TSMC Q2 Earnings Call Preview: AI Chip Demand, CoWoS Capacity, and 2nm Progress

The most important issue in TSMC’s Q2 earnings call is not a single quarter’s revenue, but whether AI demand is still pushing up advanced process and advanced packaging capacity. You need to watch Q2 guidance delivery, HPC revenue share, CoWoS expansion, 2nm production ramp, N3 capacity planning, and gross margin pressure together. For investors following TSM ADR, Taiwan-listed 2330, Nvidia’s supply chain, ASML, HBM, and AI servers, this earnings call is more like a stress test for the semiconductor upcycle.

Key Takeaways

• The first focus is whether Q2 revenue, gross margin, and operating margin land near the high end of guidance.
• AI demand should be judged through HPC share, advanced-node utilization, and long-term customer signals.
• CoWoS is a key bottleneck between AI chip wafer production and final delivery.
• 2nm production will shape TSMC’s next round of process pricing power.
• High capital expenditure must be assessed alongside future revenue, depreciation, and overseas fab dilution.
• Post-call stock reaction will depend on guidance and management’s tone versus expectations.

Which Metrics Should You Watch First in TSMC’s Q2 Earnings Call?

You should first look at whether TSMC’s Q2 results come close to the high end of its previous guidance, then examine gross margin, operating margin, advanced-node revenue mix, and next-quarter outlook. Revenue growth alone is not enough, because the market has already priced in AI chip demand, high N3 utilization, and CoWoS expansion. What truly defines the quality of the earnings call is whether revenue growth, profitability, capacity, and order visibility all confirm the same trend.

TSMC has listed its Q2 earnings call on July 16, 2026 among its upcoming financial events. The most important pre-call benchmark comes from its Q1 disclosures: TSMC reported Q1 revenue of USD 35.90 billion, gross margin of 66.2%, and operating margin of 58.1%. Its Q2 2026 guidance called for revenue of USD 39.0–40.2 billion, gross margin of 65.5%–67.5%, and operating margin of 56.5%–58.5%.

Monthly revenue offers an early read on Q2 momentum. TSMC’s 2026 monthly revenue shows April revenue of NT$410.726 billion and May revenue of NT$416.975 billion. Its May revenue release also showed January–May revenue of NT$1.9618 trillion, up 30.0% year over year. This suggests that the first two months of Q2 remained on a high revenue base, but full-quarter results and management guidance are still needed to judge sustainability.

Summary: The first layer of TSMC’s Q2 earnings call is not whether revenue increased, but whether revenue, margins, advanced-node mix, and next-quarter guidance send a consistent message. If Q2 results come near the high end of guidance and management continues to confirm demand for AI, HPC, N3, CoWoS, and 2nm, the market is more likely to interpret the call as evidence that the upstream semiconductor cycle remains resilient. If revenue meets expectations but gross margin comes under pressure, or if next-quarter guidance is cautious, the stock reaction may be weaker than the headline numbers suggest.

Is AI Chip Demand Still TSMC’s Main Growth Driver?

AI remains the core issue in TSMC’s Q2 earnings call because TSMC’s growth has become highly dependent on HPC, advanced nodes, and AI accelerator-related customers. You should not focus only on whether a single customer’s orders increased. Instead, you should watch HPC revenue share, cloud service provider capital expenditure, AI inference demand, HBM base dies, advanced packaging, and N3/N5/N2 capacity together.

TSMC’s Q1 disclosures showed that HPC platform revenue accounted for 61% of total revenue and rose 20% quarter over quarter, while smartphone revenue accounted for 26%. This structure is important because HPC is not limited to GPUs. It also includes AI accelerators, custom ASICs, cloud computing chips, parts of the HBM base die supply chain, and high-performance networking chips. In other words, TSMC’s AI exposure is not a single-customer story. It is driven by multiple cloud providers, chip designers, and system platforms.

In its Q1 call, management emphasized that the AI megatrend continues to be supported by rising token consumption, agentic AI, and demand signals from cloud service providers. For investors, the implication is clear: if AI usage expands from training to inference, agentic workflows, and enterprise deployment, chip demand will not be limited to one-time server purchases. It can become a sustained source of compute consumption.

You can track five major AI demand signals:

• Whether HPC revenue share remains high or continues to rise.
• Whether management continues to highlight strong cloud service provider demand signals.
• Whether N3, N5, and N2 capacity remains tight because of AI and HPC demand.
• Whether CoWoS, HBM base dies, and advanced packaging remain supply-constrained.
• Whether full-year revenue growth expectations remain above a normal semiconductor cycle level.

