Click here: to donate by Credit Card
Or here: to donate by PayPal
Or by mail to: Free Republic, LLC - PO Box 9771 - Fresno, CA 93794
Thank you very much and God bless you.
Posted on 08/31/2026 11:46:46 AM PDT by SunkenCiv
SpaceX's plan to build the world's largest facility -- 100 million square feet in Grimes County, Texas, near Houston -- to do advanced computing and also manufacture, test and package silicon chips for its own use and by other Elon Musk-owned companies, will be self-powered using a gigawatt-scale natural gas power plant and battery storage, according to the company.
The operations "are not expected to use electricity from the grid," the company said in a tax abatement agreement it entered into with the county in June.
(Excerpt) Read more at facilitiesdive.com ...
|
|
Click here: to donate by Credit Card Or here: to donate by PayPal Or by mail to: Free Republic, LLC - PO Box 9771 - Fresno, CA 93794 Thank you very much and God bless you. |
Electricity supply in the U.S. is a limiting factor preventing AI data centers from coming online as quickly as possible. Because of this, many projects are turning to portable natural gas turbine generators to get the electricity they need and bypass long wait times to connect to the grid. This has driven demand for turbines, so Elon Musk said on X that SpaceX will bring turbine blade and vane casting in-house...
This sudden influx of demand has led to jet engine shortages for data center operators. Beyond the fact that engine manufacturers weren't prepared for the AI boom, commercial aviation has already been suffering from its own shortages due to supply chain and engine design issues even before data center developers started buying up turbines.SpaceX starts in-house turbine blade manufacturing to boost gas-powered generator output for Elon's AI data centers -- new manufacturing strategy cuts generator delays by 18 months | Tom's Hardware | Jowi Morales | August 30, 2026
🚀🛰️🔌💡🔦
No pocket nuke plant?
Growing single crystal hot section turbine blades for high efficiency gas turbines is far harder than casting. But I’m sure Musk has hired the best manufacturering people away from Siemens and GE and greatly boosted their incomes.
So they will not be allowed to hire any workers named Ingrid.
No pocket nuke plant?
That would also be my question.
I understand that Musk does a lot of high tech stuff and does it very well. But I hope he understands that turbine blade manufacturing is very high tech as well.. you better do it right or you will tear your turbine to bits.
“commercial aviation has already been suffering from its own shortages due to supply chain and engine design issues”
The supply chain issues pertain in this context to these very same turbine blades.
The Nancy Grace Roman telescope that just got launched relies in part on surplus parts that didn’t make it to recon sats.
Building their own power generation capacity is what ALL AI sites should do. Way to set the example, Elon.
Roger that. Thanks.
Power has become the constraint that decides that race. Training and inference at the scale now contemplated require gigawatts, delivered on a schedule measured in months rather than the many years a conventional grid interconnection often takes. Tesla and SpaceX are racing to build solar manufacturing at enormous annual capacity. Even at that pace, solar and batteries will not close the gap for several years. Natural gas generation is the bridge: turbines that can be placed beside a data center and brought online far faster than a new transmission line.
Those turbines are themselves scarce. The major manufacturers’ order books stretch years forward. Everything on a large industrial machine can often be sourced on a shorter cycle except one class of parts. The hottest section of the turbine uses blades and vanes that sit in gas far hotter than the melting point of the metal they are made from. They survive only because of internal cooling passages, ceramic thermal-barrier coatings, and a microstructure grown as a single crystal or as tightly controlled directional grains. Metal grain boundaries are weak points for creep and cracking. Removing them is what allows the metal to live in that environment.
The manufacturing process is vacuum investment casting. A wax pattern is built around a ceramic core that will become the cooling channels. A ceramic shell is formed around the wax. The wax is melted out. Superalloy is melted under vacuum and poured. The mold is then withdrawn through a controlled temperature gradient so that solidification proceeds as one crystal, or as aligned grains, rather than as a random jumble. Yields on large industrial blades are harsh. Inspection rejects parts for the wrong grain structure even at shops that have done this work for decades. Only a handful of Western foundries cast these airfoils at industrial scale. Their furnaces are full. A set of hot-section parts can take longer to produce than almost every other component on the turbine.
