TSM’s Expansion Gamble and the Next Phase of Foundry Dominance

Taiwan Semiconductor Manufacturing Company (NYSE: TSM), is making a huge bet that its unprecedented expansion in advanced nodes and overseas fabs will secure its dominance in the next era of chipmaking.

The numbers are staggering: capex above $40 billion, capacity plans stretching across continents, and customer commitments spanning AI, smartphones, PCs, and hyperscaler applications-specific integrated circuits (ASICs). For investors, the question is not whether TSM will grow—but how the costs, risks, and external pressures reshape the economics of this growth. That in turn is a story about capital intensity, geopolitical insurance, and the strategic pivot to advanced nodes that now account for the lion’s share of TSM’s revenues.

The Shift to Advanced Nodes

TSM’s revenue composition has dramatically transformed in the last seven years. Advanced nodes—defined as 7nm and below—once accounted for less than 10% of revenues, but now represent nearly 70% of the company’s topline. The progression from 7nm to 5nm and now to 3nm has been relentless, with 2nm on track for volume production in the second half of 2025.

As Table 1 shows, advanced node revenues have expanded at an average growth rate above 60% annually, while legacy nodes have stagnated or declined. This divergence explains why TSM’s capital allocation is overwhelmingly focused on Extreme Ultraviolet (EUV)-intensive processes, which not only command higher Advanced Semiconductor Packaging & Chiplet Show (ASPs) but also reinforce customer dependence.

Chart 1 highlights TSM’s decisive pivot toward advanced nodes. In 2018, the advanced nodes represented less than 10% of the topline. But by 2025, TSM is expected to reach nearly three-quarters of all revenues. The steady decline of legacy processes underscores the company’s willingness to sunset older nodes in favor of high-margin EUV-based production. This transformation illustrates the growing dependence of global customers on TSM for cutting-edge technologies and the capital intensity required to maintain such leadership.

CapEx and Capacity Expansion

The pivot to advanced nodes, however, has required massive capital investment. TSM’s 2025 capex target of $38-42 billion underscores its willingness to sacrifice near-term margins for long-term technological dominance. Capex per wafer has climbed steadily since 2017, driven by the proliferation of EUV systems costing $150 million apiece, far above older Deep Ultraviolet (DUV) lithography tools. The next generation High-NA EUV machines run more than double that price.

According to Table 2, advanced node wafer capacity is projected to rise nearly 45% in 2025, as 3nm ramps and 2nm investments accelerate. By 2027, advanced nodes are expected to account for more than one-quarter of TSM’s installed base, up from under 10% just three years ago.

Chart 2, showing capacity at advanced nodes vs. total, demonstrates how aggressively the company is ramping its advanced nodes. While overall installed capacity rises steadily, the growth of ≤7nm far outpaces the total, reflecting both customer demand and strategic intent. By 2027, advanced nodes will account for nearly 30% of the company’s output—a massive increase from less than 10% just five years prior. The chart shows clearly why capex intensity is rising and why TSM remains the primary enabler of Moore’s Law scaling.

U.S. Expansion and Tariff Hedge

The company’s Arizona investment, branded Fab 21, reflects both customer demand and geopolitical calculus. The three-phase expansion totals $65 billion and progresses from 4nm to 3nm and eventually sub-2nm production by 2028.

According to Table 3, phase one began with 20k wafers per month of 4nm production this year, while phase two will scale 3nm at higher cost and complexity. The third phase, scheduled for 2028, positions the United States to host a meaningful share of global 2nm output, insulating TSM against tariff or supply-chain disruptions.

Competitive Position in Advanced Nodes

While TSM expands, rivals stumble. Samsung continues to face yield issues reportedly under 50%, with its Texas fab delayed into 2026 and its 1.4nm roadmap slipping toward 2029. Intel has pushed its 18A into 2026 and is hesitating on 14A commitments pending customer traction. Meanwhile, SMIC has demonstrated technical ingenuity by advancing 7nm and 5nm nodes via DUV multipatterning, but commercial scale remains unproven.

This contrast is illustrated in Table 4. Execution and yields—not just process roadmaps—remain the defining differentiators, and here TSM’s advantage is decisive.

Customer Pipeline and Market Share

TSM’s moat is not only technological—it is commercial. Major U.S. customers such as Apple (NASDAQ: AAPL), Nvidia (NASDAQ: NVDA), AMD (NASDAQ: AMD), and Intel (NASDAQ: INTC), alongside hyperscalers like Google (NASDAQ: GOOG), Microsoft (NASDAQ: MSFT), and Amazon (NASDAQ: AMZN), have locked in roadmaps tied directly to TSM’s advanced nodes.

According to Table 5, TSM’s market share has rebounded to above 60% as customers defect from Samsung. The company’s strong pipeline includes GPUs, CPUs, custom AI ASICs, and smartphone SoCs, while Samsung’s most notable recent win is Tesla’s Dojo accelerator—a far thinner roster.

Chart 3 shows TSM has widened its lead over Samsung, cementing its dominance in foundry market share. While Samsung’s position has eroded due to yield challenges, SMIC has made incremental gains by bypassing EUV restrictions through DUV multipatterning. Intel’s share has remained stagnant, reflecting its execution delays. The chart makes clear that TSM’s leadership is both resilient and expanding despite global headwinds.

Tariff Exposure and Risk

The final risk vector is tariffs. With Trump’s administration threatening duties up to 100% on imported semiconductors, exemptions for companies expanding U.S. capacity are crucial. TSM’s Arizona fabs thus serve a dual function: proximity to U.S. customers and protection against policy shocks.

According to Table 6, automotive chips—the most exposed to tariffs due to global car production—account for only 5% of TSM’s revenues. The majority of revenues are in HPC (high performance computing) and smartphones, categories that are largely tariff-exempt. The estimated direct tariff impact is therefore under 0.1% of revenues—immaterial by any investor standard.

Investor Takeaway

TSM’s expansion plans underscore the defining reality of the semiconductor industry: the road to dominance runs through advanced nodes, and the road to resilience runs through geographic diversification.

The company’s capex intensity signals near-term margin volatility, but the long-term reward is technological and commercial entrenchment. Competitors remain far behind in execution, yields, and customer wins. The U.S. fabs add both strategic resilience and tariff protection, while the customer pipeline—spanning Apple, Nvidia, AMD, Intel, and hyperscalers—ensures demand visibility.

For investors, TSM is no longer simply a foundry stock. It is the geopolitical linchpin and technological cornerstone of the AI and computing era. The risk lies in capital intensity and macro uncertainty, but the upside is the preservation of near-monopoly status at the leading edge of semiconductor manufacturing.