- Scale. TSMC reported Q2 2026 revenue of $39.63 billion—a 36% year-over-year surge that landed at the upper bound of its own guidance and set an all-time company record for the month of June at $13.2 billion.
- Signal. The world’s largest contract chipmaker, which commands 73% of the global foundry market, said its most advanced production node (N3) is fully booked, with analysts describing AI chip supply as “still quite tight.”
- Stakes. The numbers, released July 13, set the tone for semiconductor earnings season and pushed consensus forecasts for TSMC’s Q3 profit to 58.8% above year-ago levels ahead of the company’s full earnings report Thursday.
Taiwan Semiconductor Manufacturing posted Q2 2026 revenue of NT$1.27 trillion, or $39.63 billion, according to results released July 13—a 36% year-over-year increase that arrived at the upper limit of the company’s own guidance range of $39.0B–$40.2B. June monthly revenue of NT$442.68 billion ($13.2 billion) represented a 67.9% jump year-on-year and a 6.2% sequential rise from May, setting an all-time record for a single month in TSMC’s history. First-half 2026 revenue reached $74.99 billion, up 35.6% from H1 2025.
Why the numbers are so large
TSMC derives its outsized position from an unusually concentrated market: the company holds approximately 73% of the global pure-play foundry market as of Q1 2026, meaning that most of the world’s leading-edge chips—including those from Nvidia, Apple, AMD, and Qualcomm—are manufactured in its facilities. The current demand cycle is driven primarily by AI infrastructure buildout: hyperscalers ordering AI accelerators, and those accelerators being manufactured on TSMC’s N3 process node, which is producing at or near capacity.
Sravan Kundojjala of SemiAnalysis described the conditions as a seller’s market: “demand supply situation in AI is still quite tight and TSMC is sold out on N3.” Advanced packaging, specifically TSMC’s CoWoS (Chip-on-Wafer-on-Substrate) process used to stack memory alongside AI processors, is also booked out—a point that has constrained TSMC’s ability to convert demand into faster revenue growth even further.
What the numbers imply for the sector
TSMC’s results are closely watched as a leading indicator for the broader semiconductor and AI hardware supply chain, because its customers include most of the world’s major chip designers. A 36% revenue gain—and a June print that significantly exceeded combined April–May growth rates—signals that AI server buildout accelerated into mid-year rather than plateauing as some analysts had projected.
The figures arrive as context for a sector still absorbing prior volatility. The full earnings release Thursday will provide gross margin and operating income detail that analysts will use to assess whether pricing power is holding as the company scales production on next-generation nodes. Consensus currently projects a 58.8% year-on-year profit gain for Q3, according to Yahoo Finance analyst data.
The trajectory has implications for the memory side of the AI chip stack. Earlier this year, Micron guided Q4 revenue to $50 billion on AI memory demand that outstripped supply—a signal that TSMC’s logic-chip bottleneck is mirrored by equivalent tightness in the high-bandwidth memory market that sits alongside its processors.