Musk's Terafab Chip Plan: Betting on ASML Lithography Technology

Musk has announced a chip manufacturing initiative called Terafab, centered on adopting advanced lithography equipment from Dutch supplier ASML. The move is being interpreted as an attempt by Musk to bring more chip production in-house in response to the growing computing demands of his companies.

However, the information made public so far is quite limited. There is no cooperation agreement, no procurement contract, and no disclosure of investment amounts, timelines, locations, or capacity targets. This makes Terafab look more like a declaration of strategic intent than an engineering project that can be immediately evaluated.

Why bring chip manufacturing in-house

The rationale behind Terafab is not complicated: Musk's businesses are becoming increasingly dependent on processors. Tesla needs automotive chips to support autonomous driving and vehicle software, xAI needs large volumes of AI training and inference chips, and SpaceX's communications and aerospace systems likewise rely on specialized processors. Concentrating these needs onto a proprietary manufacturing platform could, in theory, strengthen control over the supply chain, design priorities, and delivery schedules.

This complements Musk's previously discussed plans to expand computing infrastructure. Building more data centers addresses the question of "where computing power runs," while building its own chip production lines addresses "where processors come from." Both point to the same goal: reducing passive dependence on external supply systems amid accelerating AI demand.

According to a report by The Daily Flare, Terafab has not yet specified how capacity will be allocated among Tesla, SpaceX, xAI, or other entities. This detail is precisely the key to judging its nature—if capacity mainly serves internal needs, it is more like a vertically integrated cost center; only if it serves external customers would it take on the character of a commercial wafer fab.

ASML's role: critical but undecided

ASML is the focus of this plan because its extreme ultraviolet (EUV) lithography systems are irreplaceable in advanced process nodes. EUV lithography uses extreme ultraviolet light to print extremely fine circuit patterns onto silicon wafers, an indispensable step in manufacturing high-end processors. ASML is currently the world's only supplier capable of providing commercial EUV lithography systems, and any plan to produce leading-edge chips cannot avoid this Dutch company.

But "centered on ASML technology" and "having reached a cooperation with ASML" are two different things. The original report explicitly states that Terafab's public information does not confirm a procurement agreement or formal partnership. An advanced wafer fab also requires specialty materials, precision engineering, software systems, and numerous equipment suppliers—lithography machines are only one piece of the puzzle.

The more realistic challenge lies in yield. Even with the most advanced EUV equipment, going from process validation and materials supply to stable mass production still requires substantial time and engineering accumulation. Companies such as TSMC and Samsung took years to push advanced-process yields to commercially viable levels, and this is not a process that can be compressed simply by capital investment.

How high is the barrier to EUV lithography?

Extreme ultraviolet lithography (EUV) uses extreme ultraviolet light with a wavelength of about 13.5 nanometers, projecting circuit patterns onto silicon wafers through a multilayer mirror system, enabling process nodes below 7 nanometers. Because extreme ultraviolet light is absorbed by almost all materials, the entire optical path must operate in a vacuum, and the light source relies on high-power lasers striking tin droplets to generate plasma—an extremely complex technology. A single EUV lithography machine contains more than 100,000 components, weighs hundreds of tons, and typically costs more than $150 million to $200 million. Beyond the equipment itself, supporting photoresists, mask pellicles, and process integration capabilities also constitute barriers. This is why only a handful of companies worldwide can compete in advanced process nodes.

Three tests at the execution level

For Terafab to truly materialize, it faces tests at at least three levels.

The first is process and yield. Chip manufacturing is not just about buying equipment; it also requires establishing reproducible process flows and continuously optimizing production to improve yield. Customers also need to confirm that new processors are compatible with their software and systems, which requires long-term validation and collaboration.

The second is deep dependence on the supply chain. Building its own wafer fab does not eliminate dependence on external suppliers, especially for lithography tools and other specialized equipment. ASML's EUV systems are themselves the product of global collaboration, and the more Terafab pursues advanced process nodes, the more deeply it becomes embedded in this global network.

The third is alignment with product strategy. Tesla has already developed processors for vehicle systems, while AI workloads place entirely different demands on chip design. A shared manufacturing base can provide flexibility, but one chip design cannot serve all scenarios. How to coordinate production priorities for automotive-grade chips and AI chips on the same line is a question that needs to be answered in advance.

What yield means

Yield refers to the proportion of qualified chips on a wafer and is the core metric for measuring the economics of a wafer fab. Initial yields at advanced process nodes are often only 50% or even lower, and must be gradually ramped up by adjusting process parameters, controlling defect density, and optimizing equipment matching. Every percentage point increase in yield directly corresponds to lower costs and released capacity. For a new production line, it usually takes two to three years from equipment installation and process debugging to stable mass production, during which customer validation and reliability testing are also required. Therefore, yield is not just a technical issue but a commercial one that determines whether the investment can be recouped.

What to watch next

For Terafab, the reasonable attitude at present is to focus on verifiable progress rather than over-interpreting the vision itself. Signals worth tracking include disclosed investment and construction plans, confirmed equipment orders, manufacturing targets, and evidence that actual production lines are running. Until these details emerge, Terafab's scale and schedule remain open questions.

ASML's role provides a broader footnote: the highly specialized nature of advanced chip manufacturing means that any unilateral in-house effort must ultimately return to the network of global collaboration.