Elon Musk has once again stirred up the tech sector with a minimal message, but enough to spark speculation. On March 14th, he posted on X that the “Terafab Project” will launch “in 7 days,” placing the announcement around March 21, 2026. Reuters later confirmed that the project is linked to the manufacturing of AI chips for Tesla, although there are no official details yet on location, partners, final investment, or actual production schedule.
The key point is precisely there: currently, what has been confirmed is not a new operational foundry, but a presentation or formal launch of a project. Musk has been hinting since 2025 that Tesla might need to build a “gigantic chip fab” to meet future AI chip demands, considering that even in the best-case scenario, their current suppliers would be insufficient. In that same context, he mentioned possible talks with Intel and noted that Tesla continues collaborating with TSMC and Samsung for its AI chips.
From the idea of a “giga-factory” of chips to industrial reality
Musk’s vision aligns with a logic that’s easy to understand. Tesla wants more chips for autonomous driving, robotics, data centers, and future AI systems. If the company aims to rely less on NVIDIA and major Asian foundries, controlling a larger part of its supply chain seems like a natural move. Reuters explained in November 2025 that Tesla is designing its fifth-generation AI chips and that Musk saw a need for a manufacturing scale-up to support that roadmap.
However, designing chips and operating an advanced foundry are very different things. It’s important to temper expectations regarding TeraFAB. According to Reuters, at Tesla’s annual meeting, Musk spoke of a plant capable of starting with at least 100,000 wafers per month. That sounds huge—and it is for a single facility—but it doesn’t mean “surpassing TSMC” or dominating the global semiconductor market. TSMC states on its official website that its combined annual capacity surpassed 17 million 12-inch wafer equivalents in 2025. Even a very ambitious Tesla plant would still be far from the total volume handled by the world’s leading foundry.
Additionally, the chip industry isn’t built overnight. Cutting-edge fabs depend not just on capital, but on processes, equipment, personnel, materials, validation, wafer yield, and years of experience. ASML emphasizes that their EUV systems are crucial for volume production of advanced logic chips and that the next-generation High-NA tools are destined for nodes below 2 nm. In other words, competing at the forefront requires access to the most sophisticated machinery in the world and an extremely complex manufacturing chain.
The role of Intel, TSMC, and Samsung
One of the most interesting nuances of the project is that Musk has never clearly indicated a complete break with existing manufacturers. Reuters explained both in November and March that Tesla remains partnered with TSMC and Samsung, and Musk has left the possibility open for collaboration with Intel. This suggests that TeraFAB might end up being less a fully autonomous, from-scratch foundry and more a hybrid platform supported by industry partners for some critical phases.
This nuance also explains why many headlines claiming “immediate independence” from TSMC seem premature. Tesla may want more control, vertical integration, and volume, but there’s no public evidence suggesting it will replace its current manufacturing partners in the short term. What is clear is that Musk considers chip supply a strategic issue for Tesla’s future.
The cleanroom controversy and why caution is needed
Another part of the debate revolves around Musk’s provocative comments on cleanrooms. In January, his remarks circulated widely, defending that the industry is “doing a poor job” with modern cleanrooms and joking that one day he might eat a burger and smoke a cigar inside a 2 nm factory. But this does not constitute a confirmed technical plan for TeraFAB. In fact, there’s no official documentation so far describing an advanced factory without a cleanroom environment.
It’s important to add some serious industrial context. ASML explains that their manufacturing facilities are cleanrooms precisely because cleanliness is a critical condition, measured against international particle standards. In advanced semiconductors, contamination is not a minor detail; it directly impacts defects, yield, and process viability. Therefore, literal interpretations of Musk’s jokes should be viewed more as cultural provocation than as reliable descriptions of an operational foundry.
An eye-catching project, but still very early stage
What seems clear is that Musk wants to send a message to the market: Tesla doesn’t just want to design chips; it also aims to influence how they are manufactured and at what volume. The company has been talking for months about a more aggressive cadence of new chips, with AI5 nearly completed, AI6 in early development, and Dojo 3 making its way back into the public conversation. DataCenterDynamics summarized in January that Musk envisions a nine-month design cycle for future processors, aligning with the idea of a more ambitious manufacturing infrastructure.
But industrial ambition doesn’t yet mean industrial execution. Today, TeraFAB is primarily a statement of intent with a launch date. It could become a transformative project for Tesla, a more complex manufacturing alliance with Intel, Samsung, or TSMC, or even a less revolutionary initiative than some headlines suggest. What’s unwise is to already claim that Musk will inaugurate the world’s largest chip factory or that he’s about to outdo TSMC. For now, the official announcement is the only reality; everything else remains promise, vision, and a lot of work ahead.
Frequently Asked Questions
What has Elon Musk actually confirmed about TeraFAB?
He has only publicly confirmed that the “Terafab Project” will launch in seven days from his March 14 message, which points to March 21, 2026. Reuters describes it as a project to manufacture Tesla AI chips, but no further official details have been released yet.
Will TeraFAB be a chip factory bigger than TSMC?
There’s no basis for that claim today. Musk spoke in 2025 about a plant with at least 100,000 wafers per month—a very high figure for a single facility—but TSMC asserts that its combined annual capacity surpassed 17 million 12-inch wafer equivalents in 2025.
Will Tesla stop relying on TSMC and Samsung?
Not imminently. Reuters has indicated that Tesla continues working with TSMC and Samsung, and Musk has mentioned possible talks with Intel. For now, TeraFAB does not imply a confirmed break with those partners.
Is it true that the factory will not have a cleanroom?
This has not been confirmed as an official plan. Musk’s provocations about cleanrooms are just that—provocations. There’s no official technical documentation of TeraFAB supporting an advanced foundry without controlled environments. ASML emphasizes that cleanliness is a core requirement in semiconductor manufacturing.
