Nvidia Dethrones Apple as TSMC's Biggest Customer in AI Era Shift

Decades ago, Jensen Huang made a bold promise to Morris Chang, founder of Taiwan Semiconductor Manufacturing Company: Nvidia would one day become the chip foundry's biggest customer. That prediction, once seeming audacious, has now become reality.

Nvidia will claim the top spot as TSMC's largest customer this year, displacing Apple from a position it has held for years. This isn't merely a ranking shuffle—it represents a fundamental realignment of the semiconductor industry, where AI infrastructure has definitively overtaken consumer devices as the primary growth engine.

The Numbers Tell the Story

The scale of this transformation is staggering. Creative Strategies analyst Ben Bajarin projects Nvidia will generate $33 billion in TSMC revenue this year, accounting for 22% of the foundry's total business. Apple, by comparison, is expected to contribute $27 billion, or 18% of revenue.

"The scale of this drastically changed," Bajarin noted. "A couple years ago, you could just see how much more capacity Nvidia was demanding from TSMC."

TSMC's financial results underscore this shift. High-performance computing sales, dominated by Nvidia's AI chips, jumped from 40% of net revenue in 2022 to 55% in the fourth quarter of 2024. AI accelerators alone now represent the "high-teens" percentage of TSMC's total sales.

The growth trajectories of both companies explain this reversal. Nvidia is expected to report 66% growth to $213 billion in sales for its fiscal 2026, while Apple's growth in fiscal 2025 was a modest 6.4%. Moreover, AI chips are significantly larger and more complex than smartphone processors, commanding higher prices per unit.

A New Supply Chain Dynamic

This customer swap reveals something profound about modern tech economics. Where Apple once provided TSMC with predictable, high-volume orders for relatively standardized chips, Nvidia represents a different kind of partnership—one driven by urgent, almost desperate demand from cloud providers racing to build AI infrastructure.

"Our customers' customer, who are mainly the cloud service providers, are also providing strong signals and reaching out directly to request the capacity to support their business," TSMC CEO C.C. Wei revealed during the company's recent earnings call.

Huang's commitment to this relationship is evident in his actions. He visited Taiwan five times last year, even donning TSMC's signature red shirt at the company's annual sports day. These aren't typical customer visits—they're the moves of someone who understands that TSMC's manufacturing capacity is the bottleneck constraining Nvidia's growth.

TSMC is responding with unprecedented investment plans. The company expects to spend up to $56 billion on capital expenditures this year, with potential increases to capture AI demand. But CEO Wei admits the stakes make him nervous: "I'm also very nervous about it, you bet. Because we have to invest about $52 billion to $56 billion for the CapEx, right?"

The Ripple Effects

This shift extends far beyond two companies changing positions on a customer list. It signals a broader transformation in how the semiconductor industry allocates its most precious resource: advanced manufacturing capacity.

For years, Apple's steady demand for cutting-edge chips justified TSMC's massive investments in new "leading node" technologies. The iPhone maker wanted the latest manufacturing processes for better power efficiency and longer battery life. Now, Nvidia has become that anchor customer, but for different reasons—AI systems require the most advanced chips not just for performance, but because energy efficiency directly impacts the return on investment for data centers.

"It just changes the dynamic where what was the driving force for TSMC—Apple—now shifts to Nvidia, and to some degree AMD, which is sort of the guarantee-scale customer that helps you justify the increase in CapEx to each new node," Bajarin explained.

This evolution also highlights TSMC's unique position as the world's dominant contract manufacturer, holding an estimated 70% of chip manufacturing revenue globally. While competitors like Intel promise to build leading-edge fabs in the US, they have yet to secure anchor customers of Nvidia's scale.

The Broader Questions

The semiconductor industry's pivot toward AI raises important questions about sustainability and market concentration. TSMC's dominance means that much of the world's most advanced chip production flows through a single company, creating potential vulnerabilities in global supply chains.

Moreover, the massive capital investments required to serve AI demand—TSMC alone is spending over $50 billion annually—represent bets on a technology trend that, while currently booming, remains relatively young. Wei's nervousness about these investments reflects a broader industry concern: What happens if AI demand doesn't justify these enormous expenditures?