Why Is SpaceX Making Gas Turbine Blades?
SpaceX builds rockets.
So why is it building a factory to make parts for gas turbines?
The answer has surprisingly little to do with space.
It has a lot to do with artificial intelligence.
SpaceX is preparing a foundry in Bastrop, Texas, to manufacture blades and vanes used inside large gas turbines. These are highly specialized metal components that operate in some of the hottest and most demanding environments in industrial machinery.
Elon Musk says making them in-house could help bring new gas turbines online as much as 18 months sooner.
Why does that matter to SpaceX?
Because AI has a power problem.
And the deeper companies go into solving it, the more the AI industry begins to look like heavy industry.
First, AI Needed Chips
The first obvious bottleneck in the AI boom was computing power.
Companies wanted more GPUs.
Nvidia couldn't make enough of them.
Billions of dollars flowed into chips, servers and data centers.
But eventually another problem appeared.
You can own thousands of GPUs and still have a useless data center.
The machines need electricity.
A lot of it.
And electricity cannot always be added as quickly as servers.
Connecting a giant new data center to the power grid can require new substations, transmission equipment, transformers and additional generating capacity.
Suddenly, the race to build AI wasn't only a race for chips.
It became a race for power.
So Data Centers Started Looking at Gas Turbines
One way to get electricity faster is to generate it yourself.
That has made natural gas turbines increasingly attractive to data center developers.
Instead of waiting entirely for additional grid capacity, a large campus can potentially produce substantial amounts of electricity on-site.
SpaceX has already gone much further down this path.
In April 2026, the company agreed to purchase approximately $2 billion of mobile gas turbines and related equipment to provide power to its data centers.
Its prospectus also disclosed additional turbine commitments, taking planned spending on gas turbines for AI infrastructure to more than $2.8 billion over several years.
At its Colossus data centers near Memphis, SpaceX was operating 69 gas turbines as of July while planning a permanent 1.2-gigawatt natural gas power plant.
But buying turbines created another problem.
Everyone else wants them too.
Now There Aren't Enough Turbines
The AI boom didn't create the gas turbine industry.
Power plants have used these machines for decades.
But enormous new electricity demand from data centers is arriving at a time when turbine manufacturers are already dealing with rapidly growing order books.
Siemens Energy says it is effectively booked out through 2028.
The company has raised its outlook for global annual gas turbine additions to roughly 110 to 120 gigawatts, up from an earlier estimate of 90 to 100 gigawatts.
In its latest quarter, data center operators and Middle Eastern customers accounted for about half of Siemens Energy's gas turbine orders.
The company is even investing $1 billion to expand U.S. manufacturing of gas turbines, transformers and grid equipment.
So the bottleneck moved again.
First:
GPUs.
Then:
Electricity.
Then:
Gas turbines.
And inside the gas turbine, SpaceX found another bottleneck.
A Piece of Metal That Is Surprisingly Hard to Make
A turbine blade sounds simple.
It isn't.
Inside a modern gas turbine, blades spin in an environment of extreme temperature, pressure and mechanical stress.
They require sophisticated high-temperature materials, precision manufacturing and complicated cooling designs.
Producing them reliably is difficult.
And when demand for turbines suddenly rises, increasing production of these components isn't as simple as opening another assembly line.
This is where SpaceX's plan becomes interesting.
The company has been hiring for a new blades and vanes foundry in Bastrop.
Job postings describe work involving high-temperature superalloys, casting and manufacturing equipment for industrial gas turbine components.
Instead of waiting for the existing supply chain to produce more parts, SpaceX wants to make some of the bottleneck components itself.
Musk says that could shorten turbine deployment schedules by as much as 18 months.
Think about what just happened.
The race to build artificial intelligence has somehow become a race to cast metal.
AI Keeps Moving Down the Supply Chain
The progression is remarkable.
AI models
↓
GPUs
↓
Data centers
↓
Electricity
↓
Gas turbines
↓
Turbine blades
A software revolution has worked its way down into metallurgy.
And this reveals something important about the AI boom.
AI may feel digital.
Its physical infrastructure isn't.
Every new layer of computing creates demand for another layer underneath it.
A GPU needs a server.
A server needs cooling.
A data center needs electricity.
Electricity needs generation and transmission.
A gas power plant needs turbines.
And turbines need highly engineered pieces of metal that someone has to manufacture.
Eventually, the speed of the digital economy can become limited by the speed of a factory.
This Is a Familiar SpaceX Strategy
There is another reason SpaceX's move makes sense.
Vertical integration has long been part of how the company operates.
When a supplier is too expensive, too slow or unable to produce at the scale SpaceX wants, the company has often tried to bring more of the process inside.
That philosophy helped shape its approach to rockets.
Now something similar appears to be happening with AI infrastructure.
The difference is how far down the supply chain SpaceX is moving.
It isn't merely building an AI model.
It is securing computers.
Building data centers.
Buying power-generation equipment.
And now preparing to manufacture components used inside that equipment.
The boundary between a technology company and an industrial company is becoming increasingly difficult to see.
But Making the Blade Doesn't Solve Everything
There is an important catch.
Gas turbines burn fuel.
And deploying them quickly can create environmental and regulatory problems.
SpaceX's use of gas turbines around its Memphis AI facilities has already generated lawsuits, complaints and regulatory scrutiny over emissions and permitting.
The company has said it will remove unpermitted turbines as it transitions toward a permanent 1.2-gigawatt natural gas power plant, although the transition is expected to continue into 2027.
So manufacturing turbine blades faster doesn't magically solve AI's electricity problem.
It moves one bottleneck.
Others remain.
Fuel.
Pipelines.
Permits.
Emissions.
Transmission.
Transformers.
Water.
Land.
And eventually, another piece of equipment that nobody thought much about until everyone suddenly needed it.
The Bottleneck Keeps Moving
This may be one of the most useful ways to understand the AI infrastructure boom.
Don't just follow the technology.
Follow the bottleneck.
When GPUs were scarce, GPUs became extraordinarily valuable.
As computing capacity expanded, electricity became more important.
As demand for electricity increased, turbines and grid equipment became scarce.
And as turbine manufacturers tried to increase production, specialized components inside those machines became more important.
The economic opportunity often moves with the constraint.
Today's AI bottleneck might be a GPU.
Tomorrow it might be a transformer.
A power plant.
A cooling system.
Or a turbine blade.
That's why a company famous for launching rockets is now preparing to cast metal parts for power-generation equipment in Texas.
SpaceX isn't suddenly fascinated by gas turbines.
It needs electricity.
And when electricity becomes the constraint on AI, even a turbine blade can become part of the technology race.
BEYOND THE OBVIOUS.
Sources
The Information — SpaceX Turbine-Blade Factory
WIRED — SpaceX's $2.8 Billion Gas Turbine Commitments
Siemens Energy — U.S. Manufacturing Expansion
S&P Global — Gas Turbine Demand and Data Centers
Reuters — AI Data Centers Drive Siemens Energy Turbine Demand