Elon Musk’s Terafab may well be the most ambitious undertaking Tesla (NASDAQ:TSLA | TSLA Price Prediction) or any of his other companies has ever attempted. Highly capable Optimus robots, reusable rockets that deliver payloads into orbit, and longer-term visions of human settlement on Mars are all genuinely audacious. But building a domestic semiconductor foundry capable of weaning Musk’s empire off outside suppliers takes audacity to another level entirely.
Consider what is at stake. The project would reduce dependence on Taiwan Semiconductor (NYSE:TSM), which today holds a near-monopoly on leading-edge chip manufacturing. Even with Intel (NASDAQ:INTC) signed on as a manufacturing partner, the Terafab is a profoundly expensive endeavor. There is no margin for error, and no guarantee that building a world-class fab from scratch can be compressed into anything approaching a normal construction timeline.
That said, Musk’s track record of doing what established players said could not be done is hard to dismiss. The potential upside is substantial, particularly as Musk and his team look to inject their own engineering culture into a manufacturing process that has historically moved at the pace of institutional inertia.
The Terafab is a go, and things are moving rapidly
Musk officially unveiled the Terafab concept on March 21, 2026, at the defunct Seaholm Power Plant in Austin. The vision is sweeping: a vertically integrated facility that consolidates chip design, lithography, memory production, advanced packaging, and testing under one roof. The stated goal is to produce more than one terawatt of AI compute capacity annually, a scale that dwarfs anything currently operating anywhere on Earth.
Public filings with Grimes County, Texas, where SpaceX is seeking a property tax abatement, put the first phase of construction at $55 billion. A full multi-phase buildout could reach $119 billion. Those numbers are already daunting, but Bernstein analysts have noted that achieving chip independence at genuine terawatt scale could ultimately demand somewhere between $5 trillion and $13 trillion in total capital expenditure. The $119 billion figure, in that context, is the opening bid on a very long game. Ben Bajarin, a chip analyst at Creative Strategies, described Musk’s approach as a “15-year strategy.”
Intel’s role in all of this has become clearer since the Q1 2026 earnings call. Tesla confirmed it will use Intel’s forthcoming 14A manufacturing process at the facility, making Tesla the first major outside customer for a node Intel has been pitching for months. Intel shares had their best month ever in April, more than doubling in value after the partnership was disclosed. CEO Lip-Bu Tan had previously warned that Intel would exit contract manufacturing altogether if it could not secure an external buyer for 14A, which makes Tesla’s commitment something of a lifeline for that business unit.
A clear division of labor
The structure of the project is worth understanding. Tesla is building a roughly $3 billion semiconductor research and development fab at its GigaTexas campus in Austin. That facility, capable of processing a few thousand wafers per month, is intended for prototyping and validating new manufacturing ideas. Tesla’s fifth-generation AI chip, AI5, is among the first products the pilot facility is designed to produce, with small-batch output anticipated in 2026 and volume production targeted for 2027.
SpaceX, meanwhile, is handling the initial phase of the full-scale Terafab in Grimes County. Musk has been explicit that any intercompany arrangements between Tesla and SpaceX require board approval and a formal conflict-resolution process, a governance layer that adds complexity but also provides some protection for Tesla shareholders. Tesla raised its 2026 capital expenditure guidance to over $25 billion at the Q1 earnings call, up from the “over $20 billion” figure issued just three months earlier. CFO Vaibhav Taneja confirmed the company expects negative free cash flow for the remaining quarters of the year as a result.
Can Terafab actually expedite the build?
The honest answer is that nobody outside Musk’s inner circle knows. Building a competitive fab from scratch requires years of equipment procurement from companies like ASML, Lam Research, KLA Corporation, and Tokyo Electron, all of which receive orders years in advance. A typical sub-3nm fab costs over $20 billion before accounting for packaging, memory, and testing infrastructure. Terafab’s goal of consolidating all of those steps multiplies the cost and the execution risk in ways that are genuinely unprecedented.
What Musk does have in his favor is urgency and capital. He has argued publicly that current global fab capacity produces only about 2% of what Tesla and SpaceX will eventually require across all their projects. In his framing, the choice is binary: build the Terafab, or fall short on chips. That framing may be theatrical, but the underlying constraint, that leading-edge capacity at TSMC is booked years out by Apple, NVIDIA, and other hyperscalers, is real.
Intel’s 14A node adds a credible technological foundation. The process builds on RibbonFET gate-all-around transistor architecture and extends backside power delivery through PowerVia-derived technology. Musk noted on the earnings call that by the time Terafab scales up, 14A “will be probably fairly mature or ready for prime time,” which is a candid acknowledgment that the node is still in development but a reasonable bet on its trajectory.
The SpaceX IPO adds a new dimension
The article’s original framing positioned the SpaceX IPO as a future event. It is no longer future. SpaceX priced its shares at $135 each on June 11, 2026, raising $75 billion in what became the largest IPO in financial history, surpassing Saudi Aramco’s 2019 record by a wide margin. Trading began on NASDAQ under the ticker SPCX on June 12. At the IPO price, SpaceX carried a valuation of approximately $1.75 trillion, ranking it among the ten largest listed companies globally.
The IPO has direct implications for Terafab. SpaceX’s prospectus stated that proceeds will fund continued growth of its AI compute infrastructure, launch facilities, and satellite constellations. With fresh capital on the balance sheet and public market scrutiny now in play, the pressure on SpaceX to execute the Terafab timeline has intensified. Tesla itself owns 18.99 million SpaceX shares, valued at roughly $2.56 billion at the IPO price, giving Tesla a financial stake in SpaceX’s success that runs alongside the manufacturing partnership.
Whether investors are fully pricing the Terafab opportunity into Tesla’s current valuation remains an open question. The project is long-dated, capital-intensive, and technically demanding. But for a company that has already built its own AI training clusters, its own vehicle chips, and its own battery cells, vertical integration into semiconductor fabrication is a logical next step, if an unusually expensive one.
Editor’s note: This version corrects the SpaceX valuation from a speculative “climbing toward $2 trillion” to the confirmed $1.75 trillion IPO valuation at $135 per share, notes that SpaceX’s record-breaking $75 billion IPO priced on June 11, 2026 and began trading on NASDAQ under SPCX on June 12, clarifies the Terafab cost structure as a $55 billion first phase with a total potential buildout of $119 billion per Grimes County public filings, adds the Tesla $3 billion research fab versus SpaceX’s full-scale Grimes County facility distinction, updates Tesla’s 2026 capital expenditure guidance to over $25 billion per the Q1 2026 earnings call, and adds the Bernstein analyst estimate of $5 trillion to $13 trillion in long-term chip-independence capex.
Top Gaining Stocks
Top Losing Stocks