Taiwan Semiconductor Manufacturing Company (NYSE: TSM), better known as TSMC, is the world’s leading dedicated semiconductor foundry. Rather than designing consumer chips under its own brand, TSMC manufactures advanced semiconductors for customers across artificial intelligence, high-performance computing, smartphones, automotive electronics, and connected devices.
This position gives TSMC exceptional exposure to rising computing demand, but it also makes the company unusually capital intensive and strategically important. Investors must weigh technology leadership and pricing power against enormous fabrication costs, customer concentration, execution risk, overseas-fab dilution, and the geopolitical risk surrounding Taiwan.
Last reviewed: September 11, 2026. Financial figures are based primarily on TSMC’s 2025 annual report and Form 20-F and its second-quarter 2026 results.
TSMC at a glance
| NYSE ticker | TSM |
| Business model | Dedicated semiconductor foundry |
| 2025 revenue | NT$3.81 trillion |
| 2025 net income attributable to parent | NT$1.72 trillion under Taiwan-IFRS |
| 2025 free cash flow | NT$1.00 trillion |
| 2Q 2026 revenue | US$40.20 billion |
| 2Q 2026 gross margin | 67.7% |
| 2Q 2026 HPC share of revenue | 66% |
| 2Q 2026 advanced-node share of wafer revenue | 77% |
| 2026 capital-budget guidance | US$60–64 billion |
Why TSMC’s foundry model matters
TSMC manufactures chip designs supplied by customers and generally avoids competing with them in branded semiconductor products. This dedicated-foundry model has helped the company build trust, concentrate research and capital on manufacturing, and spread the cost of leading-edge fabs across many customers and applications.
The scale advantage is difficult to replicate. In 2025, TSMC manufactured 12,682 products using 305 process technologies for 534 customers. Its managed annual capacity exceeded 17 million 12-inch-equivalent wafers. High utilization generates cash that can fund the next process generation; the next generation attracts demanding customers; and customer volume improves learning and yield. This feedback loop is one of the company’s strongest competitive advantages.
Revenue growth and extraordinary margins
TSMC’s 2025 revenue increased 31.6% to NT$3.81 trillion. Gross profit rose 40% to NT$2.28 trillion, implying a gross margin near 59.9%. Taiwan-IFRS net income attributable to shareholders of the parent increased 46.4% to NT$1.72 trillion. The company’s Form 20-F reports slightly different IFRS figures because of the timing of Taiwan tax on unappropriated earnings; investors should avoid mixing the two accounting bases.
Momentum accelerated in 2Q 2026. Revenue was US$40.20 billion, up 33.7% in U.S.-dollar terms and 36.0% in New Taiwan dollars from a year earlier. Gross margin reached 67.7%, operating margin was 60.3%, and net profit margin was 55.6%. Reported net income attributable to the parent was NT$706.56 billion, up 77.4% year over year.
The quarter also included NT$63.2 billion of disposal and mark-to-market gains on shares of Vanguard International Semiconductor. That benefit sits outside core operations, so the 60.3% operating margin is a cleaner indicator of manufacturing economics than the exceptionally high net margin alone.
Advanced nodes are the economic engine
Leading-edge manufacturing is increasingly important to TSMC’s mix. In 2Q 2026, 2-nanometer technology represented 3% of wafer revenue, 3nm represented 30%, 5nm represented 33%, and 7nm represented 11%. Together, technologies at 7nm and below accounted for 77% of wafer revenue.
The first revenue contribution from N2 is strategically important. New nodes initially carry start-up costs, low utilization, and yield-learning expenses. TSMC said the steep N2 ramp could dilute gross margin by roughly three to four percentage points in the second half of 2026. Investors should not automatically interpret that temporary dilution as weakening competitive position. Successful ramps usually require accepting lower near-term margins before volume and cost improvements mature.
