Key Takeaways:
TSMC's chairman sees no letup in AI-driven chip demand as the industry shifts from generative to agentic artificial intelligence.
TSMC expects full-year 2026 dollar revenue to grow more than 30%, Chairman C.C. Wei said, as a structural shift toward agentic AI drives demand for advanced semiconductors.
"We continue to see increasing adoption of AI models across consumer, enterprise and sovereign AI applications," Wei said at the company's annual shareholders meeting in Hsinchu, Taiwan, on June 4. "This trend is driving demand for greater computing power, which in turn supports strong demand for advanced semiconductor chips."
The forecast, delivered during Computex Taipei week where Nvidia Corp. and Intel Corp. executives gathered in the city, highlights TSMC's position as the primary beneficiary of AI infrastructure spending. Wei attributed the sustained demand to a shift in how AI is used — from early generative AI and query-based models toward agentic AI and instruction-action modes — which increases the token consumption required by large language models and pushes compute requirements higher.
TSMC, the world's largest contract chipmaker, produces advanced processors for Nvidia, Advanced Micro Devices Inc. and Apple Inc. The 30%-plus growth target implies revenue exceeding $110 billion in 2026, based on the company's 2025 dollar revenue of about $86 billion. Any slowdown in AI spending would hit not just TSMC but the entire semiconductor supply chain from ASML Holding NV to chip packaging providers.
Wei said TSMC's customers and their customers continue to provide positive outlooks for the AI industry. The company maintains strong confidence supported by technology differentiation and a broad customer base, he added.
The comments come as Intel Corp. Chief Executive Officer Lip-Bu Tan, speaking at Computex, described TSMC as a "trusted partnership" and said Intel would continue outsourcing advanced chip production to the Taiwanese foundry. "We are a big, top customer for them, and we're going to continue doing that," Tan said.
The AI-driven demand surge is also rippling through the semiconductor supply chain. Aspeed Technology Inc., a supplier of baseboard management controllers used in servers, expects revenue to grow significantly this year and next year, Chairman Chris Lin said at Computex, citing "skyrocketing server demand" as agentic AI applications gain traction. Aspeed has secured large-scale orders from customers, with shipment volumes doubling from forecasts provided at the end of last year, Lin said.
TSMC's growth trajectory faces some near-term constraints. Aspeed flagged supply shortages of key components, particularly substrates, which have pushed some shipments into next year. The company said it expects the supply issue to improve in the fourth quarter, later than previously anticipated.
For investors, TSMC's reaffirmation of above-30% growth signals that the AI capex cycle remains intact despite concerns about diminishing returns on large language model training. TSMC trades at about 20 times forward earnings, a discount to Nvidia's 35 times, reflecting its role as a manufacturing proxy rather than a direct AI platform play. The widening adoption of agentic AI — which requires more inference compute per user interaction than simple query-based models — could extend the investment cycle beyond initial training infrastructure buildouts.
This article is for informational purposes only and does not constitute investment advice.
