During the fourth-quarter conference call, TSMC’s CEO C.C. Wei described himself as "very nervous" about sustaining AI demand, but immediately backed that caution with action: a planned capital expenditure of up to $56 billion to expand production capacity. This investment represents roughly 45% of the company’s 2025 revenue and underscores management’s conviction that long-term AI computing needs justify significant new fabs and equipment purchases. Wei noted that extensive due diligence with hyperscale cloud and data-center customers confirmed that generative AI workloads remain under-provisioned, driving his decision to accelerate spending on advanced process nodes (3 nanometer and below) through 2027.

In the most recent quarter, TSMC delivered 26% year-over-year revenue growth, building on a full-year increase of 36% in 2025, when total revenue reached $122 billion. Management has guided for another 30% top-line advance in 2026 and forecasts a compound annual growth rate of 25% through 2029. High-performance computing, which includes AI training and inference chips, accounted for 58% of revenue last year and grew 48% year over year. With hyperscalers planning to invest hundreds of billions more in data-center expansion through 2030, TSMC’s expansion plans align directly with the strongest demand segment of the semiconductor industry.

Despite a more than 300% share-price increase since early 2023, TSMC trades at approximately 25 times forward earnings—below the typical 30-times multiple assigned to large U.S. technology companies and only slightly above the broader market’s 22-times multiple. Analysts expect 2026 earnings per share to rise by roughly 23% to $13.05, reflecting management’s guidance for a 30% revenue increase. Gross margin expanded from 59% in 2024 to 62.3% in the fourth quarter of 2025, with long-term targets above 56%, suggesting the company can sustain profitability even as it ramps up capital intensity.