A look at the symbiotic relationship between TSMC and Apple: TSMC gets a predictable, demanding, high volume customer and Apple gets cutting-edge chips at scale
The TSMC-Apple partnership has been one of the defining relationships of the mobile computing era.
Context & Ripple Effects
Apple’s chip sourcing has long combined deep dependence with selective flexibility: it relied on TSMC as its sole supplier by 2021, after an earlier dual-sourcing period for the A9 chip with Samsung. Subsequent reporting described a defect-cost arrangement tied to early 3nm access, showing how the two companies shared manufacturing risk as well as volume.
The relationship has grown far larger over time, with Apple’s annual TSMC spend reported to have risen from $2 billion in 2014 to $24 billion in 2025. But the AI boom has made TSMC capacity less exclusively available: Apple is now reported to be competing for production slots as Nvidia’s demand rises.
First-order effects
- TSMC receives predictable high-volume demand from Apple, helping support its investment in leading-edge manufacturing capacity.
- Apple gains scaled access to TSMC’s most advanced chips, while its reliance on one foundry concentrates supply exposure with that supplier.
Second-order effects
- Apple’s early-access and defect-cost terms raise the competitive bar for rival chip designers seeking leading-edge TSMC output.
- As AI customers absorb more TSMC capacity, Apple’s historical purchasing scale no longer assures preferential allocation, making chip supply a more active product-planning constraint.
Third-order effects
- The partnership illustrates how advanced-chip manufacturing concentrates leverage in a small number of specialist suppliers: large buyers can shape process investment, but remain constrained when shared capacity tightens.
- If AI-driven demand persists, foundry relationships will be defined less by stable bilateral preference and more by allocation competition among Apple, Nvidia, and other major customers.
The trend: Advanced-chip procurement is shifting from long-term preferred access toward contested foundry capacity as AI demand competes with mobile computing for the same leading-edge nodes.