Taiwan Semiconductor Manufacturing Company reported first quarter 2026 results on April 16 that were ahead of every major analyst estimate. Revenue came in at NT$1.134 trillion, roughly $35 billion at the current exchange rate. Net profit was up 58% year over year, which is the fastest quarterly profit growth the company has posted since the original artificial intelligence capital expenditure cycle began in 2023. Management raised its full-year revenue outlook and increased its planned capital spending range for the year. The result landed at a moment when Wall Street had been asking out loud whether the AI spending wave was finally cooling off. The numbers say not yet.
The headline explanation for the quarter is straightforward. Advanced node chips built for AI training and inference still account for a larger share of TSMC's revenue mix each quarter. Three nanometer and five nanometer processes together represented more than 55% of wafer revenue in Q1 2026. That is the highest concentration ever reported. The company's largest customers on those nodes remain the hyperscale cloud providers and the dedicated AI accelerator designers. Nvidia, AMD, Google, and Amazon are each expanding their chip orders for 2026 and 2027 to meet training demand for next-generation foundation models and inference demand for production deployments inside enterprise software.
What makes this cycle different from earlier semiconductor cycles is the pattern of customer concentration. Semiconductor demand has historically moved with consumer electronics replacement cycles and with enterprise server refresh cycles. Those drivers still exist, but they are now second-order compared to AI infrastructure buildout. TSMC told analysts on the earnings call that it expects AI-related revenue to grow at a faster rate in 2026 than in 2025. That is notable given how large the base has become. The mathematical implication is that the dollar increase from AI this year will be larger than the increase from any other product segment combined.
Capital spending plans were raised. The company now expects to spend between $42 billion and $46 billion on capital expenditures in 2026. That range is up from a prior estimate of $38 billion to $42 billion. A significant portion of the new spending is earmarked for advanced packaging capacity, which has become the genuine bottleneck for AI chip production. Chip-on-wafer-on-substrate packaging, the specific technology used in most high-end AI accelerators, has been sold out through 2026 for nearly a year. Management said capacity is expanding but will not meet demand at scale until 2027. That is useful information for anyone tracking the downstream supply picture.
For investors, the result has three implications. First, the AI capital expenditure cycle is not decelerating on the demand side. Second, the supply constraints that have held back AI deployment are real and will take another year to ease. Third, companies on the right side of both of those dynamics are generating extraordinary cash flow. TSMC's operating margin in Q1 was 49%, up from 42% a year earlier. That level of margin at this revenue base produces free cash flow that funds future expansion without outside capital. The company is also continuing to pay dividends and execute buybacks.
The broader market context matters. According to Goldman Sachs research published in April, consensus estimates for 2026 global AI capital spending among the major hyperscalers now stands at $527 billion. That figure is up from an earlier estimate of $465 billion at the start of the fourth quarter 2025 earnings season. Information technology is projected to grow earnings per share by roughly 44% in the first quarter. Goldman estimates that AI infrastructure investment will account for 40% of all S&P 500 earnings growth in 2026. These numbers place TSMC's report in context. It is not a single company story. It is the infrastructure story.
The risks are also real. Geopolitical tension between the United States and China over Taiwan remains the structural risk that investors price against TSMC every quarter. The company has been diversifying manufacturing with facilities in Arizona, Japan, and Germany, but the core advanced node production stays on the island. Export controls from the United States on the most advanced chips continue to move. Any change to the current export framework affects revenue composition and the pace at which capacity utilizes. Customer concentration is also a risk. A slowdown in orders from one or two large AI accelerator customers would move the numbers meaningfully.
What to watch next is the Q2 guidance and the pace at which AI-specific revenue growth accelerates compared to other segments. Analysts will also be watching margin trends. If pricing power holds at current levels into the second half of 2026, earnings estimates will move higher. If pricing pressure emerges as packaging capacity expands, margins could compress from the current peak. Either way, the conversation about whether the AI spending cycle was ending has a clearer answer now. The cycle is still running. The next question is what the ceiling looks like, and nobody has a good estimate on that yet.