News

According to reports on monthly financial results, TSMC, the world’s leading semiconductor foundry, recorded its second-highest consolidated revenue in April, while cumulative revenue for the first four months of the year rose nearly 30% year-on-year. The strong performance reflects sustained demand for AI-driven chips, even as recovery in smartphones and consumer electronics remains gradual.

In April, TSMC reported consolidated revenue of approximately NT$410.726 billion, slightly down 1.08% from March’s all-time high, but still marking the second-highest monthly level in the company’s history. Compared with the same period last year, revenue increased by 17.5%. For the January–April period, total revenue reached NT$1.54 trillion, representing a robust 29.9% year-on-year growth. The figures underscore continued strength in artificial intelligence (AI) and high-performance computing (HPC) demand, with advanced process nodes and advanced packaging capacity remaining highly utilized.

Market observers note that orders from major clients such as NVIDIA, AMD, and several leading North American cloud service providers continue to flow strongly into TSMC’s advanced manufacturing lines. Demand is particularly strong for 3nm and 5nm process technologies. As agentic AI applications expand rapidly, both AI model scale and inference workloads are increasing significantly, driving a new cycle of global capital expenditure in AI infrastructure and further boosting demand for advanced chips.

During its recent earnings call, TSMC provided a strong second-quarter outlook, forecasting revenue between US$39 billion and US$40.2 billion. At the midpoint, this represents approximately 10.3% quarter-on-quarter growth. The company also expects gross margin to remain at historically high levels of 65.5% to 67.5%. Based on April’s performance, analysts believe TSMC is well positioned to meet its guidance, with potential upside if AI-related demand continues to accelerate.

Beyond advanced semiconductor processes, advanced packaging has become an increasingly critical growth driver. The evolution of AI GPUs toward larger-scale architectures and high-bandwidth designs has significantly increased demand for technologies such as CoWoS (Chip-on-Wafer-on-Substrate) and SoIC (System on Integrated Chips). Supply chain sources indicate that CoWoS capacity remains tightly constrained, with AI customers actively competing for allocation. In response, TSMC continues to expand capacity to support surging demand.

Industry trends also show a clear shift toward chiplet-based architectures and high-bandwidth memory (HBM) integration, further elevating the importance of advanced packaging solutions. Next-generation AI platforms such as Rubin and Rubin Ultra are expected to impose even higher requirements on package size, power efficiency, and bandwidth, positioning CoWoS and SoIC as key enabling technologies for the AI era.

Overall, AI remains the primary growth engine for the semiconductor industry. With long-term AI demand continuing to strengthen, market expectations remain positive that TSMC will achieve new highs in both revenue and profitability this year.