That is the bottleneck SpaceX is trying to break. The company is standing up a blades-and-vanes foundry in Bastrop, Texas, adjacent to existing Starlink manufacturing, on land assembled for the purpose. This is not an entirely foreign craft. SpaceX already casts nickel-superalloy, directionally solidified and single-crystal components for Raptor engine turbopumps. The same family of alloys, vacuum furnaces, and skilled operators applies to both rocket turbomachinery and land-based hot-section airfoils. Hiring has targeted materials engineers with years of single-crystal and directional-solidification experience on blades and vanes, operations leads to own construction and ramp-up of a new foundry, automation engineers for wax injectors, ceramic shells, and solidification furnaces, and programmers for industrial-gas-turbine airfoil machining. The equipment is equally specialized: vacuum induction melting and directional-solidification furnaces, typically from a small set of builders with long delivery times, plus shell systems, hot-isostatic pressing, heat treatment, and coating lines. The stated aim is to bring natural-gas turbines online as much as eighteen months sooner by making the scarce parts in-house rather than waiting in the existing queue. Complete machines continue to come from suppliers; SpaceX has purchased large-frame units and acquired a mobile fleet. The foundry is aimed at the parts that actually set the pace.
None of that work is easy. Large power-plant blades are bigger than rocket parts. Maintaining a clean solidification front is harder. Cooling geometries and specific alloys are closely held intellectual property. Qualification for use in a commercial turbine takes time. A vacuum foundry is typically many months from groundbreaking to first commissioning, and longer to stable output. Permitting and process development still have to be done. The claim is not that a rocket company becomes a full turbine original-equipment manufacturer overnight. The claim is that relevant process knowledge, captive demand from both data centers and engines, capital, and a habit of attacking manufacturing bottlenecks already exist in one place.
But there is a big cost trade-off available to Musk. The highest-efficiency combined-cycle plants exist to minimize fuel burned over decades of baseload running. They do it by pushing turbine inlet temperature as high as materials will allow. Those temperatures are precisely what demand the most difficult single-crystal blades, the most intricate cooling, and the most demanding coatings. A first-mover race does not require winning that metallurgical contest on day one. If the firing temperature is relaxed by several hundred degrees, simpler directional castings become adequate. Cooling and coating requirements ease. Yields rise. A new line can produce usable parts sooner, including parts that do not yet carry decades of pedigree on the latest high-temperature class. Fuel consumption per megawatt-hour rises, but that increased fuel cost is small compared to speed to produce MW. Time to first useful electrons falls significantly if you do not chase the very best thermal efficiency.
That trade is rational because fuel is cheap relative to lost position. Extra natural gas is an operating cost that can be paid continuously. Months of delayed training are not. The early pattern already looks like this: mobile and readily available machines first, purchased large frames next, an internal foundry to unstick the hot-section queue, and higher-temperature single-crystal work continuing in parallel for rockets and for later, more efficient plants. The immediate product is not the last two points of combined-cycle efficiency. The immediate product is power on a date that still decides who leads the AI race.
(Full disclosure: my ideas, my power industry knowledge, my outline, my drafting and Grok’s able writing assistance)
The man has built work forces that do more than NASA, Boeing, Lockheed, Martin-Marietta, Johnson Controls, etc., on a daily basis. Do you think they can’t now produce jet engines after already revolutionizing commercial and scientific rocketry?
I’m all in on Elon’s track record for doing incredible stuff.
Exactly. Set the bar high because none of his competitors can catch up.
Ugh…nice, but definitely a Googly TL;DR, so I skipped it.
No nuke, but a synchrotron driving a free electron laser, which is cool enough.
With all this debate about data centers and their projected power needs, I keep thinking back to early 80’s and power projections that WPPSS (Washington Public Power System and Supply) used to determine Washington atate needed 5 1,300 Megawatt nuke plants. Of course they famously defaulted on $25B in bond debt when the will of the political class abandoned nuclear power. Part of it was the public was used to paying next to nothing for cheap hydropower in those days and I think the power cost was going to be in the neighborhood of 8 to 10 cents per kilowatt to buy power from these nukes.
All of those plants would have been online by 1990 and we’d be sitting pretty.
Very informative. Ya done good.
Yeah, I could have tightened it up by half. But I got tired of working on it. Sorry it was TL.
I imagine the regulatory time frame for implementation doesn't work well for Musk. As well as avoiding NIMBY protests.
Military facilities that can be exempted by the CINC from regulatory hurdles are where we're going to see those tested.
“Electricity supply in the U.S. is a limiting factor preventing AI data centers from coming online as quickly as possible. “
These shortcomings courtesy of the leftist luddites and GND advocates who consider planning the evening meal as detailed planning for the future. Beyond that Gaia will provide ...
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.