Article: TSMC is on track to post a net profit of about T$632.6 billion for the April-June quarter, a rise of roughly 59 % from a year ago, and to break its own record for quarterly earnings. The surge comes as AI-driven demand for its 3-nm and 2-nm processes and its CoWoS packaging hits an unprecedented level, a development that could reshape the economics of the whole semiconductor supply chain.

The AI engine behind the earnings

The numbers aren’t a surprise to anyone watching the chip market this year. TSMC’s customers are flooding orders into the most advanced nodes because today’s AI models need billions of parameters and can only run when the underlying silicon delivers massive compute per watt. Nvidia’s latest data-center GPUs and Apple’s upcoming AI-focused silicon both sit on TSMC’s 3-nm and 2-nm processes, and the company’s CoWoS (chip-on-wafer-on-substrate) packaging lets those dies communicate at the speeds required for large-scale inference and training workloads.

A LSEG SmartEstimate consensus expects revenue to have already jumped 36 % in the quarter, well ahead of market forecasts, and the profit margin is set to expand as higher-value AI chips replace lower-margin logic and mobile business. The premium product mix and the ability to charge a price premium for the most advanced geometry turn a strong quarter into a record-breaking one.

What the market value says

At a market capitalization near $1.95 trillion, TSMC now tops Asia’s listed companies and is almost twice the size of South Korean rival Samsung Electronics. The share price in Taipei has risen 57.4 % so far this year, reflecting investor confidence that the AI wave will keep the fab’s utilization high and its pricing power intact.

That valuation isn’t just vanity. It gives TSMC the financial bandwidth to fund a $165 billion program to build new fabs in Arizona, a move aimed at diversifying production away from Taiwan’s geopolitical risk profile. In April the company hinted that its 2026 capital-expenditure plan could land at the top of the $52-$56 billion range it gave earlier, signalling that cash flow from the current quarter will be reinvested to lock in future capacity.

Risks that could temper the rally

The upside isn’t without headwinds. The same geopolitical tensions that motivate the Arizona build also threaten supply-chain stability. Any disruption in Taiwan would force TSMC to shift production to overseas sites, which are still ramping up and may not yet match the yield and volume of the home-base fabs. Competitors are also accelerating their own advanced-node roadmaps; a successful 2-nm rollout by rivals could erode TSMC’s technology moat.

Capacity constraints loom as well. While demand for AI chips soars, the advanced nodes are inherently low-yield. If wafer yields fall short of expectations, TSMC may have to turn away orders or raise prices, which could dampen the very AI-driven demand fueling its current growth. Customers such as Nvidia and Apple have been known to diversify across multiple fabs to hedge against supply shocks, a practice that could limit TSMC’s share of future AI spend.

What to watch next

  • Quarterly earnings release: The actual profit figure will confirm whether the company clears the T$572.5 billion threshold needed for an all-time high.
  • Yield reports for 2-nm: Early data on the new node’s production yield will show how quickly TSMC can scale the most advanced chips.
  • Progress on the Arizona fabs: Construction milestones and any regulatory hurdles will affect the timeline for adding U.S. capacity.
  • Customer order trends: Shifts in the order books of Nvidia, Apple, and other AI players will reveal whether the current demand surge is a temporary spike or the start of a sustained uptrend.

The profit forecast signals that AI is no longer a niche add-on for TSMC; it is now the central driver of its financial engine. If the company keeps yields high, expands capacity without major geopolitical disruptions, and stays ahead of competing fabs, the record profit could be the first of many quarters where AI chips dominate the bottom line. If any of those variables wobble, the same premium pricing that fuels the current boom could quickly turn into pricing pressure, dragging earnings back toward more modest levels.