The 2nm Overflow: Why AMD and Tesla are Deserting TSMC for Samsung’s Texas Fab

As of 25 December 2025, the piece notes that TSMC remains dominant in advanced AI chip manufacturing, with capacity increasingly constrained as Apple and NVIDIA reportedly absorb the initial 2-nanometre capacity while AMD, Google and Tesla have begun discussions with Samsung about its 2nm offerings.

Samsung’s Taylor, Texas fab is seen as a potential first-mover, with volume production of a 2-nanometre GAA process expected as early as 2026, offering a U.S. manufacturing option in a market where overflow orders spill toward Samsung and Intel. The article highlights that roughly half of TSMC’s 2nm capacity is rumoured to be reserved for Apple, with NVIDIA the other top-priority customer, and notes that Intel has conducted chip tests for NVIDIA on the 18A process under CEO Lip-Bu Tan’s direction.

It also cites Intel’s pragmatic path, including demonstrations of combining 14A and 18A with Foveros packaging for future AI workloads, while predicting 2026 could reshape the global foundry landscape. According to Reuters, TSMC’s third‑quarter 2025 market share sits around 71%, underscoring the pressure on rivals to offer viable alternatives.