Tesla Motors CEO Elon Musk announced on X this past Saturday that the company's Terafab project, aimed at manufacturing artificial intelligence chips, will commence operations in seven days. While Musk did not provide specific details, he and his team are expected to soon explain how the chip fabrication facility will be implemented. This move signifies another expansion for Tesla beyond its core electric vehicle business and marks the beginning of a large-scale, capital-intensive endeavor.
Currently, the majority of the world's advanced chip manufacturing is handled by a few foundries, including TSMC, Samsung, and Intel. Tesla both purchases chips from companies like Nvidia and designs its own chips internally. The company has established manufacturing agreements with foundries such as Samsung and TSMC. Tesla is currently designing its fifth-generation AI chip, known as AI5, to advance its autonomous driving vision. According to Tesla, this chip, targeted for mass production in 2027, is expected to deliver 50 times the performance of the current AI4 chip. The AI5 will feature 9 times the memory capacity and 10 times the raw computing power of AI4, and it is intended for use across a wide range of Tesla applications, from electric vehicles and robots to AI training and data centers. Meanwhile, the AI6 chip is in the early stages of development. Musk has stated that subsequent versions, including AI7, AI8, and AI9, are planned, with a target design cycle of nine months for each chip. It is understood that AI6 will be primarily used for the Optimus robot and data center computations, while AI7 will shift focus to space-grade AI computing.
Driven by the artificial intelligence boom, the semiconductor industry is experiencing unprecedented growth, with chip demand so immense that companies like Tesla are facing supply constraints. Over the past few months, Musk has repeatedly indicated that Tesla would build a massive chip factory to produce AI chips and alleviate supply bottlenecks. At Tesla's annual shareholder meeting last November, Musk first introduced the concept of "Terafab"—a facility similar to a Gigafactory but dedicated to AI chip production, with the goal of achieving full vertical integration. At that time, Musk expressed his view that Tesla would need to establish a massive chip factory to realize its ambitions for manufacturing self-driving cars and humanoid robots, stating he saw no other way to achieve the required chip output. He also mentioned the possibility of collaboration with Intel, suggesting that talks with the company might be worthwhile, though no agreements had been signed.
Earlier this year, Musk announced plans to build a 2nm chip factory and challenged the industry's cleanroom standards, famously remarking in an interview that he wanted to create a facility where one could "smoke cigars and eat hamburgers." During Tesla's earnings call in January, Musk again discussed the plan for an in-house chip factory, noting that existing chip suppliers could not meet the company's demand.
According to Musk's previous statements, he aims for an annual production output of 100 to 200 billion chips. Technically, this would make his factory one of the largest chip manufacturing plants in the world. While Musk appears determined to enter the chip manufacturing sector, some experts have expressed skepticism, calling his ideas far-fetched. Critics point out that the barriers to entry in semiconductor manufacturing are extremely high, and building a chip factory from scratch is one of the world's most formidable technical challenges. Some also believe that Musk's concept of operating without a cleanroom is unworkable. There is speculation that one way Tesla might realize TeraFab is through licensing agreements with chip manufacturers like Intel and TSMC, providing the necessary funding to help them establish production lines.
Comments