Taiwan semiconductor manufacturing controls critical supply path for AI chips as demand surges.
Taiwan Semiconductor Manufacturing Company (TSMC) is a semiconductor foundry that manufactures chips for Advanced Micro Devices, Inc. (AMD), NVIDIA Corporation (NVDA), and 530 other companies. TSMC dominates the foundry business with a 72.3% market share. The global semiconductor foundry industry expanded from $28.3 billion in 2010 to $169.5 billion in 2025, while TSMC's revenue increased from $13.3 billion to $122.5 billion over the same period. Notably, TSMC's $122.5 billion in 2025 revenue alone exceeds the combined revenue of all other foundry companies, which totaled $47.0 billion.
As demand for artificial intelligence infrastructure has accelerated, the key constraint has shifted from chip design to manufacturing capacity at advanced nodes. TSMC dominates other foundries, particularly Samsung Electronics (SSNLF), at nodes to 5nm. This reflects a structural shift in the industry: as semiconductor manufacturing has moved to more advanced process nodes, the number of companies capable of competing at the leading edge has narrowed. TSMC's scale and execution at these nodes have allowed it to pull away from the rest of the industry. Conversely, at nodes greater than 7nm, where chips have lower manufacturing costs and revenues dilute earnings, leaders include Chinese foundries Semiconductor Manufacturing International Corporation (SMIC) and HHGrace.
TSMC's High Performance Computing revenues demonstrate its exposure to the AI boom. In Q2 2023, NVIDIA's data center revenues jumped 140%, marking the start of significant demand as its H100 GPU began deployment in data centers. Between the H100 ramp in 2023 and the Rubin generation in 2026, NVIDIA introduced the H200 in late 2023 and the Blackwell-based B100/B200 platform, announced in Q1 2024 and ramping through 2024–2025. AMD introduced the MI300A/MI300X platform in late 2023 and began ramping deployments through 2024–2025, but clearly without the same scale or acceleration demonstrated by NVIDIA's data center growth, as AMD continues to see a mixed environment with cloud customers optimizing data center compute while enterprise customers remain cautious with new deployments.
TSMC's advanced technology nodes have become the primary driver of both revenue growth and average selling price (ASP) expansion. Technologies at 7nm and below, including N7, N5, and N3, represented 74% of wafer revenues in Q1 2026, reflecting continued migration toward more complex and higher-value process technologies. ASP expansion correlates not only with price increases but also with product mix, as smaller node wafers are priced higher; over time, a greater portion of TSMC's wafer revenue has shifted toward advanced process nodes.
While NVIDIA captures direct exposure to AI demand through its GPU products and AMD participates as an emerging competitor, TSMC's role in the AI ecosystem is fundamentally different: it captures the manufacturing economics that enable both companies' growth. TSMC is not simply a supplier to NVIDIA and AMD—it is the manufacturing foundation that determines how much of the current AI demand cycle can be realized. As AI workloads continue to drive demand for advanced processors, TSMC's position in the semiconductor value chain remains central to that growth, capturing the underlying economics through advanced-node manufacturing where supply remains constrained.