TSMC posts strong Q1
TSMC reported a big revenue jump this quarter, driven by ongoing demand for AI compute rather than a short-term blip. The foundry posted T$839.25bn in first‑quarter revenue, up 35% year‑on‑year and above market forecasts, with analysts tying the strength to AI workloads and server shipments. That surge signals sustained wafer demand and underlines why system-level constraints—memory, power delivery and board integration—are dominating design tradeoffs today. (reuters.com)
Taiwan Semiconductor Manufacturing Company just put up another quarter that was too big to dismiss as a one-off spike: March 2026 revenue hit New Taiwan dollar 415.19 billion, up 45.2% from a year earlier, and first-quarter revenue reached New Taiwan dollar 1.134 trillion. The company said those January-through-March sales were 35.1% higher than the same period in 2025. (pr.tsmc.com) That number matters because Taiwan Semiconductor Manufacturing Company is not a chip brand like Nvidia or Apple. It is the factory behind many of their most advanced chips, so its monthly sales are a direct read on how many leading-edge wafers customers are actually buying. (cnbc.com) The jump also landed above what the market had penciled in. Reuters reported analysts were looking for about New Taiwan dollar 1.12 trillion for the quarter, which means the demand wave was stronger than expected even after a year of artificial intelligence spending headlines. (reuters.com) This is showing up first in servers, not phones. Reuters said analysts tied the quarter’s strength to artificial intelligence workloads and server shipments, which means cloud companies are still ordering the processors that train and run large models inside data centers. (reuters.com) Taiwan Semiconductor Manufacturing Company had already warned investors in January that 2026 would stay heavy on artificial intelligence. In its fourth-quarter 2025 results, the company guided first-quarter 2026 revenue to 34.6 billion to 35.8 billion US dollars and said revenue from artificial-intelligence accelerators was expected to double in 2026 after tripling in 2025. (investor.tsmc.com) The bottleneck is no longer just making the main chip. DigiTimes reported on April 10 that Taiwan Semiconductor Manufacturing Company’s Chip on Wafer on Substrate packaging capacity remains in severe shortage, and that packaging has become one of the hardest constraints in getting artificial-intelligence systems out the door. (digitimes.com) Chip on Wafer on Substrate is the step that turns a bare processor into something a server can actually use. It stacks and connects the compute chip with high-bandwidth memory, which is the fast memory sitting right next to the processor so data does not have to travel across the whole motherboard. (digitimes.com) That is why a strong quarter at Taiwan Semiconductor Manufacturing Company tells you more than “chip demand is good.” It says the artificial-intelligence buildout is still being limited by system plumbing like memory supply, power delivery, and board-level integration, because those are the parts that decide how many servers can be assembled after the wafer leaves the fab. (reuters.com; digitimes.com) The next checkpoint is April 16, 2026, when Taiwan Semiconductor Manufacturing Company is scheduled to publish full first-quarter earnings with profit and margin details. The revenue release answered one question already: customers are still placing large orders for advanced silicon, and they are doing it at a scale big enough to push quarterly sales past New Taiwan dollar 1.1 trillion. (cnbc.com; pr.tsmc.com)
