launch costs provides another potential support for SpaceX's future deployment of orbital data centers. If Starship can achieve high-frequency reusability, the deployment cost of space infrastructure is expected to decrease further.
Expansion Speed Remains the Biggest Test: 2028 May Be a Key Turning Point
Although both investment banks are optimistic about SpaceX's long-term potential, achieving the above commercial prospects still faces significant challenges. Bernstein noted that SpaceX previously set a record of deploying 210 megawatts of facilities in 91 days when building the Colossus II data center in the Memphis and South Haven areas, with the second cluster's construction period further shortened to 64 days. However, this speed was partly due to the renovation of existing industrial facilities and simultaneous construction of power supply infrastructure. As the company expands to larger new-build projects, whether it can replicate its previous construction efficiency remains uncertain. Bernstein also expects that around 2028, large US technology companies and cloud computing service providers will bring substantial new data center capacity to market. According to the bank's project pipeline analysis, the scale of related new facilities is close to 75 GW, and even after accounting for project delays and actual landing probabilities, there remains approximately 55 GW of potential new capacity. This means that 2028 may become an important turning point in the supply-demand relationship of the AI computing market. If SpaceX fails to complete construction in time, its newly invested computing facilities may face the risk of declining leasing prices and extended payback periods. By contrast, the core risk that Morgan Stanley focuses on mainly comes from Starship technology verification and commercialization progress. The bank expects SpaceX to achieve a once-per-week Starship launch cadence by the end of 2027, but this target still depends on technological breakthroughs such as rocket and booster reusability and full satellite payload deployment.
Two Major Investment Banks Maintain Bullish Stance: Price Targets Up to $300
On valuation, both institutions believe SpaceX has further upside potential, but their forecasts for future business development differ. Morgan Stanley maintains its $300 price target, with its sum-of-the-parts valuation attributing $170 to enterprise AI business, $111 to satellite connectivity business, $11 to X and Grok business, and $8 to space business. Bernstein maintains its $268 price target. While the bank raised its computing pricing assumptions for the coming period, it simultaneously slowed its forecast for ground data center construction progress and slightly lowered its 2027 earnings per share estimate from $1.44 to $1.43. The common judgment of both investment banks is that SpaceX is building a mutually supportive infrastructure system through Starship, Starlink, and AI data centers. Starship is expected to reduce space network deployment costs, while ground data centers may leverage short-term computing supply-demand imbalance to accelerate investment recovery and provide support for future orbital AI computing business. However, whether it is Starship's theoretical 19-week payback period or ground data centers' less-than-one-year investment recovery expectation, both are built on specific assumptions about revenue, pricing, costs, and execution efficiency, and actual performance remains to be verified.
Comments