TSMC to raise chipmaking prices by up to 10% from 2027

What's happening

Taiwan Semiconductor Manufacturing Company (TSMC), the world's largest dedicated semiconductor foundry with a market capitalization of approximately $2.09 trillion, is set to implement base price increases of 5–10% on advanced-node chipmaking services beginning January 2027, Nikkei Asia reported on July 21, 2026. Potential service premiums, particularly on high-performance computing workloads, could push total price increases for some customers to as high as 25%. Mature-node processes — including 12-nm, 16-nm, and 28-nm geometries — will also see increases of up to 10%, meaning the repricing is not confined to leading-edge production.

Customer negotiations began in June 2026 and concluded in July 2026, with the new pricing structure set to take effect at the start of 2027. TSMC CEO C.C. Wei signaled the direction of travel in June 2026, stating he would like to raise prices while also indicating the company would refrain from abrupt hikes. A TSMC spokesperson elaborated on the company's position: "Our pricing strategy is strategic, not opportunistic. We will continue to work closely with customers and sell our value to them." The increases are attributed to rising costs for materials, equipment, and the buildout of overseas manufacturing facilities.

Why it matters for markets

Cutting-edge nodes at 7-nm and below accounted for approximately 77% of TSMC's revenue in the second quarter, with mature nodes contributing the remaining 23%. That revenue concentration means the 5–10% base price increase on advanced nodes — and the potential 25% ceiling on HPC services — applies to the substantial majority of TSMC's foundry business, and by extension to the cost structures of the fabless companies that depend on it. Nvidia, whose data-center GPU and AI accelerator portfolio is manufactured on leading-edge TSMC nodes, and AMD, whose EPYC server processors and Instinct AI accelerators similarly rely on advanced TSMC processes, are both named as affected clients. For Nvidia, which reported annual revenue of $253.49 billion, and AMD, which reported annual revenue of $37.45 billion, higher wafer costs flowing through from January 2027 onward represent a direct input-cost pressure that will need to be absorbed, passed on to customers, or offset through other operational efficiencies.

The downstream implications extend beyond the two primary fabless clients. Electronics manufacturers and system integrators that purchase chips incorporating TSMC-produced silicon — across data-center servers, consumer devices, and automotive platforms — may face higher component costs as the repricing works through the supply chain. The breadth of the increase, spanning both cutting-edge and mature nodes, limits the ability of buyers to substitute toward older process geometries as a cost-avoidance strategy, since mature-node pricing is also rising by up to 10%.

Sectors and assets to watch

The most direct exposure sits with the two named fabless clients: Nvidia (NVDA), whose H100 and successor AI accelerators are central to data-center infrastructure spending, and AMD (AMD), whose EPYC and Instinct product lines compete in the same high-performance computing and AI training markets. Both companies design chips without owning fabrication capacity, making TSMC's pricing a non-negotiable input cost. Any margin compression or pricing adjustments at either company would have secondary effects on the hyperscale cloud providers and enterprise customers that purchase their products.

Beyond Nvidia and AMD, the semiconductor supply chain more broadly warrants attention. Companies reliant on TSMC for mature-node production — spanning automotive semiconductors, industrial chips, and consumer electronics components — face the same up-to-10% increase on 12-nm, 16-nm, and 28-nm processes. Equipment suppliers and materials vendors cited by TSMC as cost drivers behind the repricing are also implicated, as their pricing power appears to be flowing upstream into foundry economics and, ultimately, downstream into end-product costs.

What to watch next

Key developments to monitor include how Nvidia and AMD address the higher input costs in their own pricing and margin guidance when they next report earnings or provide forward-looking commentary. The extent to which the 25% HPC service premium is applied broadly or selectively — and which specific node generations or service tiers attract the highest surcharges — will clarify the full financial impact on leading AI chip customers. Additionally, any public response from other major TSMC clients such as Apple, which is not named in the current reports but relies heavily on TSMC's leading-edge nodes, could indicate how broadly the repricing is being contested or accepted across the customer base. Progress on TSMC's overseas fab expansions, cited as a cost driver behind the increases, will also be relevant context for assessing whether further pricing adjustments follow in subsequent years.