Tesla Selects Intel 14A for Terafab Chip Production

Elon Musk announced that Tesla plans to use Intel's 14A fabrication technology for chips made at the Terafab complex, with Tesla responsible for a pilot research fab at Giga Texas and SpaceX lined up for high-volume manufacturing. Musk described a near-term research fab costing about $3 billion, capable of a few thousand wafers per month, and framed Terafab as a path to secure the chip supply needed for Tesla, Optimus humanoid robots, and xAI data-center workloads. He also stated an eventual Terafab target of one terawatt of computing capacity per year in public remarks.

The plan centers on Intel's 14A process, a node positioned after 18A in Intel's roadmap that is not yet fully qualified in production. Musk said, "By the time Terafab scales up, 14A will be probably fairly mature or ready for prime time." Practical implications for practitioners include:

• •Potential licensing, tech-transfer, or foundry partnerships to deploy 14A technology at facilities operated by Musk-led entities
• •A staged build: a $3 billion research/pilot fab at Giga Texas for design validation and iterative experimentation, followed by higher-volume manufacturing under SpaceX stewardship
• •Near-term production volumes described as a few thousand wafers per month, insufficient for full product supply but adequate for R&D and pilot silicon

This move is material on three fronts. First, it is an operational signal that large AI consumers are moving from buying silicon to vertically integrating production capacity, driven by forecasts that demand will outstrip existing foundry output. Second, it represents a strategic commercial victory for Intel, which has struggled to regain leadership against TSMC and Samsung; landing Tesla as a marquee partner strengthens Intel's foundry narrative. Third, it exposes risk: 14A is still under development, so Tesla is betting on a future node maturing on schedule. That gamble links Tesla's AI timeline to a process technology that must hit yield and performance targets to justify Terafab economics.

Securing a long-term route to advanced nodes could reshape bargaining power with external suppliers, including TSMC and Samsung, and may accelerate similar vertical moves by other hyperscalers and OEMs. For Intel, involvement could mean licensing or close technical collaboration rather than a pure customer-supplier deal. For practitioners, the practical consequence is a potential new source of advanced node capacity that could open alternative supply for AI accelerators, but with uncertainty on timelines and unit economics.

Monitor concrete agreements between Intel and Musk entities (licensing terms, IP transfer, and capacity commitments), yield and qualification milestones for 14A, and funding or permitting signals that move Terafab from concept to large-scale construction. The critical open questions are timing, yield trajectories for 14A, and whether Terafab economics scale beyond pilot experimentation to high-volume, cost-competitive production.