Saturday, October 3rd 2026

TSMC Explores Collaboration with Elon Musk's Terafab Project

Taiwanese semiconductor giant TSMC is considering collaborating with Elon Musk's Terafab project to establish advanced end-to-end semiconductor manufacturing in the United States. According to an exclusive report from Tim Culpan's Culpium newsletter, later confirmed by Elon Musk on X, the two companies are in discussions about a joint venture to boost American semiconductor manufacturing. The plan involves TSMC potentially developing its own semiconductor manufacturing capabilities, with Terafab providing financial backing, equity participation, purchase agreements, and other necessary steps to keep the new fab capacity operational. As some managers believe that Intel might not come through with its promises, Terafab would need to dual-source expertise.

Previously, we learned that Intel has partnered with Terafab, with Tesla becoming Intel's first major customer for the 14A node at Elon Musk's Terafab AI chip complex in Austin. However, the specifics of this deal remain unclear, including whether Terafab will license the 14A node PDK. While both TSMC and Intel can coexist within the Terafab project, it is important to note that TSMC might not bring its latest semiconductor technology outside of Taiwan, keeping its most advanced nodes at its headquarters, while Terafab remains a secondary project. Perhaps Intel and TSMC will compensate each other where the technology needs balance, but that is yet undetermined.
The goal of Terafab is to consolidate the entire chip manufacturing process under one roof. The plant aims to integrate several stages of semiconductor production at a single site, including logic fabrication, memory, packaging, testing, and mask production. This setup is unusual, as these steps are typically spread across multiple specialized facilities and companies.

Additionally, simply investing more funds does not guarantee a solution. Building a competitive fab from scratch requires years of execution, including sourcing parts from companies like ASML, Lam Research, KLA Corporation, and Tokyo Electron. These companies often receive orders years in advance before delivering machinery to TSMC, Samsung, or Intel. Thus, creating a supply chain for equipment, raw materials, and other necessary components will take years to develop. Terafab has reportedly already begun ordering tools for its manufacturing process from these firms, so we can expect the first silicon in a few years.
Sources: Culpium, Elon Musk on X
Add your own comment

14 Comments on TSMC Explores Collaboration with Elon Musk's Terafab Project

#1
AsRock
TPU addict
TerraFab sounds like a terrible name.
Posted on Reply
#2
Steevo
Whatever you think about Musk is irrelevant, he sees an opportunity and takes it and has made humanity better overall, while this might not be a better for humanity project it will be better for the US and show that manufacturing can still be done here despite the claims otherwise.
Posted on Reply
#3
TechBuyingHavoc
AsRockTerraFab sounds like a terrible name.
TerraBad, the new and more accurate name.
Posted on Reply
#4
Six_Times
Exciting. Long term or short, advancements will come of this.
Posted on Reply
#5
Onasi
AsRockTerraFab sounds like a terrible name.
Agreed. The only Fabricators in Sector Solarum are to be on Mars, as per the Treaty of Olympus. This Elon of Musk is clearly a Heretek and an investigation should be launched posthaste.

Praise the Omnissiah.
Posted on Reply
#6
Athena
Not a bad thing, but we will see how fast they can get things up and running
Posted on Reply
#7
kondamin
SteevoWhatever you think about Musk is irrelevant, he sees an opportunity and takes it and has made humanity better overall, while this might not be a better for humanity project it will be better for the US and show that manufacturing can still be done here despite the claims otherwise.
its a pity this isn’t going to happen anymore.
To much money has been wasted on data centres
Posted on Reply
#8
user556
The article is unclear what Terafab will bring that TSMC doesn't already do itself. Maybe packaging, since that is currently done outside of USA. Interposing assembly could be another area since that is now lacking everywhere. *Shrug*
Posted on Reply
#9
kondamin
user556The article is unclear what Terafab will bring that TSMC doesn't already do itself. Maybe packaging, since that is currently done outside of USA. Interposing assembly could be another area since that is now lacking everywhere. *Shrug*
The idea behind Terafab is that it will do everything on a massive scale
From creating silicon boules to leading edge logic and dram.

Loads of it is needed to make every car a smart car and by scaling everything up things should get cheaper.
Like what happened with the gigafab and batteries
Posted on Reply
#10
user556
Ouch, modern cars are already properly bork'd. Key-FOBs priced 100x a cut key typifies the breakdown.
Posted on Reply
#11
kondamin
user556Ouch, modern cars are already properly bork'd. Key-FOBs priced 100x a cut key typifies the breakdown.
It's pretty obvious we will be going in a direction that is going to make cars with human drivers only illegal
Those vehicles will need chips, big powerful chips.
If that is what will be happening a similar demand for high end """"AI""""" chips will be needed as we have seen in the 2025 AI build out on a yearly basis.
Posted on Reply
#12
user556
Maybe, they've still got a very long way to go for cars. We'll see stuff like planes and trucks first. Freighting being the leader. Where the uses are more controlled.

As for the chips, those are not big at all when compared to what is being piled into data centres now. The self-driving chips were in cars already, before the data centre boom, and that didn't even register on the demand curve.
Posted on Reply
#13
kondamin
user556Maybe, they've still got a very long way to go for cars. We'll see stuff like planes and trucks first. Freighting being the leader. Where the uses are more controlled.

As for the chips, those are not big at all when compared to what is being piled into data centres now. The self-driving chips were in cars already, before the data centre boom, and that didn't even register on the demand curve.
The BYD uses 480mm² chips 4nm by tsmc which is larger than the 5080, the difficulty in full self driving is all the special conditions that can happen during a commute. So they aren't that small.
a car needs more than just one.
Posted on Reply
#14
user556
It's small compared to the total amount of hardware in a car. It's a tiny amount compared to what is being thrown at the data centres.

BTW: If you're throwing kilowatts of power at the AI then that's substantial drain on the battery. Not desirable at all.
Posted on Reply
Add your own comment
Nov 12th, 2026 04:39 CST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts