According to Zhitong Finance APP, Tesla (TSLA.US) CEO Elon Musk said last Sunday that two advanced chip factories will be built at a large industrial site in Austin, Texas. One will provide computing power support for automobiles and humanoid robots, while the other will supply chips for artificial intelligence (AI) data centers deployed in space. This statement followed Musk’s announcement the previous day of plans to construct an advanced AI chip complex named ‘Terafab’ in Austin. Musk posted on social media: ‘Technically, Terafab will consist of two wafer fabs, each producing only one type of chip design.’

Musk previously announced the construction of one of the largest chip manufacturing plants in history, named ‘TeraFab.’ This super factory aims to achieve an annual computing power production capacity exceeding 1 terawatt (1000 gigawatts), covering logic, storage, and advanced packaging. It is jointly initiated by SpaceX, Tesla, and xAI.

It is reported that TeraFab is planned to have an annual production capacity of 100 billion to 200 billion AI and memory chips, equivalent to approximately 100,000 wafers per month, with an estimated total investment of $20 billion. The first mass-produced chips, AI5, are scheduled for production in 2027 and will be used for full self-driving (FSD), the humanoid robot Optimus, Cybercab autonomous taxis, and data centers. Musk further revealed on social media that after the factory is completed, about 80% of its capacity will be used for space applications, while the remaining 20% will be for ground-based applications.

This is ‘one of the largest semiconductor manufacturing operations ever planned by a private company,’ which will ‘make Tesla one of the largest semiconductor manufacturers in the world,’ allowing Tesla to no longer rely on Taiwan Semiconductor, Samsung, or any external suppliers, controlling every layer of the AI stack from chips to software. It is also one of Tesla’s largest capital investments, funded by Tesla’s $44 billion cash reserves and supported by Musk’s long-term AI vision.

Musk stated: ‘We are very grateful to our existing supply chain, including companies like Samsung and Taiwan Semiconductor, and hope they can expand quickly. We will buy all their chips. I have already communicated this to them. There is a limit to how fast they can scale, but it is much slower than our expectations, so we need to build Terafab.’

Notably, at Tesla’s annual shareholders’ meeting in November 2025, Musk first mentioned the idea of building a chip factory. He stated that to support the continuous iteration of Tesla’s Full Self-Driving (FSD) software and the large-scale deployment of the Optimus robot, Tesla’s annual demand for chips will reach 100 billion to 200 billion units. Even if global foundries expand according to the most optimistic forecasts, their capacity will not meet Tesla’s explosive demand for AI chips in the future. Therefore, building a large-scale wafer fab is ‘inevitable.’ Musk pointed out during a demonstration at a facility in Austin last Saturday: ‘Either we build Terafab, or we will run out of chips.’

However, the construction of the Terafab project requires a substantial financial threshold. Industry insiders stated that building a wafer fab using advanced 2-nanometer process technology typically requires an investment of $25 billion to $40 billion, with a construction period of 3 to 5 years. This poses significant pressure on Tesla’s current financial situation.

Data shows that Tesla’s revenue for the full year of 2025 decreased by 3% year-over-year to $94.8 billion, while net profit plummeted by 46% to $3.79 billion. Despite holding over $44 billion in cash and investments, Tesla’s capital expenditure budget for 2026 has exceeded $20 billion, which does not fully cover the enormous costs of the Terafab project. Market analysis suggests that Tesla may need to raise funds through equity financing to support this project.

Gartner analysts stated that the joint chip development efforts of Tesla, SpaceX, and xAI will promote cross-scenario efficient allocation of computing resources, providing a new model for the integration of AI and aerospace industries. Despite numerous challenges, the successful implementation of this project will significantly enhance the resilience and diversity of the global chip industry.