According to Jensen Huang, CEO of NVIDIA, TSMC will need to double its manufacturing capacity over the next ten years just to meet NVIDIA’s own demand. This statement highlights how strong and long-lasting the growth of artificial intelligence is becoming.
The rapid expansion of AI is putting heavy pressure on the entire semiconductor supply chain, with TSMC at the center of this shift. Speaking to local media, Huang explained that TSMC is planning an expansion of more than 100%, calling it the largest industrial infrastructure build in human history.
TSMC’s Massive Expansion Driven by AI Demand
TSMC’s recent capital expenditure clearly supports Huang’s view. Over the past quarters, the company has launched multiple new factory projects and sharply increased its investments. This growth is also influenced by geopolitics, with production spreading across strategic regions such as the United States, the European Union, and Japan.
In the US alone, TSMC is planning investments worth around $250 billion, covering advanced chip manufacturing, packaging, and research and development. The Arizona facilities are already moving toward 3nm production, with plans to later adopt the A16 node, while still following a careful “N-2” technology strategy.
Another key point is NVIDIA’s growing role as TSMC’s main client. With platforms like Grace Blackwell and Vera Rubin, NVIDIA now uses a large share of TSMC’s production lines, overtaking Apple as the company’s top customer in just a few years.
TSMC’s introduction of capacity prepayments further strengthens this position, giving NVIDIA priority access to future manufacturing lines. Thanks to its scale and strong relationship with TSMC, NVIDIA holds a major competitive advantage, making pressure from rivals like AMD a secondary concern for now.
