AI Chips Are Challenging Apple’s Longtime Dominance at TSMC

A new report argues that Apple helped build the modern leading-edge foundry model at TSMC, but AI computing is now changing who pays for new chip technology and who sets the pace.

Apple and the TSMC Model

The report explains how Apple became TSMC’s most important customer by committing early to new manufacturing nodes, starting with the A8 chip in 2014. Apple paid upfront costs, helped improve yields, and aligned its yearly product cycle with TSMC’s roadmap. That approach allowed TSMC to move faster than rivals.

Apple’s spending at TSMC grew from about $2 billion in 2014 to roughly $24 billion in 2025. At its peak, Apple generated up to 25 percent of TSMC’s revenue. For many years, Apple consumed most of the early output of each new node, effectively funding advanced manufacturing when no one else could.

The rise of AI accelerators introduced new customers that can absorb massive chip capacity.

Smartphones once made up nearly half of TSMC’s revenue. High-performance computing, driven by AI, is now the largest segment. Companies like NVIDIA now fund large parts of new capacity, reducing Apple’s share of early production.

This shift shows up in future nodes. Apple is expected to take a smaller share of early production for TSMC’s N2 and A16 processes. A16 focuses on high-performance computing rather than mobile chips.

Apple’s Role Going Forward

The report says Apple’s influence grows again at later nodes such as A14, which target both mobile and high-performance products. At that stage, iPhone and Mac chips once again drive volume.

Apple still relies on TSMC for its most advanced silicon. At the same time, the report says Apple is exploring alternatives for lower-risk components to diversify its supply chain. Intel’s upcoming 18A-P process is mentioned as one possible option for select Apple silicon without affecting flagship products.

See SemiAnalysis’s full report for deeper analysis.