What Is Terafab? Musk's Newly-Launched Project that Aims to Produce 1 Terawatt Yearly Compute

Elon Musk on Sunday unveiled Terafab, a $25 billion chip manufacturing project that will be built in Austin and jointly operated by Tesla Inc and SpaceX.

Announced by SpaceX on X as a “next step towards becoming a galactic civilization,” the facility is aimed at giving Musk’s companies greater control over the chips required for AI, robotics and future space-based computing.

What will Terafab Focus On?

Terafab is expected to function as a semiconductor fabrication plant that will design, test and manufacture custom AI chips exclusively for Musk’s businesses.

The project is likely to be located on Tesla’s campus in eastern Travis County, Austin, Texas. Musk said the facility would initially operate as an “advanced technology fab,” capable of producing and testing chips of various kinds, while laying the groundwork for a much larger computing infrastructure.

The primary objective behind Terafab is to reduce dependence on external chip manufacturers and accelerate development across Tesla, SpaceX, and xAI.

Musk argued that the semiconductor industry is not evolving quickly enough to meet the anticipated surge in demand driven by advancements in AI and robotics. He emphasised that building Terafab is essential to ensuring a reliable supply of chips for his companies’ future needs.

Musk’s Vision for Terafab

Musk also outlined an ambitious long-term target for the project, stating that it aims to produce “a trillion watts of compute per year.” He added that a significant portion of this capacity may eventually need to be deployed in space, given that the United States currently generates only about 0.5 terawatts of electricity.

As such, Terafab is being positioned not merely as a manufacturing initiative, but as a foundational element of a broader infrastructure strategy.

Unlike the traditional fabless model, where companies design chips but outsource production, Terafab is intended to internalise the entire semiconductor stack.

This would include advanced packaging, memory integration, and inference-optimised silicon tailored for specific applications. The first generation of chips is expected to focus on AI inference and could power Tesla’s self-driving vehicles, robotaxi fleet, and the Optimus humanoid robot.