Rachel Warren

So TerraFab is going to be a vertically integrated semiconductor factory designed to produce an unprecedented one terawatt or one trillion watts.

of AI computing power annually.

And unlike traditional factories that specialize in one part of the process, Musk envisions that TeraFab will handle everything, design, lithography, fabrication, packaging, under a single roof to bypass global supply chain bottlenecks.

for Tesla, this is really, and Musk, this is really about achieving total self-reliance for the physical AI ambitions.

I mean, the idea is the facility will mass produce the inference chips needed to power cyber cab robotaxis and optimist humanoid robots.

Musk estimates that all current global chip makers combined can only provide about 2% of the compute capacity his companies will eventually require.

One final thing I want to note, I mean, there's some

kind of major implications here for SpaceX as well, that 80% of TeraFab's output is earmarked for space-based applications, specifically for a new constellation of orbital AI satellites.

Basically, the satellites will use custom chips to run massive AI workloads in orbit and take advantage of the vacuum of space for better thermal management and more solar energy than on Earth.

So it's a very big promise that Musk is making, a massive vision that he has for TeraFab.

It'll be very interesting to see how this plays out.

Yeah, I mean, so gigawatts is a billion watts, right?

A terawatt is a trillion watts.

I mean, this is a massive leap that almost no one else is currently making.

To kind of put it in perspective, a single terawatt is about the total power capacity of the entire U.S.

grid, right?

And basically, this is a bet that the future requires a nation's worth of computing power to run things like Tesla's global robot fleets and orbital AI networks.

But I think it is important to kind of put some of this in context, right?

Really, the effectiveness of Terafab is going to really depend on whether Musk can solve the task of entering a field, semiconductor manufacturing, where his companies have zero experience.

I mean, there's been a lot of critics and industry experts that have come out saying, you know, this plan is virtually impossible.