TSMC spends $56 billion to expand production, but the CEO admits that the chip shortage will continue beyond 2027.

TSMC is making a record-breaking capital expenditure bet on AI chip demand, but even so, the world’s largest foundry admits it is unable to keep pace with demand.

In this week’s earnings call, TSMC announced it expects capital expenditure in 2026 to reach $56 billion, with all funds going toward building new fabs and upgrading existing production lines. However, CEO C.C. Wei made it clear that despite the huge investment, the company still expects supply shortages to persist until 2027 and possibly longer, not just through 2026.

This statement directly reveals the core contradiction facing the semiconductor industry in the current AI super cycle: the speed of demand growth far outpaces the capacity-building cycle. From GPUs and CPUs to memory, as well as supporting materials such as voltage regulators, integrated circuits, and cables, the entire supply chain is currently in a state of shortage, restricting wafer order renewals for major customers like NVIDIA, AMD, and Apple.

Global Fab Construction Plans: Taiwan, USA, and Japan Advance in Parallel

To address the steadily increasing demand for 3-nanometer processes, TSMC is executing parallel capacity expansion plans worldwide.

In Taiwan, TSMC will add a new 3-nanometer fab within the GIGAFAB cluster at the Tainan Science Park, with mass production expected to begin in the first half of 2027. In Arizona, USA, the second fab will also use the 3-nanometer process; construction is complete, and mass production is scheduled for the second half of 2027. In Japan, TSMC plans to upgrade its second fab to the 3-nanometer process, with mass production expected in 2028.

C.C. Wei stated in the earnings call that, in addition to building new fabs, the company is continuously converting existing 5-nanometer capacity in Taiwan to support 3-nanometer production, and deploying a flexible capacity allocation mechanism between N7, N5, and N3 nodes to maximize support for all customers. He also emphasized that, amid tight capacity, TSMC will not show favoritism among clients.

Demand Pressure: AI Applications Penetrate the Entire Industry Chain

The main driving force behind this round of supply tensions is the explosive rise in AI demand. As Agentic AI surges, demand for high-bandwidth memory (HBM) and LPDDR is soaring, and the infrastructure consumption from AI applications has expanded from data centers to automobiles, IoT, and other endpoints.

Tech giants like NVIDIA, AMD, and Apple continue to update wafer orders, further intensifying TSMC's capacity pressure. C.C. Wei noted that once existing production lines reach peak capacity, upgrades and transformations will be initiated to meet incremental demand, but this process takes time.

Supply bottlenecks are also pushing some manufacturers to diversify their foundry arrangements. It is reported that Tesla is working with both TSMC and Samsung to develop next-generation AI chips, and is advancing a partnership with Intel on Terafab; Intel is rumored to win major customers this year with its 14A process technology; Samsung is seeing more inquiry from clients about its foundry business, but its main focus remains on HBM and other memory chip production.

Capacity Bottlenecks Hard to Resolve in the Short Term; Investors Should Mind Long-Term Risks

TSMC’s $56 billion capital expenditure plan is massive, but the cycle from fab construction to mass production usually takes years, meaning that even if funding is in place, capacity release cannot be immediate. The new fabs in Taiwan and the USA will not begin mass production until at least the second half of 2027, while the Japan plant must wait until 2028.

This timeline suggests that for the next two to three years, the tight supply situation for AI chips will likely persist, with delivery times and procurement cost pressures across related industries unlikely to ease quickly. For tech companies relying on advanced process chips, supply chain management and capacity-locking ability will be key variables influencing their competitive landscape.

Risk Warning and DisclaimerThe market entails risks; investments require caution. This article does not constitute personal investment advice, and does not take into account individual user’s particular investment goals, financial circumstances, or needs. Users should consider whether any opinions, viewpoints, or conclusions in this article suit their specific situations. Invest at your own risk.