Nvidia is gradually displacing Apple as the key customer of Taiwanese chip manufacturer TSMC. The reason for this shift is the explosive growth in demand for artificial intelligence chips, backed by multi-billion dollar investments.
The essence of the problem lies in the scale of production. The Apple A19 processor, which is used in the iPhone 17, is manufactured using TSMC's N3P process and occupies a relatively small area of about 100 square millimeters. This allows hundreds of such chips to be placed on a single standard 300-millimeter wafer.
At the same time, the Nvidia Rubin graphics processor consists of two huge chips, each measuring 750-800 square millimeters — the maximum size that TSMC is capable of producing. As a result, no more than 100 chips can be produced from a single wafer, and not all of them are suitable for use.
Although Apple sells hundreds of millions of iPhones annually, Nvidia's wafer orders were significantly higher due to the enormous size of its AI chips. This led to a redistribution of TSMC's production capacity in favor of the AI accelerator developer.
However, Nvidia's current dominance is unlikely to be long-lasting. Apple still has serious advantages: huge cash reserves, stable annual sales of millions of devices, and a long-standing partnership with TSMC. These factors make the company a reliable long-term partner, despite current market shifts.
Against the backdrop of order redistribution, other significant trends are also emerging. Intel is actively expanding its market presence, bidding for contracts with major players such as Apple, Broadcom, and Google, as well as potentially Nvidia. Taiwan, meanwhile, continues to pursue its strategy of localizing advanced manufacturing. For its part, TSMC is continuing to develop and improve its technical processes—from N4 to N3—which are necessary for the production of new generations of AI accelerators.