NVIDIA CEO Jensen Huang has thrown cold water on Elon Musk’s ambitious plan to build a massive semiconductor fabrication plant, warning that creating advanced chip manufacturing capabilities is “extremely hard” and goes far beyond simply constructing facilities.
Table of Contents
The TeraFab Vision
| Aspect | Details |
|---|---|
| Musk’s Plan | Build a “TeraFab” – beyond TSMC’s Gigafab scale |
| Target Output | 1 million chips per month |
| Purpose | Power Tesla’s AI5 chips for vehicles and Optimus robots |
| Alternative | Considering outsourcing to Intel Foundry |
| Context | Announced after $1 trillion pay package approval |
| Challenge | TSMC’s Arizona Fab 21: $165 billion for Gigafab capacity |
Why Huang Says It’s Nearly Impossible
Speaking at a TSMC event in Taiwan, Huang emphasized that advanced chip manufacturing involves not just building plants, but mastering the engineering, science, and artistry that TSMC has perfected. His timing was strategic—NVIDIA depends entirely on TSMC as its sole semiconductor supplier for powering the AI revolution.
The reality check comes as Tesla needs tens of thousands of NVIDIA GPUs while planning to use AI5 processors across its vehicle and robotics ecosystem, especially after canceling its Dojo supercomputer project.
The ASML Bottleneck
Even with unlimited capital, manufacturing equipment giant ASML can’t magically double its extreme ultraviolet (EUV) lithography tool production—the backbone of advanced chip manufacturing. Add the staffing challenge of recruiting thousands of specialized engineers, and the timeline extends to decades, not years.
Yet dismissing Musk entirely would be premature. He’s known for achieving things initially deemed impossible, and companies like Samsung and Intel would eagerly compete for Tesla’s business. The semiconductor industry desperately needs competition as TSMC dominates with rising prices.
Whether Tesla’s TeraFab materializes or remains aspirational, one thing is clear: the tension between chip demand and manufacturing capacity will define the next decade of tech innovation.
