Elon Musk has announced the launch of an ultra-large-scale chip manufacturing initiative codenamed TERAFAB, targeting an annual production capacity of 1 terawatt (TW) of computing power chips. This figure is approximately 50 times the current global annual chip output, with about 80% of the capacity intended to directly serve space missions. The project is jointly led by Tesla Motors, SpaceX, and xAI, and is located in Austin, Texas, marking the largest publicly announced manufacturing initiative in human history to date.
At a press conference held in downtown Austin on March 21, attended by Texas Governor Greg Abbott, Musk stated that the current global annual chip production capacity is approximately 20 gigawatts (GW), which represents only about 2% of his targeted requirement. He indicated that existing suppliers, including TSMC and Micron Technology, can no longer keep pace with the demand growth from Tesla Motors in the fields of robotics, autonomous driving, and artificial intelligence. "It's either build TERAFAB or have no chips," Musk said. He plans to first construct an advanced facility near the Giga Texas plant in Austin, integrating logic chips, memory chips, and advanced packaging within a single complex, and equipping it with lithography mask manufacturing equipment to create a closed-loop process encompassing design, production, and testing.
TERAFAB represents the first unified strategic project announced by Tesla Motors, SpaceX, and xAI as a joint entity. The announcement timing coincides with SpaceX's planned large-scale Initial Public Offering (IPO) this summer.
According to prior Bloomberg reports, SpaceX is expected to raise up to $500 billion, with a potential valuation exceeding $1.75 trillion. Deploying AI data centers into space is cited as one of the core rationales for this IPO financing. Musk did not provide a specific timeline for the factory's construction or the achievement of the capacity targets.
Analysts generally express reservations regarding the engineering feasibility of the plan, pointing out that semiconductor manufacturing involves investments amounting to hundreds of billions of dollars, multi-year construction cycles, and a highly scarce pool of specialized talent. However, some views suggest that TERAFAB's strategic significance extends beyond the chip industry itself—it integrates computing power manufacturing with space expansion under a single industrial logic, directly providing industrial-level support for SpaceX's valuation narrative.
The primary driver behind Musk's push for TERAFAB is his extreme assessment of future computing power demands. According to his roadmap, the Optimus humanoid robot project alone is projected to consume 100 to 200 GW of chip computing power, while the demand for space-based solar-powered AI satellite clusters reaches the terawatt scale. He anticipates that annual production of humanoid robots could eventually reach 1 to 10 billion units—compared to the current global annual automobile production of approximately 100 million units, suggesting robot production could be 10 to 100 times greater.
In Musk's view, terrestrial computing expansion is approaching physical limits. The total capacity of the U.S. power grid is about 0.5 TW, insufficient to simultaneously support large-scale AI training, robot operation, and data center demands. Suitable land for construction is becoming increasingly scarce, and opposition from local communities (NIMBYism) is growing. Space, however, presents a different scenario—no day-night cycle or atmospheric attenuation means solar efficiency can be over five times higher than on Earth, with marginal costs decreasing as scale increases. Musk estimates that within 2 to 3 years, the cost of deploying AI chips in space will be lower than deploying them on the ground.
Existing suppliers are unable to fill this gap. While Tesla Motors has a cooperation agreement with Samsung's Austin factory for next-generation chips and maintains supply relationships with TSMC and Micron Technology, Musk stated that the supply growth from these manufacturers is "far below our expectations." He noted that while the semiconductor industry is expanding output overall, the pace of expansion remains insufficient.
The TERAFAB plan involves completing the entire process—mask manufacturing, chip production, packaging, testing, and design iteration—within a single building, creating a high-speed recursive iteration loop. Musk claimed that, to his knowledge, this level of integration is unprecedented globally and is expected to accelerate iteration speed by an order of magnitude compared to existing solutions. Musk had previously mentioned a target process node of 2 nanometers.
The factory will produce two types of chips: one optimized for edge inference, primarily for Optimus robots and Tesla Motors vehicles; the other is a high-power chip specifically designed for the space environment, capable of withstanding high-energy particle bombardment, radiation accumulation, and extreme temperatures. These space chips are designed to operate at higher temperatures than terrestrial chips to reduce the weight requirements of cooling systems. In terms of demand structure, space chips are expected to dominate absolutely—Musk projects terrestrial computing power will stabilize at the 100-200 GW level, while space-based computing will ultimately reach the terawatt scale.
