Elon Musk's New Obsession Is Actually Crazy

Summary

The video examines why SpaceX is pursuing its own gas-turbine blade and vane casting capability while continuing to buy complete turbines. The presenter connects this effort to the power needs of SpaceX’s AI data centers and asks whether in-house production could bring electricity online sooner.

Why turbine blades matter to SpaceX’s AI expansion

• The presenter cites SpaceX CFO Bret Johnson’s forecast of roughly $100 billion in annual recurring revenue pace by the end of 2026, and Elon Musk’s projection of as much as $1 trillion by the end of 2027. “Annual recurring revenue” here means multiplying the revenue pace for one month by 12; it is not the total revenue collected during that year.
• SpaceX operates large AI data centers, including Colossus sites in Memphis and Southaven. The video says SpaceX had 1.7 gigawatts of computing capacity at the end of September 2026, with a target of 13 gigawatts by the end of 2027.
• Musk has described power, cooling, and infrastructure—not chips—as major limits on this expansion. More computing capacity requires more electricity, and on-site gas turbines are one way to supply it.
• The presenter calls this a “last part problem”: a project can be largely ready but unable to operate until a crucial component arrives. Musk says turbine blades and vanes can trail the rest of a turbine by 12–18 months, potentially delaying power projects.

What the turbine parts do—and why they are hard to make

A gas turbine compresses air, mixes it with fuel, and burns the mixture. Hot gas passes through stationary vanes, which direct it onto rotating blades. The blades turn a shaft; some of that energy drives the compressor, and the remainder can power an electricity generator.

The hot-section parts must retain their shape and strength while exposed to:

• High temperatures and mechanical loads, which can cause gradual deformation known as creep.
• Repeated heating and cooling, which creates thermal stress.
• Chemical attack from hot gases.

The video notes that turbine gas can reach about 2,300°F, while some critical metals tolerate only around 1,500°F. Blades therefore rely on carefully selected nickel-based superalloys and cooling systems. Internal passages carry compressor air through the blade, and small holes can release air over its surface for film cooling. A ceramic exterior coating also slows heat transfer. Cooling improves durability but uses air that could otherwise contribute to turbine performance.

How a blade is cast

The presenter walks through the manufacturing process:

1. A ceramic core is shaped to define the blade’s internal cooling passages.
2. Wax is formed around the core to create the blade’s exterior shape.
3. A ceramic shell is built around the wax. The wax is then removed, leaving a cavity with the core inside.
4. Molten alloy is poured into the space between the shell and core. A controlled, often vacuum-based process helps protect reactive alloy ingredients.
5. The foundry controls how the metal solidifies. Directional solidification aligns grains; single-crystal casting uses a selector to produce one continuous crystal, avoiding grain boundaries.
6. The shell and core are removed after solidification. The part then undergoes heat treatment, finishing, precise cooling-hole production, coating, and inspection.

A blade can fail its intended purpose even if its exterior looks right: the core may shift, cooling passages may be wrong, or defects such as pores or unwanted grains may compromise it. The presenter emphasizes that factory activity is not the same as accepted production. Repeatability, inspection, and customer qualification are essential before a new source can reliably affect delivery schedules.

Single-crystal construction is not used for every industrial turbine blade. The video cites Howmet’s description of industrial blades as mostly conventional-grain or directionally solidified, with single-crystal parts making up a smaller share; the required process depends on the component.

SpaceX’s plan and what remains unproven

• SpaceX is not described as building the entire turbine. Its stated plan is to make blades and vanes through casting while continuing to buy larger turbine systems.
• The presenter says SpaceX has had an in-house superalloy foundry since at least 2019 and has experience with hot-gas rocket-engine hardware. Musk has described an alloy called SX500 for the Raptor engine’s turbopump environment. The video cautions that its chemistry is undisclosed and there is no verified evidence that SX500 is intended for industrial turbine blades.
• SpaceX job postings in Bastrop are presented as evidence of work on power-generation components, including nickel-superalloy castings, wax-and-ceramic casting, and an industrial gas-turbine blade-and-vane line.
• Tesla and SpaceX also share materials expertise, according to a report cited in the video. However, Tesla’s aluminum high-pressure die casting for vehicle bodies is a different process from ceramic-shell casting of turbine blades.
• The presenter stresses that SpaceX’s claimed 12–18-month power-delivery advantage is unproven. Analysts cited by the Wall Street Journal estimate that establishing a single-crystal casting line from scratch can take four years or more. The video says no SpaceX industrial turbine blade had been publicly shown or verified as shipped, and public hiring information does not establish accepted production volumes or qualification.

Industry context and alternatives

• Turbine manufacturers must supply both new installations and replacement parts for existing plants. Siemens Energy says about 70% of its blades and vanes go into service, illustrating the ongoing maintenance demand.
• GE Vernova has said that meeting its end-of-2026 bookings target would leave it mostly sold out through 2030. Established suppliers are also investing in added capacity, but new furnaces must become part of a repeatable production process before they increase finished-turbine deliveries.
• Other AI and data-center companies described in the video generally use partner-based approaches. Meta’s Hyperion project involves Entergy building combined-cycle power plants; Crusoe has ordered GE Vernova turbines. The presenter distinguishes securing capacity through suppliers from bringing manufacturing under a company’s direct control.
• The broader energy discussion includes solar, batteries, grid capacity, and placing AI workloads where electricity is available. Batteries can help shift electricity to when it is needed, but generation, storage, grid connections, local approvals, and regulation all remain part of the problem.
• The presenter’s central conclusion is that owning a factory could give SpaceX more control, but only dependable, qualified output would make that control useful. Earlier power could enable earlier computing and revenue only if the rest of the site and customer demand are also ready.

Reviews, guides, or tutorials: No product review is presented. The video is an explanatory analysis and manufacturing walkthrough of gas turbines, blade design, casting, and the data-center power supply chain.

Main speakers and sources

• Farzad Mesbahi (presenter): Provides the technical explanation and analysis.
• Elon Musk: Quoted on SpaceX’s growth ambitions, turbine-part bottlenecks, and the company’s manufacturing plans.
• Bret Johnson, SpaceX CFO: Cited for the revenue forecast.
• Organizations and reporting cited: U.S. Department of Energy, Howmet, Siemens Energy, GE Vernova, the Wall Street Journal, and Foundry Management & Technology.