The key indicators are customer adoption, yield improvement, utilization, and the pace at which N2 becomes a meaningful revenue contributor. Inventory days rose to 87 in 2Q 2026 from 80 in the prior quarter, primarily because of the N2 ramp. That is understandable during a major transition, but it deserves continued monitoring.
AI and HPC now dominate the growth story
High-performance computing contributed 66% of 2Q 2026 revenue, up from 60% a year earlier, while smartphones represented 22%. HPC revenue grew 20% sequentially during the quarter. This category includes more than AI accelerators, but its rising share demonstrates how decisively the company’s center of gravity has shifted toward data-center and advanced-computing workloads.
TSMC benefits from AI demand in several ways: leading-edge logic wafers, specialized variants, and advanced packaging that combines compute dies and high-bandwidth memory. The opportunity is structural if computing requirements continue to rise. The risk is that customers and investors may extrapolate unusually rapid infrastructure spending too far. Capacity built for aggressive forecasts can pressure utilization and returns if end demand slows.
Customer geography also shows concentration. North America represented 78% of revenue in 2Q 2026, although this measures customer headquarters rather than final product destination. A small number of large chip designers can materially influence TSMC’s growth and capital plan. Strong customer relationships are an advantage, but dependence reduces bargaining flexibility if purchasing plans change.
Capital expenditure is both moat and risk
TSMC spent US$40.9 billion on capital expenditures in 2025, compared with US$29.8 billion in 2024. In the first half of 2026, capital expenditure totaled US$26.8 billion, including US$15.7 billion in the second quarter. Management raised its full-year 2026 capital-budget guidance to US$60–64 billion, citing stronger customer demand and equipment-cost inflation.
This spending is necessary because leading-edge tools and facilities become more expensive with each generation, and advanced packaging capacity must grow alongside wafer output. Large capital requirements discourage new entrants and reinforce TSMC’s scale advantage. They also raise the penalty for forecasting errors: a fab costs money long before it reaches efficient volume production.
Investors should compare capital spending with future revenue growth, gross margin, utilization, and incremental free cash flow—not simply reward a higher budget. Management views elevated spending as evidence of future opportunity, but the eventual return on that investment remains the essential test.
Free cash flow and capital allocation
Despite its capital intensity, TSMC generated NT$1.00 trillion of free cash flow in 2025, up from NT$870.17 billion in 2024, because operating cash-flow growth exceeded the increase in capital expenditure. In 2Q 2026, operating cash flow was NT$783.36 billion and capital expenditure was NT$496.00 billion, leaving NT$287.36 billion of free cash flow.
The balance sheet provides additional flexibility. At June 2026, cash and marketable securities were NT$3.52 trillion versus NT$1.03 trillion of interest-bearing debt, producing net cash reserves of NT$2.49 trillion. TSMC paid NT$467 billion in cash dividends during 2025 and has stated its commitment to a sustainable, steadily increasing dividend per share.
Free cash flow will be volatile while investment accelerates. A temporary decline can be rational if new capacity earns attractive returns. More concerning would be sustained capital growth accompanied by slowing revenue, weak utilization, deteriorating margins, or rising net debt.
Overseas fabs reduce concentration but cost more
TSMC is expanding beyond Taiwan to provide customers with geographic flexibility and respond to government incentives. Arizona’s first fab entered volume production using 4nm technology in late 2024. The second fab is being prepared for 3nm and more advanced processes, and construction of a third began in 2025. The company also plans further U.S. fabs, advanced-packaging facilities, and an R&D center.
In Japan, the first Kumamoto fab began volume production at the end of 2024, and a second fab is under construction. TSMC is also building a specialty-technology fab in Dresden, Germany, for automotive and industrial applications. These investments improve supply resilience but operate in higher-cost environments. Management expects overseas-fab ramps to dilute gross margin by two to three percentage points in early stages and three to four points later as their scale increases.
How to think about TSM’s valuation
TSM should be valued on normalized earnings and free cash flow rather than one exceptional quarter. The ADR represents five ordinary shares, so investors must use the correct share ratio and a consistent exchange rate when converting New Taiwan dollar results into U.S. dollars.