At the press conference, Musk also displayed a prototype of a 100-kilowatt AI micro-satellite, suggesting that "future satellites could reach the megawatt level." In January of this year, SpaceX applied to the U.S. Federal Communications Commission (FCC) for permission to launch one million data center satellites into orbit.
The announcement of TERAFAB closely aligns with SpaceX's IPO preparations. According to Bloomberg, SpaceX plans to complete its IPO this summer, with an expected fundraising amount of up to $500 billion. If successful, this would set a record for IPO fundraising, potentially valuing the company at over $1.75 trillion. The deployment of AI data centers in space is a core part of the financing rationale, and the TERAFAB announcement provides concrete industrial support for this narrative.
SpaceX completed its acquisition of xAI in February of this year, with xAI now operating as its wholly-owned subsidiary. Musk stated that the vast majority of chips produced by TERAFAB are expected to be consumed by xAI, primarily for training space-based AI models and processing satellite data.
Synergies between Tesla Motors and xAI are already evident on multiple fronts: Tesla Motors sells its Megapack energy storage products to xAI, some Tesla vehicle models have integrated xAI's Grok chatbot, and in January, Tesla Motors announced a $2 billion investment in xAI along with a framework cooperation agreement. The TERAFAB announcement represents a further escalation of the synergy between the three companies—moving from product-level collaboration to jointly leading a unified industrial project.
The strategic significance of TERAFAB lies in its integration of the disparate capabilities of Musk's three companies into a complete industrial chain. Under this framework, Tesla Motors handles the chip demand side for Optimus robots and electric vehicles, SpaceX is responsible for transporting chips and computing infrastructure to orbit using its large-scale launch capacity, and xAI operates the space-based AI satellite system and consumes the majority of the chip output. Together, they form a closed loop from chip manufacturing to orbital deployment and AI computation. This is the fundamental logic behind Musk defining the project as a "joint project of three companies" rather than a single corporate initiative.
Musk positioned TERAFAB as "the first step for humanity to become a multi-planetary civilization" and outlined a subsequent roadmap: building an electromagnetic mass driver on the Moon, utilizing the Moon's low gravity and lack of atmosphere to accelerate materials directly to escape velocity. At that stage, computing scale could expand a thousandfold, entering the petawatt (PW) range. He expressed his hope to see the lunar mass driver constructed within his lifetime.
During an earnings call in January, Musk had already indicated that building TERAFAB was intended to "resolve a highly probable production bottleneck within three to four years." If the project materializes, its impact will extend beyond the semiconductor industry—affecting the global computing power supply landscape, orbital data center infrastructure, and even the engineering foundation for SpaceX's deep-space missions.
Analyst reservations focus on three main areas. Financially, Morgan Stanley estimates the cost of building a factory producing 100,000 advanced logic chip wafers per month at approximately $45 billion; UBS estimates start from $30 billion. Baird analyst Ben Kallo directly raised the market's key concern: "Where will the money come from?" Musk has not yet disclosed any financing arrangements.
On the supply chain front, high-end extreme ultraviolet (EUV) lithography machines are almost entirely dependent on Dutch company ASML, with delivery lead times of 1 to 2 years, often longer for new customers. Integrating logic chips, memory chips, and advanced packaging processes within a single factory would exponentially increase system complexity. New semiconductor facilities typically cost tens of billions of dollars and often require several years to reach full production capacity.
Regarding talent, Bernstein semiconductor analyst Stacy Rasgon commented, "Because it's Musk, I won't dismiss it outright, but I suspect this is actually harder than putting a rocket on Mars." He cited the example of TSMC's Arizona factory—a project that has faced years of delays, necessitating the airlifting of engineers from Taiwan to the U.S. to support production ramp-up. "That talent isn't just lying around," Rasgon said.
It is noteworthy that Musk himself has no background in semiconductor manufacturing and is perceived by outsiders as having a history of over-promising on targets and timelines. During the recent press conference, he did not provide any specific deadlines for factory construction progress or production ramp-up.