The question institutional investors are sitting with this Monday morning is not whether TSMC and Terafab will reach a deal. They probably will not, at least not soon. No formal agreement has been signed, and both sides are still at the discussion stage. The more important question is what the conversation itself signals about where the AI capital-spending cycle is heading, and who gets hurt if it goes the distance.
Why the Debate Opened
TSMC and Elon Musk’s Terafab semiconductor production initiative are discussing a potential collaboration under which TSMC would build and operate a facility that will exclusively serve Elon Musk’s Tesla, SpaceX, and xAI companies, according to Tim Culpan’s Culpium newsletter. Musk kept it short on X: “Just discussions, but something may come of it.” TSMC has not commented. TSM closed at $472.78 on October 2, 2026, marking a 2.96% gain.
The timing matters because TSMC was already under scrutiny for a separate expansion. TSMC is weighing the construction of a new Texas campus that could pump tens of billions of dollars into its growing U.S. chipmaking footprint, Reuters reported, separate from the $265 billion it has committed to an Arizona campus. An interesting background detail from Culpium is that TSMC’s rumored intentions to build a fab complex in Texas are tied to the foundry’s effort to support Terafab.
The Bull Case: Volume, Captivity, and U.S. Soil
Bears on TSMC’s overseas strategy always point to margin dilution. Overseas fab ramps are expected to cause roughly 2 to 3 percentage points of gross margin dilution per year in the early stages, a management-acknowledged headwind. The bull counter is that a Terafab arrangement could flip that math. In the scenario observers consider most likely, TSMC would own and operate a dedicated fab complex in Texas with Terafab as an anchor customer, while Musk’s companies provide capital, long-term chip volume commitments, or both, similar in structure to how TSMC has partnered with local stakeholders for expansion projects in Japan and Germany while keeping process technology and operations under TSMC’s control.
Terafab, jointly backed by Tesla, SpaceX, and xAI, involves a first-phase investment of $16.8 billion, with total potential investment reaching up to $119 billion. That is not a small anchor. TSMC’s own playbook endorses the concept: management has long emphasized stable, predictable pricing and avoiding opportunistic, short-term actions during shortages as part of how it sustains customer relationships.
The Bear Case: Pricing Discipline at Risk
Here is what most of the weekend coverage missed. TSMC’s moat is not simply technical; it is behavioral. Pricing restraint is part of the moat, leaving enough value with customers to keep their best designs, commitments, and future road maps concentrated at TSMC. A factory built and run for a single customer, one who is simultaneously a competitor to some of TSMC’s existing clients in AI compute, breaks that covenant in a way that Japan’s JASM and Germany’s ESMC did not. Those were multi-customer structures. Terafab, by design, is aimed at securing capacity for advanced semiconductors needed for Tesla and its robotics ambitions, as well as for AI and data center projects tied to xAI and SpaceX.
If Apple and Nvidia watch TSMC redirect leading-edge capacity and operational expertise toward a captive Musk fab, the implicit promise that TSMC competes with nobody gets harder to defend. The potential Texas investment is also contingent on U.S. tax policy. The CHIPS advanced manufacturing investment credit under Section 48D does not apply to property the construction of which begins after December 31, 2026, unless extended. That is another variable TSMC cannot control.
What Investors Are Missing: Intel’s Position Just Got Shakier
Until this week, Intel was the only other named manufacturing partner in the public discussion around Terafab. Musk has said Terafab plans to use Intel’s 14A process technology when it has its own production capacity later this decade and when the process matures. The framing has been treated as Intel’s first marquee outside-customer alignment around 14A.
That win still stands on paper. But TSMC’s arrival at the table changes the room. Intel’s filings have acknowledged the stakes plainly: if Intel is unable to secure sufficient committed demand for Intel 14A through product design wins with potential significant external customers, it faces the prospect that it may pause or discontinue its pursuit of Intel 14A and other next-generation leading-edge process technologies. Terafab was supposed to be the proof point that made 14A credible to the next customer. TSMC’s presence in these discussions introduces doubt about whether that proof point holds.
Stocks to Watch
TSM gets the near-term bid, but the real valuation question is longer-range: does a dedicated-customer fab model compress or expand its multiple? TSLA and the broader Musk ecosystem benefit from having the world’s best foundry potentially underwriting their chip supply security. INTC is the most exposed: Intel says it is actively engaged with multiple prospective external customers, and expects customers to begin making 14A decisions in the second half of 2026 and into the first half of 2027. That window just got more crowded. AMAT and LRCX sit behind any new fab regardless of who operates it, equipment spending is the one part of this story that does not depend on who wins the ownership argument.