A practical framework is to estimate revenue growth by platform and node, apply a through-cycle gross margin that allows for N2 start-up costs and overseas-fab dilution, subtract sustainable operating expenses and taxes, and then compare the resulting earnings with the current ADR price. A second check divides normalized free cash flow—after the capital spending required to maintain leadership—by market capitalization.
A premium multiple may be justified by technological leadership, high margins, customer trust, a net-cash balance sheet, and AI-related growth. That premium should be moderated for cyclicality, capital intensity, customer concentration, and Taiwan-related geopolitical risk. Investors should test slower AI spending, lower utilization, several points of margin pressure, and a higher required return rather than rely on a single optimistic forecast.
Key risks for TSM shareholders
- Geopolitical risk: Most leading-edge capacity remains in Taiwan, exposing operations and valuation to cross-strait tension, trade controls, and supply disruption.
- AI-cycle risk: Customers may slow accelerator and data-center investment after a period of rapid expansion.
- Execution risk: Delays or weak yields at N2, future nodes, or advanced packaging could reduce returns and customer confidence.
- Capital intensity: The US$60–64 billion 2026 budget increases the cost of demand-forecasting errors.
- Overseas costs: Arizona, Japan, and Germany improve resilience but are expected to dilute margins during ramp-up.
- Customer concentration: Changes in orders from a small number of major customers can materially affect utilization.
- Technology and competition: Alternative architectures, manufacturing breakthroughs, or aggressive competitor investment could narrow TSMC’s advantage.
- Currency risk: Revenue, costs, capital spending, financial reporting, and ADR returns span multiple currencies.
What investors should monitor
- HPC growth and the durability of AI-related customer demand.
- N2 revenue, yield progress, and inventory during the ramp.
- Gross margin after overseas-fab and new-node dilution.
- Capital expenditure relative to revenue and future free cash flow.
- Advanced-packaging capacity and bottlenecks.
- Customer and geographic concentration.
- The timing, utilization, and economics of overseas facilities.
Bottom line
TSMC entered the second half of 2026 with exceptional operating momentum. AI and HPC demand, advanced-node leadership, rising N2 adoption, and powerful operating leverage drove rapid revenue growth and a 67.7% gross margin. Its scale, manufacturing expertise, customer trust, and net-cash position are formidable strengths.
The same investment case contains unusual risks. TSMC must deploy tens of billions of dollars before demand is fully known, manage several international fab ramps, preserve technology leadership, and operate within a difficult geopolitical environment. A disciplined investor should focus on returns on new capacity and normalized free cash flow—not AI enthusiasm alone.
For a reusable research process, see our guide to analyzing a stock. Investors studying the broader AI supply chain may also compare TSMC with our NVIDIA (NVDA) analysis.
Frequently asked questions
Why is TSMC important to artificial intelligence?
TSMC manufactures many leading-edge processors used in AI and high-performance computing and provides advanced packaging capabilities needed to combine compute and memory components.
Why can a new process node reduce TSMC’s margin?
A new node begins with start-up costs, lower utilization, and yield-learning expenses. Margins can improve as production volume, yield, and manufacturing efficiency rise.
What is the biggest risk to TSM stock?
There is no single risk, but Taiwan-related geopolitical exposure is unusually important because much of TSMC’s leading-edge production remains concentrated there. Capital intensity, customer concentration, and semiconductor cyclicality also matter.
Primary sources
- TSMC 2025 Form 20-F — SEC
- TSMC 2025 annual report
- TSMC 2Q 2026 earnings release
- TSMC 2Q 2026 management report
- TSMC 2Q 2026 results page
Disclosure: This article is for educational and informational purposes only and does not constitute investment, tax, or legal advice. Currency conversion, ADR ratios, accounting bases, and company-defined measures should be checked against current filings before making an investment decision.