SpaceX Targets Gas-Turbine Hot-Section Casting Bottlenecks to Secure Time-to-Power for AI Compute Through Vertical Integration

Overview

This report focuses on a potential pathway for SpaceX to address the “time-to-power” bottleneck for AI data centers. Morgan Stanley believes SpaceX may be building blade and vane casting capabilities in Bastrop, Texas to secure supply of critical gas-turbine hot-section components. These capabilities could support both terrestrial AI data-center gas turbines and Starship Raptor-engine turbopumps, creating dual-use scale benefits across its space and AI businesses.

Core views

The report's core conclusion is that, in the competition for AI infrastructure, value is determined not only by compute per watt or investment per dollar, but also by the speed at which energy can be converted into intelligence. Although gas turbines can be purchased, supplier schedules, component shortages, and high prices make the ability to bring power online on one's own timetable a scarce capability. If successful, SpaceX's vertical integration would improve its data-center expansion pace, cost control, and equipment-design autonomy; however, complex casting processes, equipment lead times, talent development, and qualification requirements mean that meaningful capacity is unlikely to be contributed quickly in the near term.

Analysis framework

The report combines SpaceX job postings, publicly disclosed turbine procurement commitments, industry order and capacity data, management interview content, and sum-of-the-parts valuation to assess how blade and vane supply bottlenecks could affect SpaceXAI compute expansion and long-term valuation.

Methodology notes

• Valuation methodsSum-of-the-parts valuation

  Values the Space, Connectivity Services, X & Grok, and Enterprise AI businesses separately and then aggregates them.

  The report's $300 price target consists of $8 for Space, $118 for Connectivity Services, $8 for X & Grok, and $165 for Enterprise AI.

• Valuation methodsLong-term cash flow valuation

  Assesses business values using a long-term forecast horizon, cost of capital, and terminal growth rate.

  The valuation date is June 30, 2027, with a forecast period through 2040, a WACC of 11.1%, and a cost of equity of 11.9%; Enterprise AI valuation additionally applies a 50% execution-risk discount.

• Strategic AnalysisVertical integration and bottleneck analysis

  Identifies critical supply-chain nodes that constrain scale expansion and assesses the time, cost, and supply-security benefits of internalizing production.

  The report identifies hot-section blades and vanes as the narrowest point in the gas-turbine supply chain and analyzes their shared constraint on AI data-center power supply and Raptor turbopump production.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• SPACE EXPLORATION TECHNOLOGIES CORP (SPCX.US)

  Covered company

  Strengths

  Possesses an asset base spanning space, connectivity services, and AI businesses; may gain control over the supply of critical hot-section castings through vertical integration; data-center gas turbines and Raptor turbopumps can share alloys, furnaces, and specialized talent.

  Weaknesses

  AI compute expansion is highly dependent on access to off-grid power and project deployment speed; process, yield, and qualification ramp-up for a new single-crystal casting facility are significantly challenging.

  Comparison

  Compared with buyers relying solely on external OEM production schedules, internalization can improve scheduling and design autonomy; however, compared with established casting suppliers, SpaceX lacks decades of accumulated volume-production process experience and an existing qualification system.

  Risks

  Power resources may not materialize as expected, casting-facility commissioning may be delayed, AI commercialization may be weaker than expected, capital expenditures may rise, financing dilution may occur, and regulatory delays may emerge.

• Gas-turbine blade and vane supply chain

  Critical upstream constraint and potential beneficiary segment

  Strengths

  AI-driven power demand has increased the strategic value of hot-section components; the industry is concentrated, qualification barriers are high, and capacity expansion is slow.

  Weaknesses

  New qualified capacity requires long-term capital investment, technical accumulation, and qualification, making it difficult to ease supply tightness quickly in the short term.

  Comparison

  Turbine assembly capacity can expand relatively quickly, whereas qualified hot-section casting capacity is more difficult to increase and is therefore more likely to become the industry's actual bottleneck.

  Risks

  Long-term new capacity additions or alternative power-supply solutions may ease supply tightness; if demand falls short of expectations, elevated prices and profit margins may decline.

Key data

• Rating and price targetOverweight, $300The current share price is $141.29, and the report's bull-case scenario is $600.
• Power potentially available by end-20273–4GWBased on reported but not fully confirmed procurement and related transactions; Morgan Stanley's model assumes 5GW, below management's stated 10GW target.
• AI infrastructure gas-turbine commitmentsApproximately $2.8bnIncludes $805mn in procurement agreements through 2029 and an approximately $2.0bn mobile gas-turbine acquisition arrangement.
• Revenue opportunity per GW of computeMore than $50bn/yearThe report's estimate of the commercialization opportunity for SpaceXAI compute.
• Global data-center power demandApproximately 55GW (2023) to 210GW (2028)The report expects demand to grow by approximately fourfold.
• Gas-turbine orders and pricing100GW in 2025; 117GW annualized based on 1Q 2026; approximately $3,000/kWThe report believes that orders, pricing, and prepayment requirements collectively reflect tight supply.
• Blade casting volume-production cycleAt least approximately 2 years, with stable capacity typically requiring more than 3 yearsSingle-crystal casting involves complex processes, specialized equipment, yield ramp-up, and aerospace qualification.

Impact & implications

If the Bastrop casting facility is built as planned and achieves qualified volume production, SpaceX could shorten the deployment cycle for AI infrastructure through internal supply, cooperation with original equipment manufacturers, or stronger procurement bargaining power, while spreading fixed costs across its space and AI businesses. The report believes that, after deducting the value of the Space and Connectivity Services businesses, the current share price implies only approximately $13 per share for Consumer and Enterprise AI, or roughly 1x 2028 EV/Sales, below certain emerging cloud-computing peers, thereby offering valuation re-rating potential.

Risks

• The Bastrop casting facility could be delayed by equipment lead times, talent constraints, process yields, or qualification issues, and initial volume production will take at least approximately two years.
• The non-grid gas-turbine capacity SpaceX can actually secure and its deployment progress remain unclear, and the report's identification of some supply arrangements is not company-confirmed.
• Weaker-than-expected monetization of Enterprise AI and Consumer AI could reduce the economic returns on compute expansion.
• Slower-than-expected Starship reusability progress or Starlink capacity and user growth could weigh on segment valuations.
• Rising compute cost per watt, capital expenditures, funding needs, and potential dilution could reduce shareholder returns.
• Regulatory approvals or infrastructure-construction delays could extend time-to-power.
• Morgan Stanley has participated in or co-managed SpaceX securities offerings during the past 12 months and has received investment-banking compensation from SpaceX, creating a potential conflict of interest.

What to watch

• Whether SpaceX formally discloses construction, equipment procurement, hiring, and commissioning timelines for its Bastrop blade and vane casting facility.
• Whether gas-turbine procurement agreements, APR Energy-related assets, and other power resources can support the model assumption of approximately 5GW by the end of 2027.
• Further company disclosures regarding its 10GW terrestrial-compute target, data-center locations, fuel supply, and grid-connected/off-grid power arrangements.
• Single-crystal casting yields, qualification progress, and whether internal supply partnerships are established with existing gas-turbine OEMs.
• Progress in Starship reusability, Starlink capacity expansion, enterprise AI customer acquisition, and declining AI infrastructure costs.
• Whether gas-turbine orders, delivery lead times, prices, and hot-section casting capacity expansion continue to reflect tight supply.