Assessing the profitability of TSMC's nodes: Analyzing the numbers

TSMC's older nodes are surprisingly profitable due to depreciation benefits.

: In 2024, TSMC earned nearly half of its revenue from nodes over five years old, such as 7nm and older, unlike Intel's approach of shutting down old nodes. Advanced nodes like 3nm and 5nm contribute 52% of TSMC's revenue but only 27% of profit, indicating higher costs. The profitability of older nodes remains substantial as they have depreciated fully, with depreciation considered a major expense impacting Cost of Goods Sold. The model used considers allocation methods for depreciation, R&D expenses, and SG&A, highlighting the economic advantage of well-depreciated nodes.

In 2024, TSMC's revenue from nodes that are five years or older made up almost 50% of its income, crucially setting it apart from the strategy employed by Intel, which closes older nodes upon shifting to new processes. This substantial revenue from older nodes contrasted with Intel's challenges in entering the foundry business illustrates a strategic advantage for TSMC, allowing continued operation of older facilities to boost profits. The data shows that the advanced nodes, including 3nm and 5nm, collectively produce 52% of TSMC's revenue yet only contribute 27% of its operating profit. This indicates that although technologically advanced, these newer nodes incur higher operational costs.

Goldberg's analysis leverages TSMC's reported revenue by node, combined with depreciation knowledge and estimations to compute profitability. Depreciation, particularly of older, fully depreciated nodes such as the 7nm and above, significantly impacts the profit model, lowering cost due to the absence of depreciation charges. Goldberg highlights that the company depreciates equipment over a five-year cycle, marking this window as critical for maximizing profitability post-depreciation. Therefore, nodes launched since 2017, like 7nm, have completed their depreciation, unlike the 3nm and 5nm nodes still carrying active depreciation costs.

The analysis performed allocates depreciation expenses primarily to advanced nodes based on their revenue share, which provides a clearer illustration of cost structures. The R&D and SG&A (Selling, General, and Administrative Expenses) expenses were also allocated, with assumptions made that the majority of R&D is directed towards currently active and developing nodes, specifically 3nm and 5nm. This allocation might overestimate costs due to these expenses not being evenly distributed across nodes, as older nodes likely require less investment and oversight.

The profitability advantage of older nodes is driven by a lack of depreciation and lower associated costs, making them significantly valued despite lower technological advancement. As Goldberg describes, "the big driver in this model is depreciation," underscoring the model's sensitivity to asset depreciation schedules. While newer nodes are strategically positioned for future profitability gains, their higher initial costs reduce current profit margins, contrasting the unexpectedly efficiently profitable older nodes.

Overall, TSMC's approach of maintaining older node operations demonstrates a revenue strategy that capitalizes on the fully depreciated technology, yielding higher profit margins. It is crucial that as the newer technology advances, like the 3nm node, the company gradually reduces these costs and aligns its profitability closer to its impressive revenue figures. As TSMC progresses with innovation, achieving a balance between maintaining old node profitability and developing new node capabilities remains a challenge and opportunity for further growth.

Sources: TechSpot, TSMC Financial Reports

hardwareinnovationGrowth