TechFlow Logo
Login/ Sign up
ETH Gas
Gwei
Fear
gas
Morgan Stanley Research Report Analysis: Blades and Impellers, SpaceX Breaks Through AI Data Center Power Supply Timeliness Bottleneck

Morgan Stanley Research Report Analysis: Blades and Impellers, SpaceX Breaks Through AI Data Center Power Supply Timeliness Bottleneck

2026.08.18
Share
Morgan Stanley maintains SpaceX Overweight rating and $300 price target, believing the pullback after lock-up expiration provides an entry opportunity.

By: Rita

The compute race for AI data centers is shifting from chips to power. SpaceX has secured chip supply, while gas turbine blades and impellers are becoming the new bottleneck.

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 globally can produce the hot-end blades and impellers required for large gas turbines, and orders are booked until 2030. SpaceX's solution is vertical integration: building its own foundry to meet the needs of both data center turbines and Starship Raptor engine turbopumps.

Morgan Stanley maintains an Overweight rating and a $300 price target for SpaceX, believing the pullback after the lock-up period expires provides an entry opportunity.

Building Its Own Foundry to Break Through Power Supply Bottlenecks

Four positions related to the Bastrop foundry recently appeared on SpaceX's recruitment page, involving automation, operations, and materials engineering. Morgan Stanley considers these job listings important clues to SpaceX's strategic direction.

Musk revealed his intent in a February podcast: to get sufficient power online, SpaceX and Tesla may need to manufacture turbine blades and impellers in-house. Only three foundry companies globally produce these things, and orders are severely backlogged.

Gas turbine hot-end components must operate under extreme high temperature and pressure, involving complex single-crystal casting processes. This is the narrowest bottleneck for turbine capacity. The order books of global mainstream turbine manufacturers are booked until 2029 to 2030, and EBITDA margins for new orders have risen from historically about 5% to over 30%. To lock in capacity, customers are placing orders at about $3,000/kW (about $1,000/kW in the first half of the 2020s) and paying about 25% as a down payment.

There are three layers of logic behind SpaceX building its own foundry: breaking free from supplier timelines, optimizing product performance through first principles, and serving both the aerospace and AI businesses with the same foundry to amortize fixed costs.

The Next Stop in the Compute Race is Power

The compute scale of SpaceX AI depends on two variables: chips and power.

The chip side is basically resolved. An exclusive agreement with NVIDIA locks in chip supply for at least 10GW. The power side remains unclear. Musk proposed a 10GW target by the end of 2027 during the Q2 earnings call; Morgan Stanley is currently modeling only 5GW, and the gap lies in the uncertainty of power supply timing.

Morgan Stanley identifies that SpaceX has locked in about 3 to 4GW of gas turbine supply through procurement agreements and Musk's personal acquisition of APR Energy (a mobile power supplier with a fleet exceeding 11GW), with total publicly disclosed commitments approaching 8GW. The actual constraint lies in blades and impellers, rather than complete gas turbine units.

Each GW of compute represents an annual revenue opportunity of about $50 billion. Morgan Stanley's SOTP price target breakdown is: Space business $18, Connectivity business $118, X and Grok business $8, Enterprise AI business $165. With the current stock price at about $141, after deducting $127 for Space and Connectivity, the market prices Enterprise AI at only about $13, approximately 1x 2028 EV/Revenue, lower than the valuation level of new cloud service provider peers.

How to Realize GW-Scale Compute

The Morgan Stanley team estimates that the four major hyperscalers will add about 25GW of compute in 2027 (excluding TPU). If SpaceX AI reaches 5GW by the end of 2027, the scale will be about one-fifth of the total new capacity of the four major cloud providers.

Upside risks include Starship reusability progress exceeding expectations, accelerated Starlink capacity growth, Enterprise AI monetization exceeding expectations, and accelerated Cursor ARR. Downside risks include slowed Starship reusability pace, slowed Starlink user growth, Enterprise AI monetization falling short of expectations, longer power supply cycles, and higher capital requirements.

SpaceX is replicating its vertical integration experience in the aerospace field into the AI infrastructure field. Morgan Stanley believes the selling pressure after the lock-up period expires provides an entry window for long-term investors. If SpaceX can prove it will push compute scale to 5GW or even higher by the end of 2027, the Enterprise AI business valuation will be significantly revised upward.

Disclaimer

This article is a compilation and interpretation by TechFlow Research of a third-party brokerage research report (Morgan Stanley, August 16, 2026), combined with organized public market information. The ratings, price targets, earnings forecasts, and related judgments cited in the text are the views of the brokerage's analysts, represent only the position of their affiliated institution, do not represent the views of TechFlow Research, and do not constitute any investment advice.

The market carries risks; decisions must be made independently. This article should not be used as a basis for buying or selling any securities.