TSMC's Global Buildout Faces Crucial Capital Efficiency Test

TSMC expects greater manufacturing scale in Arizona, productivity improvements and cooperation with suppliers and customers to moderate the cost premium. Gross-margin dilution measures the effect on consolidated profitability. It does not show the return on invested capital of individual factories, which TSMC does not disclose.

TSMC expects a new 3-nanometer (N3) fab in Tainan to enter volume production in the first half of 2027. Management has now approved another facility at the same location, scheduled to begin production in the first half of 2028.

Wei acknowledged that TSMC has historically avoided adding capacity to a process node after reaching its original target. The additional investment reflects demand that has exceeded earlier assumptions.

TSMC is also converting compatible 5-nanometer (N5) equipment to support N3 production, alongside productivity improvements and capacity optimization. These decisions have implications for capital efficiency.

Equipment conversions allow an established manufacturing base to support additional production, although conversion and qualification still require investment. Building a new fab involves a much larger capital commitment and a longer period before production reaches mature utilization.

The decision to add another N3 fab also signals that TSMC expects substantial demand for the process well beyond the introduction of N2 manufacturing.

TSMC has completed construction of its second Arizona fab and begun installing equipment. Volume production is now scheduled for the third quarter of 2027, slightly earlier than previously planned. In Japan, the company plans to introduce N3 manufacturing at its second fab, with volume production scheduled for 2028.

Arizona's accelerated schedule brings forward the opportunity to generate revenue. The factory must still progress through equipment qualification, yield improvement and utilization ramp-up before reaching mature operating economics.

TSMC has decades of experience transferring process technology between manufacturing sites. Its overseas factories must apply that experience within different labor markets, supplier networks and infrastructure conditions.

Government incentives and customer participation can lighten the capital burden of individual projects. How much depends on the terms of support, the timing of payments and the operating costs that remain once production begins.

A successful technology ramp will demonstrate manufacturing capability. The financial outcome will depend on the operating profit generated relative to the capital committed.

On the earnings call, Huang acknowledged rising equipment, material and labor costs. He reiterated that capital expenditure during 2026–2028 would be substantially higher than in the preceding three years.

Outside estimates suggest that spending could rise sharply again in 2027. According to Bloomberg, Goldman Sachs forecasts capital expenditure of $85 billion while UBS expects $90 billion, compared with a $75 billion average among analysts surveyed by the wire service.

These are analyst forecasts, not company guidance, but they nevertheless show the scale of capital commitments the market is beginning to anticipate.

Management expects revenue growth to outpace capital expenditure if its plans are executed successfully. TSMC's strong earnings and revenue outlook provide support that expectation, although concerns persist about the sustainability of the broader AI infrastructure buildout.