CluenexHow much did TSMC make in Q3 in US dollar terms?
TSMC generated NT$1,494.24 billion in consolidated revenue for the third quarter of 2026, which converts to $46.72 billion. This represents a 41.2% increase in US dollar terms compared to the third quarter of 2025, and a 16.2% sequential increase over the previous quarter.
Because TSMC reports monthly sales in New Taiwan dollars (NT$) but bills international customers mostly in US dollars, currency swings can slightly alter growth rates. In local currency, revenue climbed 50.9% year-over-year. Net income reached NT$763.06 billion (roughly $23.85 billion), up 68.7% from a year earlier. Diluted earnings per share came in at NT$29.43, or $4.60 per American Depositary Receipt (ADR) unit.
The top-line figure cleared both Wall Street estimates and TSMC's own guidance. In July 2026, management guided Q3 revenue between $44.6 billion and $45.8 billion. The actual result of $46.72 billion beat the midpoint of that guidance by over $1.5 billion, confirming that customer orders for leading-edge silicon accelerated through August and September.
Which line items prove AI hardware demand is real?
Wafer revenue by technology node and platform segment gives the clearest confirmation of artificial intelligence infrastructure spending. In the third quarter of 2026, advanced process nodes (defined as 7-nanometer and below) generated 79% of TSMC's total wafer revenue.
| Process Node | Share of Q3 Wafer Revenue | Primary Use Case |
|---|---|---|
| 2-nanometer (N2) | 8% | Early cloud AI silicon, test units, next-gen mobile |
| 3-nanometer (N3) | 33% | Flagship mobile processors and leading AI accelerators |
| 5-nanometer (N5) | 28% | Mainstream data center AI GPUs and server CPUs |
| 7-nanometer (N7) | 10% | Networking hardware, automotive, legacy compute |
| Mature nodes (16nm+) | 21% | Power management, microcontrollers, display drivers |
Nodes at 5nm and below, which host modern graphics processing units (GPUs) and specialized AI application-specific integrated circuits (ASICs), generated 69% of all wafer sales. The 3nm node alone contributed one-third of wafer revenue. When mega-cap tech companies invest billions into data centers, their capital spending directly pays for these specific wafer runs. To understand how that capital flows from cloud balance sheets to chip foundries, review our guide to AI capex vs revenue payoff.
What do TSMC's profit margins tell us about its pricing power?
TSMC posted a gross margin of 66.4% and an operating margin of 58.7% for the quarter. These numbers show that TSMC passed on elevated energy, overseas expansion, and research costs without losing customer volume.
By comparison, the median technology benchmark operating margin sits at 15.9%, while the sector benchmark gross margin is 61.5%. TSMC operates well above these broad peer benchmarks because no rival foundry can deliver leading nodes at comparable yields and scale.
Net profit margin reached 51.1%. In fundamental analysis, when a capital-intensive manufacturer converts more than half of its gross sales into pure profit, it indicates an exceptional competitive advantage. For a detailed breakdown of how to interpret these profitability levels, see our guide on gross, operating, and net margins.
Why are investors talking about TSM losses on social media?
Online community discussions regarding losses refer to retail stock trading losses, not corporate net losses at the company. TSMC set record corporate profits in Q3 2026, earning NT$763.06 billion.
Retail traders often experience losses around earnings reports due to three common dynamics:
- Pricing in perfection: The stock rallied strongly into the earnings date. Even when results beat guidance, short-term options traders take profits, creating an immediate post-earnings dip.
- Foreign exchange shifts: TSMC's ADR shares trade in US dollars, while its headquarters and capital structure operate in Taiwan dollars. Currency volatility can cause the ADR to trade differently than the underlying Taipei-listed shares (TWSE: 2330).
- Geopolitical risk discounting: Institutional funds consistently price a geopolitical discount into TSMC due to cross-strait tensions, limiting how high its price-to-earnings multiple expands relative to US-domiciled chip designers.
What is the true bottleneck for AI hardware: fabrication or packaging?
Raw silicon wafer fabrication is no longer the main constraint on AI chip delivery: advanced packaging is. Producing high-end AI processors requires connecting logic dies to High Bandwidth Memory (HBM) using packaging technologies like Chip-on-Wafer-on-Substrate (CoWoS).
Even as TSMC's 3nm and 5nm lines run at full capacity, customers like Nvidia, AMD, and Broadcom remain constrained by CoWoS allocation. Building new advanced packaging lines takes 12 to 18 months, requiring specialized cleanrooms and tools. While wafer revenue demonstrates massive appetite, packaging throughput determines the actual volume of finished server racks delivered to cloud data centers. You can read more about industry cycles and inventory metrics in our tutorial on how to analyze semiconductor stocks.
What are the main risks to this growth trajectory?
Despite record numbers, investors must watch two distinct structural risks that could disrupt TSMC's financial momentum.
First, customer concentration remains elevated. A handful of US technology giants account for the vast majority of advanced node purchases. If any major cloud provider slows data center infrastructure spending, TSMC's capacity utilization on 3nm and 2nm lines could soften quickly.
Second, international fab expansion in Arizona, Japan, and Germany carries higher operating costs than Taiwanese operations. While these sites provide geographic resilience, lower initial manufacturing yields and higher labor expenses can dilute overall gross margins as they scale. Investors can review TSMC's valuation, insider activity, and business quality score on Cluenex's TSM research page.
What this means for you
If you manage your own portfolio and own individual semiconductor stocks or tech index funds, TSMC's Q3 release offers three practical takeaways:
- Separate trading sentiment from business fundamentals: A short-term drop in share price after an earnings beat is normal market behavior. It does not mean the underlying business is distressed or losing money.
- Track the capital expenditure chain: TSMC's results confirm that hyperscale cloud companies are still spending heavily on hardware. Watch the upcoming earnings calls of Microsoft, Alphabet, Amazon, and Meta to verify whether their AI capital expenditures continue to align with TSMC's shipment growth.
- Keep node diversification in mind: When evaluating any chip company, check what percentage of its products depend on TSMC's sub-7nm capacity. Companies with secured advanced packaging allocations are better positioned to fulfill customer orders than those stuck in foundry backlogs.