Musk limits TSMC to possible Terafab sublease; Intel says it will stay involved

Musk capping TSMC at a space-sublease tier is a structural rejection of the open foundry model. The strategic intent is clear: process IP, yield curves, and equipment tuning stay in-house, creating a walled-garden fab where the entire upstream stack—ASML EUV integration, Applied Materials deposition, back-end packaging—must recalibrate around Intel's node roadmap rather than the N3/N2 trajectory out of Taiwan, China. On compliance, sidelining a Taiwan, China-based player from the core manufacturing loop of a US strategic fab materially reduces cross-strait supply-disruption tail risk. The trade-off is an explicit bet on Intel's foundry maturity, and the 18A yield ramp remains an open wound in the industry's collective memory. Competitively, Samsung's P4 and GlobalFoundries' 2nm efforts gain a rare strategic window. TSMC's anchor customers—Apple, AMD, NVIDIA—will almost certainly accelerate multi-sourcing to hedge against this new concentration risk. Over the next 12–24 months, expect the 'fab-as-sovereign-asset' paradigm to displace cost-optimized offshore foundry logic. Government-subsidized in-country manufacturing becomes the default, and pure-play foundry share will see its first structural decline since the 1990s.