Musk: SpaceX's In-House Casting of Engine Blades Could Advance Timeline by 18 Months
Elon Musk stated that SpaceX and Tesla are each rapidly building photovoltaic manufacturing capacity of 100 gigawatts per year, but will still need natural gas as a supplement and to initiate photovoltaics in the coming years; the limiting factor for increasing natural gas turbine production is the casting of blades and vanes. If SpaceX completes the casting in-house, it could bring the grid connection of the units forward by up to approximately 18 months.
This photovoltaic goal was first mentioned in Davos to BlackRock CEO Larry Fink: the two teams are advancing domestic manufacturing in the U.S. separately, which will take about three years. He viewed energy availability as a constraint on the expansion of artificial intelligence, comparing it to China's annual manufacturing capacity of about 1500 gigawatts and over 1000 gigawatts of installed capacity. According to the U.S. Solar Industry Association, the current component capacity in the U.S. is about 74.1 gigawatts.
Tesla has applied for tax incentives for a photovoltaic factory named Project Crystal Sun in Texas, located in Fort Bend County, with an investment declaration of approximately $10.1 billion, aiming for commercial operation in Q1 2029 and creating over 9,700 full-time jobs. The application referenced the aforementioned 100 gigawatts/year statement but did not specify the gigawatt capacity of the factory. SpaceX is also linking photovoltaics with its solar AI satellite program.
Natural gas is described as a transitional power source. SpaceX has committed to over $2.8 billion in turbine procurement related to xAI, including approximately $805 million in agreements and a potential $2 billion for mobile units; the Colossus site has operated over 40 mobile gas turbines, adding more than 500 megawatts of generation. The company acquired mobile power provider APR Energy for about $1 billion, with antitrust review expected to pass in May 2026; the mobile gas and diesel fleet has a capacity of over 1 gigawatt and has procured five 380-megawatt units from Doosan Enerbility for fixed power stations. The planned super-large manufacturing and computing facility Terafab will also use on-site gas turbines and batteries as a voltage stabilization solution.
The casting of blades is concentrated in a few high-temperature alloy foundries, and Musk views the supply cycle as unacceptable. In-house casting means migrating aerospace hot-end processes into power equipment. Photovoltaics are responsible for long-term energy supply, while natural gas provides dispatchable capacity and the energy quality required for chip production lines; Musk referred to the 18 months as a decisive variable.
In market mechanisms, buyers are the electricity-deficient training clusters and future photovoltaic factories, while sellers are the traditional turbine casting supply chains. The event is driven by the gap in computing power and electricity. Funding flows from the capital expenditures of the two companies into component production lines, mobile power stations, and casting workshops. Beneficiaries are self-sufficient power sources that can shorten waiting times, while the pressure is on external foundries where blade delivery cycles still operate at the pace of the aerospace and power duopoly.
Source: Public Information
ABAB AI Insight
Musk's fixed action to solve bottlenecks is: if you can't buy it, cast it yourself. Batteries, transformers, and rocket engines have all followed the same path, and now it's the turn of gas turbine hot ends. SpaceX already has experience in high-temperature alloys and precision casting, changing blades from purchased components to in-house components is not diversifying into energy equipment, but extending the lifespan of the meters for Colossus and Terafab. The 100 gigawatts/year photovoltaic goal is a three-year slogan, while the gas timeline is currently 18 months; both are mentioned together because solar energy cannot meet the voltage quality and nighttime base load required by chip factories.
The capital path is divided into three segments: acquiring APR to obtain mobile gigawatts; ordering fixed 380-megawatt units from Doosan; and bringing the slowest castings into the aerospace workshop. The motivation is that external casting queues will cause already purchased compressor casings to idle on the ground. The strategy is to use transitional fuels to maintain the training curve while connecting the narrative of domestic photovoltaics to the component gap under tariffs in the U.S. Tesla's application in Fort Bend turns the slogan into a tax form, while SpaceX turns the slogan into the DC bus for satellites and data centers.
Similar vertical integration is seen in TSMC building its own electricity and water, Intel building its own lithography support, and wartime aircraft engine factories bringing blades from suppliers into their own factories. The power equipment industry is at the stage of controlling casting capacity, while the photovoltaic manufacturing industry is in the expansion phase where U.S. capacity is still below the slogan. The Musk system is simultaneously on two curves: one that takes three years to grow component factories, and one that aims to light up turbines 18 months early.
The essence is the reconstruction of the industrial chain. The constraints of artificial intelligence shift from model weights to high-temperature alloy molds. The mechanism is: the delivery time of dispatchable electricity is determined by the blades, and the blade delivery time is determined by a few foundries; whoever can pull the casting cycle out of the industry queue can change the power-on date for data centers from "waiting in line" to "scheduled for production." Photovoltaics are responsible for reducing long-term levelized cost of electricity, while gas is responsible for buying training days for the next few years.
ABAB News · Cognitive Law
The bottleneck is not in the fuel, but in those few factories that can cast blades.
The timeline of transitional energy determines whether the long-term energy slogan can be fulfilled.
Key castings that cannot be purchased will ultimately become the buyer's workshop.