Products & Announcements

Nvidia Edges Apple for TSMC’s Leading Edge as AI Wins the Wafers

Semiconductor IndustryAI InfrastructureTechnology EconomicsGPU Computing
Added Jan 16
Link479
Article: NeutralCommunity: NeutralDivisive
Nvidia Edges Apple for TSMC’s Leading Edge as AI Wins the Wafers

TSMC is prioritizing AI chips as Nvidia’s demand surges, likely pushing Apple from the top customer spot in some periods and forcing it to fight for capacity and accept higher prices. Current and near-term nodes (N2P, A16) are optimized for HPC workloads, while a future A14 node could swing balance back toward Apple. TSMC’s record margins and massive capex come with high risk, so it’s expanding carefully despite outside pressure to scale even faster.

Key Points

  • Nvidia has likely overtaken Apple as TSMC’s top customer in some quarters and could do so for full-year rankings as AI demand soaks up leading-edge capacity.
  • TSMC’s mix is shifting toward HPC/AI (up 48% last year) while smartphone-related revenue grows modestly, pushing capacity toward GPUs with larger die sizes.
  • Near-term nodes (N2P, A16 with backside power) are best suited for HPC, favoring Nvidia/AMD; A14 (~2028) targets both mobile and HPC, likely improving Apple’s position later.
  • TSMC’s margins and capex are surging (62.3% gross margin; $52–56B 2026 capex), but it must expand cautiously due to high capital intensity and depreciation risk.
  • Critiques that TSMC isn’t scaling fast enough miss that fabless clients bear far less risk; TSMC carries the fixed-cost burden if demand cools when the AI boom flattens.

Sentiment

The discussion is broadly analytical and accepts the article's premise that Nvidia currently outspends Apple for leading-edge capacity. However, there is notable skepticism about whether this shift is permanent. Many commenters believe Apple's long-term strategic position remains strong and that the AI spending boom may eventually cool, restoring Apple's pricing power with TSMC. The overall tone is balanced with slightly more agreement than disagreement with the article's thesis.

In Agreement

  • Nvidia's explosive revenue growth and willingness to pay premium prices naturally earns it priority over Apple at TSMC's leading edge
  • TSMC operates as an auction-like system — capacity goes to whoever pays the most, and Apple must match or lose priority
  • AI demand creates a genuine, massive need for leading-edge silicon that dwarfs smartphone processor requirements
  • TSMC's upcoming node architectures like A16 with backside power are optimized for HPC and AI routing, naturally favoring GPU designs over mobile SoCs
  • Apple's smartphone revenue growth has stagnated compared to the AI sector, weakening its position as TSMC's anchor customer

Opposed

  • Apple's predictable demand and broad multi-fab footprint provide strategic stability that TSMC cannot afford to abandon
  • AI capex may be a bubble fueled by venture capital spending; when it pops, Apple will regain leverage as the reliable anchor customer
  • TSMC needs high-volume smaller-die customers like Apple to amortize fab depreciation after early adopters subsidize yield learning
  • Apple has alternative options including Intel 18A and potential fab acquisition that provide long-term leverage against TSMC dependence
  • Nvidia's demand is concentrated and potentially volatile, while Apple spreads demand across both mature and leading-edge nodes providing diversification value