Written by: Rita
The AI data center computing race is shifting from chips to electricity. SpaceX has secured chip supplies, and turbine blades and impellers are becoming the new bottlenecks.
In a research report released on August 16, Morgan Stanley analyzed SpaceX’s strategic intent behind building a new foundry in Bastrop, Texas. Only three foundries worldwide can produce hot-section blades and wheels required for large gas turbines, with orders booked through 2030. SpaceX’s solution is vertical integration—building its own foundry to meet demand for both data center turbine engines and Starship Raptor engine turbopumps.
Morgan Stanley maintains an overweight rating and a $300 price target on SpaceX, viewing the pullback after the lock-up expiration as an opportunity to enter.
Build your own minting facility to overcome power supply constraints
SpaceX's careers page recently listed four positions related to the Bastrop foundry, involving automation, operations, and materials engineering. Morgan Stanley believes these job postings are significant indicators of SpaceX's strategic direction.
Musk revealed on a February podcast that to ensure sufficient power is available, SpaceX and Tesla may need to manufacture turbine blades and impellers in-house. Only three foundries worldwide produce these components, and orders are severely backed up.
Gas turbine hot-section components must operate under extreme high temperatures and pressures, involving complex single-crystal casting processes. This is the tightest bottleneck in turbine production capacity. Global leading turbine manufacturers’ order books are now booked through 2029 to 2030, and the EBITDA margin on new orders has risen from a historical ~5% to over 30%. Customers are placing orders at approximately $3,000 per kilowatt (compared to ~$1,000 per kilowatt in the first half of the 2020s) and paying upfront deposits of around 25%.
SpaceX's rationale for building its own foundry consists of three layers: escaping supplier timelines, optimizing product performance through first-principles thinking, and sharing the same foundry to serve both aerospace and AI businesses, thereby spreading fixed costs.
The next frontier of the hash rate race is electricity.
The computing power of SpaceXAI depends on two variables: chips and electricity.
The chip side has been largely resolved, with an exclusive agreement with NVIDIA securing chip supply of at least 10 GW. The power side remains unclear. Musk proposed a 10 GW target by the end of 2027 during the Q2 earnings call, while Morgan Stanley currently models only 5 GW—the gap reflects uncertainty around power availability timing.
Morgan Stanley identified that SpaceX has secured approximately 3 to 4 GW of gas turbine supply through procurement agreements and Elon Musk’s personal acquisition of APR Energy (a mobile power provider with a fleet exceeding 11 GW), with publicly disclosed commitments totaling nearly 8 GW. The actual constraints lie in blades and rotors, rather than complete gas turbine units.
Each GW of computing power represents an annual revenue opportunity of approximately $50 billion. Morgan Stanley’s SOTP target price breakdown is as follows: Space business at $18, Connectivity business at $118, X and Grok business at $8, and Enterprise AI business at $165. The current stock price is approximately $141; after deducting the $127 attributed to Space and Connectivity, the market is pricing the Enterprise AI business at only about $13, equivalent to roughly 1x EV/revenue for 2028—below the valuation levels of peer next-generation cloud service providers.
How to realize computing power on the scale of billions of watts
Morgan Stanley estimates that the four major hyperscalers will add approximately 25 GW of computing power by 2027 (excluding TPUs). If SpaceXAI reaches 5 GW by the end of 2027, its scale would be about one-fifth of the total new capacity added by the four major cloud providers.
Upside risks include faster-than-expected Starship reusability progress, accelerated Starlink capacity growth, stronger-than-expected enterprise AI monetization, and faster Cursor ARR growth. Downside risks include slower Starship reusability cadence, slowed Starlink user growth, weaker-than-expected enterprise AI monetization, longer power cycle timelines, and higher capital requirements.
SpaceX is applying its vertical integration expertise from the aerospace industry to the AI infrastructure sector. Morgan Stanley believes that the post-lockup selling pressure presents an entry opportunity for long-term investors. If SpaceX can demonstrate that it will scale its computing power to 5 GW or higher by the end of 2027, the valuation of its enterprise AI business could be significantly revised upward.
Disclaimer
This article is a compilation and interpretation by Chaoxiang Research of a third-party brokerage research report (Morgan Stanley, August 16, 2026), combined with publicly available market information. The ratings, price targets, earnings forecasts, and related judgments cited herein reflect the views of the brokerage’s analysts and represent the position of their respective institution only; they do not reflect the views of Chaoxiang Research nor constitute any investment advice.
The market carries risks; make decisions independently. This article should not be used as a basis for buying or selling any securities.