Customer Concentration Risk
Customer concentration risk is the exposure a foundry or supplier carries when a small number of customers, sometimes just one, account for an outsized share of its revenue, which means losing or shrinking that single relationship can threaten the business far more than an equivalent revenue loss spread across many smaller customers.
The math is simple, but the exposure compounds. A foundry with its top customer at 12 percent of revenue can absorb that customer leaving; a foundry with its top customer above 20 to 25 percent is effectively betting its own financial stability on one company's continued demand, and that customer knows it, which changes every negotiation that follows.
TSMC and Apple are the industry's most-discussed example. Public reporting on TSMC's customer mix has repeatedly placed Apple near a fifth to a quarter of total revenue in some years, a concentration that gives Apple significant influence over wafer pricing, capacity allocation priority during shortages, and in some cases how early TSMC commits a new node's first production slots to Apple's product cycle.
| Company | Top customer | Reported concentration | Mitigation underway |
|---|---|---|---|
| TSMC | Apple | historically near 20-25% of revenue | broad customer base across mobile, HPC, auto |
| Samsung Foundry | Qualcomm / internal Exynos | mixed internal and external demand | leans on internal device business as a buffer |
| GlobalFoundries | AMD (legacy), broad industrial base | diversified after 2010s restructuring | deliberately exited leading-edge chase |
| SMIC | domestic Chinese fabless firms | high reliance on domestic policy demand | state-backed demand as a concentration offset |
Concentration risk cuts the other way too. A fabless customer that depends on a single foundry for its most advanced node carries a mirrored version of the same risk, which is part of why large customers increasingly negotiate for dual-sourcing options or second-source qualification even at extra cost, trading some efficiency for protection against being the dependent party in a concentrated relationship.
Diversification is slow and sometimes deliberately avoided. GlobalFoundries reduced its concentration risk by exiting the leading-edge node race entirely and building a broader base of automotive, RF, and industrial customers after 2018; Samsung Foundry instead leans on its own internal device business as a demand floor, which is a different way of avoiding the same exposure without giving up leading-edge ambitions.
Read customer concentration through a *mutual hostage* lens rather than a *big customer is good* lens: once a single buyer is large enough to set pricing and allocation terms, the foundry has effectively built its roadmap around that one relationship, and the healthiest-looking revenue number on a quarterly earnings call can be hiding the single biggest strategic risk on the balance sheet.