Intel joins Elon Musk's TeraFab project alongside SpaceX, xAI, and Tesla to build domestic chips
Apr 7, 2026
Key Points
- Intel joins Tesla, SpaceX, and xAI's TeraFab project to design and manufacture ultra-high-performance chips targeting one terawatt per year of compute for AI and robotics.
- The Austin-based facility solves a supply constraint for Musk's companies, whose chip demand will exceed capacity from TSMC, Samsung, and Nvidia combined.
- Intel gains guaranteed near-term revenue from high-compute customers while demonstrating domestic fabs can compete with TSMC, potentially unlocking broader demand to justify its expansion.
Summary
Intel Joins Musk's TeraFab Project to Build Domestic AI Chips
Intel has joined Elon Musk's TeraFab initiative alongside SpaceX, xAI, and Tesla to design, fabricate, and package ultra-high-performance computing chips at scale. The project aims to produce one terawatt per year of compute for AI and robotics applications.
CEO Lip Bu Tan announced the partnership Tuesday, positioning Intel's manufacturing and packaging capabilities as critical to TeraFab's ambitions. The move marks a strategic reversal for Intel, which has spent years ceding ground to NVIDIA and AMD while struggling to capitalize on surging data center chip demand. For Tesla and SpaceX, the partnership solves a near-term constraint: Musk has publicly stated TeraFab is needed because his companies' chip demand will far outstrip what traditional suppliers like TSMC, Samsung, and Nvidia can provide.
The facility, planned for Austin, Texas, will produce chips for Tesla's robotaxis and humanoid robots, SpaceX's satellite constellation, and xAI's compute infrastructure. Tesla currently sources chips from multiple vendors—Samsung for some robotaxi work and TSMC for its Nvidia Dojo chips—but the unified fab would consolidate supply under domestic control.
The geopolitical and supply chain angle
Intel's partnership with Musk's companies reflects a broader acknowledgment that TSMC, the global chip leader, is not expanding capacity as aggressively as the industry demands. Multiple sources in the segment note expectations of a chip bottleneck in the coming years. The U.S. government has already moved to shore up domestic capacity: in 2024, the Trump administration acquired an 8.4% equity stake in Intel for roughly $9 billion as part of a larger CHIPS Act investment that allocated $280 billion—including $53 billion specifically for domestic semiconductor manufacturing.
Intel shares rose nearly 3% on the announcement Tuesday and are up roughly 15% over the past month and 167% over the past year, reflecting investor appetite for a resurgent domestic chipmaker.
The demand-side bet
The structural problem Intel has faced is not engineering capability but customer commitment. Multiple AI infrastructure companies have remained locked into TSMC even as they design proprietary chips, largely because TSMC's lead in yield and scale made the decision automatic. TeraFab appears designed to flip that equation by guaranteeing demand from within Musk's own ecosystem—Tesla, SpaceX, and xAI represent some of the highest-compute-intensity operations in existence. If successful, the project could demonstrate to other AI companies that domestic fabs are viable alternatives, potentially unlocking broader demand Intel needs to justify its expansion plans.
The partnership also sidesteps the political uncertainty that has surrounded Intel acquisitions in the past. In 2022, when Intel's market cap sat at $110 billion and the CHIPS Act was newly funded, observers briefly imagined a Musk acquisition. That deal never materialized, but this narrower partnership achieves some of the same strategic goal: binding one of Musk's highest-leverage companies to Intel's success and giving the chipmaker a concrete, near-term revenue stream from a customer with genuine, scale-driven demand.