Taiwan Semiconductor Manufacturing Company is warning key customers that it cannot fully meet demand for advanced artificial intelligence processors, underscoring how the global artificial intelligence boom is straining the semiconductor supply chain. The world’s largest contract chipmaker has notified major clients including NVIDIA and Broadcom that capacity at its most advanced manufacturing nodes is increasingly constrained, as artificial intelligence workloads scale across cloud computing, data centres and enterprise applications. With hyperscale providers, chipmakers and system designers all seeking priority access to cutting-edge production, industry analysts say lead times at advanced nodes are lengthening and some customers are exploring alternative suppliers to avoid delays in artificial intelligence infrastructure rollouts.
Despite these bottlenecks, Taiwan Semiconductor Manufacturing Company’s latest financial results illustrate the strength of artificial intelligence-driven demand. The company reported a 35% year-on-year increase in fourth-quarter profit, reaching a record high and beating market expectations. Revenue for the quarter came in at US33.7bn, while net income rose to US16.3bn. High-performance computing, which includes artificial intelligence and 5G applications, accounted for 55% of total revenue in the quarter, and advanced chips measuring seven nanometres or smaller made up 77% of wafer revenue. On an earnings call, chief financial officer Wendell Huang said the business will be supported by continued strong demand for leading-edge process technologies, and the company guided for current-quarter revenue of between US34.6bn and US35.8bn, representing growth of up to 38% year on year. Taiwan Semiconductor Manufacturing Company has begun mass production of 2 nm chips and expects capital expenditure to rise to between US52bn and US56bn in 2026, up from US40.9bn in 2025, as it ramps capacity for next-generation processes.
The tight supply environment is reshaping competitive dynamics, giving Intel a chance to reassert itself in chip manufacturing as a supplementary foundry for customers seeking diversification. Analysts argue Intel does not need to overtake Taiwan Semiconductor Manufacturing Company to benefit, but can act as a pressure valve by offering scarce foundry capacity, while its manufacturing footprint aligns with United States industrial policy and geographic risk concerns. Investor sentiment has improved, with Intel’s shares rising 19% so far in 2026, and momentum strengthened after supportive comments from United States President Donald Trump, who praised Intel’s latest processors following a meeting with chief executive officer Lip-Bu Tan. NVIDIA has already invested in Intel, and speculation that Apple could use Intel’s foundry services for parts of its chip production highlights growing confidence in its turnaround. In parallel, Taiwan Semiconductor Manufacturing Company is expanding globally with major projects in Japan, Europe and Arizona, where chief executive officer C.C. Wei plans a gigafab cluster to better serve United States customers, although the company warns that fabs outside Taiwan will operate at diluted margins and that global tariff policies and memory market volatility present emerging risks, even as artificial intelligence drives a structural shift in semiconductor demand.
