Elon Musk has excluded Taiwan Semiconductor Manufacturing Company from the planned Terafab chip initiative, redirecting a potential supply decision toward Intel Foundry Services.

The Terafab project, Musk's consolidated semiconductor manufacturing program spanning Tesla, xAI and SpaceX workloads, will not use TSMC's leading-edge nodes, according to StreetInsider. Intel, which has been courting external foundry customers since its 2024 restructuring, stands to absorb part of the redirected demand.

What does the TSMC exclusion change in chip supply?

TSMC operates the largest share of sub-5nm merchant wafer capacity globally, with Nvidia, Apple and AMD among its biggest external customers. The Taiwanese foundry's CoWoS advanced packaging line remains the dominant bottleneck for current AI accelerator shipments.

Removing TSMC from the supplier shortlist for a project of Terafab's stated scale shifts a major customer decision away from the world's leading foundry. Musk's companies have previously used TSMC for custom silicon programs, including earlier generations of Tesla inference and training processors. Routing new wafer starts away from Hsinchu pushes that workload onto alternative process nodes.

The decision also signals that Musk is willing to look beyond the default foundry for hyperscaler-class volume, a stance few buyers have publicly adopted to date.

Why is Intel positioned to benefit?

Intel Foundry Services has been rebuilding its external customer roster since its 2024 separation of product and foundry operations, a structural change intended to give external customers clearer visibility into capacity and roadmap.

A Terafab allocation would rank as one of the larger external commitments secured since the relaunch and would give Intel a marquee anchor customer outside its traditional PC and data center silicon base. The deal would in practice:

• Lock in multi-year wafer volume for Intel's 18A manufacturing process
• Provide a reference customer for Intel's advanced packaging capacity
• Validate Intel's foundry-as-service pitch to other hyperscalers

Specific process nodes, wafer-start numbers and contract value remain undisclosed. Intel has not issued a statement on the matter.

What is Terafab's scope?

Terafab is Musk's vertically integrated semiconductor initiative, intended to consolidate chip production for Tesla's autonomous-driving hardware, xAI's training accelerators and SpaceX's orbital compute systems under a single manufacturing program.

Musk has framed the project as a response to capacity bottlenecks in third-party fabs serving AI workloads and as a hedge against foundry concentration risk. The project has been described in terawatt-class compute terms, though concrete wafer-start targets, capital expenditure figures and geographic siting have not been released.

Who else could participate?

Samsung Foundry remains an unconfirmed candidate for the Terafab lineup. The Korean foundry operates a 2nm gate-all-around process in early production and has been seeking anchor customers to fill capacity at its Taylor, Texas site, which began equipment installation in 2024.

Any US-domiciled fab allocation would align with the $52.7 billion CHIPS and Science Act framework, though no Terafab-specific award has been announced. The program's potential federal subsidy eligibility will depend on whether any portion of Terafab capacity is set aside for non-Musk-affiliated customers.

What stays unresolved?

• Final foundry partner selection and process node allocation
• Construction timeline, capex commitments and site selection
• Split between Tesla autonomy silicon and xAI accelerator production
• Whether the announcement advances into executed offtake agreements