TSMC posts record profit for a sixth straight quarter, beating estimates on AI boom

TSMC delivered another record quarter as demand for AI chips continued to surge. It reported NT$1.49 trillion in revenue and NT$763.1 billion in net income. Net profit set a sixth consecutive quarterly record and jumped 68.7% from a year earlier.

Revenue exceeded LSEG SmartEstimates of NT$1.47 trillion by NT$20 billion. Net income beat the NT$746.34 billion estimate by NT$16.76 billion. The company’s operating margin also reached 58.7%, above the top of its 58% guidance range.

The results show how strongly AI-related semiconductor demand is supporting TSMC’s business. The company expects fourth-quarter revenue of $52.3 billion to $54.3 billion. At the midpoint, that would represent a 14% increase from the third quarter and a 58% increase from a year earlier.

A contract chipmaker, also called a foundry, manufactures semiconductors for companies that design chips but outsource production. This arrangement lets chip designers focus on architecture, software, and customers while the foundry handles complex fabrication. The source does not define the term, so this explanation uses established industry knowledge.

TSMC is especially important to AI processor designers because it offers advanced manufacturing technologies at enormous scale. Its 7-nanometer and more advanced processes generated 79% of wafer revenue. The company also expects demand for its 2-nanometer process to grow as it ramps production.

That manufacturing capability makes TSMC a key link between AI chip designs and physical products. Its results show the connection: high-performance computing produced 68% of revenue, while demand for AI accelerators and networking chips remained strong. Continued customer demand is driving further investment.

TSMC’s AI-related opportunity is large because two major parts of its business are closely connected to AI hardware. High-performance computing accounted for 68% of revenue, making it the company’s largest platform. Advanced technologies, defined as 7-nanometer and more advanced processes, contributed 79% of total wafer revenue.

The figures show strong exposure to powerful computing and leading-edge manufacturing. In the third quarter, 3-nanometer chips alone represented 33% of revenue, the largest share among the listed process nodes. TSMC also highlighted AI accelerators, processors for agentic AI applications, and networking chips.

These numbers do not mean 68% or 79% was exclusively AI revenue. They do show that TSMC is positioned where AI demand is concentrated. If adoption of agentic AI and open-source models expands, demand for advanced chips could support higher production, investment, and revenue.

Operating margin measures how much operating profit remains from revenue after operating costs. TSMC’s 58.7% third-quarter margin exceeded the upper end of its 58% guidance. The company linked this performance to pricing power as the AI boom increased demand for semiconductors.

AI accelerators and networking chips require advanced manufacturing technologies. TSMC said demand for both remained strong, alongside demand for processors used in agentic AI applications. Its leading processes also represented most wafer revenue, with advanced technologies at 79%.

When customers strongly need a specialized service, a supplier can have more ability to maintain or improve prices. That is the mechanism behind TSMC’s pricing power, according to the article. Continued demand could support margins, but the company must also spend heavily. It expects capital expenditure from 2026 through 2028 to be significantly higher than in the previous three years.

Manufacturing capacity is the amount of chip production available to meet customer orders. If demand for advanced chips rises faster than factories can expand, the global AI supply chain could face tighter availability. This matters because AI systems depend on processors, accelerators, and networking chips.

The article points to the pressure already building. TSMC reported strong demand for AI accelerators and advanced technologies, and it expects capital expenditure from 2026 to 2028 to be significantly higher than during the previous three years. Its second Arizona fab is being equipped for future production.

A sustained mismatch could lead to longer waits, tougher competition among customers, and greater urgency to build capacity. Those consequences are a general supply-chain implication, not figures stated in the article. TSMC’s planned investments and possible U.S. expansion suggest the industry is responding by increasing manufacturing capacity and geographic reach.

TSMC is building more U.S. factories as it responds to growing customer demand and seeks to strengthen its American manufacturing presence. The company has outlined a planned Arizona investment totaling $265 billion. The article also reports that it is evaluating a possible investment in Texas.

TSMC has completed construction of its second Arizona fab and is installing equipment there. Mass production is expected to begin in the third quarter of 2027, slightly ahead of its previous schedule. A Texas facility remains under consideration, rather than a confirmed project.

If these facilities operate as planned, more advanced chip manufacturing would take place in the United States instead of being concentrated mainly elsewhere. Arizona would become an important production location, while Texas could add another regional center. The article does not specify which processes a Texas facility would use or how much it might cost.

Semiconductors are materials whose electrical behavior can be controlled, making them useful for building the tiny switching devices in computer chips. A nanometer manufacturing process refers to a technology generation associated with extremely small chip features. The article identifies 7-nanometer and more advanced technologies as advanced processes.

In general industry knowledge, smaller process nodes can place more transistors in a given area. More transistors can help chips perform more calculations or use energy more efficiently. TSMC reported that 3-nanometer chips accounted for 33% of third-quarter revenue and said it was ramping its 2-nanometer process.

These technologies matter because AI workloads require powerful, efficient processors. TSMC expects fourth-quarter business to receive support from advanced chipmaking demand, including the 2-nanometer ramp-up. The source gives these manufacturing and revenue facts but does not itself explain why smaller nodes can improve performance or energy use.