If you also follow Nvidia, Broadcom, AMD, Marvell, ASML, or the HBM supply chain, TSMC’s earnings call can serve as an upstream validation tool. Orders from chip designers eventually flow into wafer manufacturing, packaging, and testing. If TSMC’s advanced-node utilization remains high, it suggests that real delivery demand across the AI supply chain has not cooled significantly.

Summary: To judge whether AI remains TSMC’s main growth driver, do not rely on isolated news about a single popular AI chip. A more reliable approach is to look at HPC share, advanced-node revenue mix, cloud service provider demand signals, CoWoS supply-demand conditions, and full-year revenue outlook. As long as these indicators remain aligned, TSMC’s AI story is not merely a short-term theme. It is a structural growth driver built on wafer foundry leadership, advanced packaging, and high-performance computing.

Why CoWoS Capacity Is a Key Bottleneck for AI Chip Delivery

CoWoS is a variable you must listen for in TSMC’s Q2 earnings call because AI chips do not become deliverable products just because advanced-node wafers are manufactured. GPUs, AI ASICs, and high-performance accelerators often need compute dies, HBM, interposers, substrates, and system-level packaging to be integrated together. Tight wafer capacity shows strength in front-end manufacturing, while tight CoWoS capacity shows that final AI compute modules still face delivery bottlenecks.

At its 2026 North America Technology Symposium, TSMC disclosed that to support AI demand for more compute and memory inside a single package, CoWoS technology continues to expand. The company has already moved 5.5-reticle-size CoWoS into production planning and is preparing larger versions. Its 14-reticle-size CoWoS is expected to enter production in 2028 and can integrate around 10 large compute dies and 20 HBM stacks. This shows that the value of AI chips does not come only from smaller process nodes. It also comes from larger package areas, higher memory bandwidth, and more complex system integration.

You need to distinguish three types of bottlenecks. The first is wafer manufacturing, such as EUV, N3, N5, and N2 capacity. The second is advanced packaging, including CoWoS, SoIC, and CoPoS. The third covers supporting links such as HBM, substrates, testing, and system integration. In its Q1 Q&A, TSMC said its current main supply approach remains large-sized CoWoS, while CoPoS is also being developed. For investors, this means that the bottleneck in AI chip delivery may shift from wafers to packaging.

Summary: CoWoS matters because it connects TSMC’s front-end manufacturing with final AI server delivery. Advanced process nodes determine chip computing power, while advanced packaging determines whether that computing power can be combined with HBM into high-value deliverable products. If the Q2 earnings call continues to confirm strong CoWoS demand, smooth expansion, and customer willingness to pay for larger packaging, the market is more likely to believe that TSMC’s AI revenue comes not only from wafer pricing, but also from increasing system-level manufacturing complexity.

How 2nm, N2P, and A16 Progress Could Shape Long-Term Growth

TSMC’s 2nm progress will determine whether the company can extend AI and high-end smartphone demand into the next process cycle. You need to watch N2 production, N2P, A16’s Super Power Rail, customer tape-outs, yield, and early-stage gross margin dilution. In the short term, 2nm will bring depreciation and ramp-up costs. In the long term, it will determine whether TSMC can maintain pricing power in advanced processes.

In its Q1 earnings call, TSMC management said N2 entered high-volume production in the fourth quarter of 2025 and is ramping in phases in Hsinchu and Kaohsiung, with demand coming from smartphones and HPC/AI. This matters because 2nm is not only a smartphone SoC node. It will also gradually move into high-performance computing and AI chip designs.

A16 represents a longer-term technology roadmap. In its description of A16 technology, TSMC highlights nanosheet transistors and Super Power Rail, targeting improvements in logic density and performance. You can think of it this way: as AI chips require higher power density, more complex routing, and stronger energy efficiency, transistor scaling alone is not enough. The power delivery structure also needs to improve.

In the earnings call, you should listen for six 2nm-related questions:

• Whether N2 yield continues to improve according to plan.
• Whether the N2 capacity ramp can serve both smartphone and HPC/AI customers.
• Whether the N2P and A16 timelines remain unchanged.
• Whether early 2nm gross margin dilution remains controllable.
• Whether additional N3 capacity affects the pace of N2 investment.
• Whether customers are already locking in next-generation advanced-node capacity.

It is important to note that 2nm may not immediately improve margins. TSMC has already stated that the initial 2nm ramp will begin to dilute gross margin in the second half of 2026, with a full-year impact of around 2–3 percentage points. You should assess this alongside the long-term revenue opportunity. If N2 demand is strong, customer quality is high, and yield improves quickly, short-term dilution may be the cost of securing the next node cycle. If the ramp is slow or demand falls short, capital expenditure pressure will become more visible.

Summary: 2nm is a core variable for TSMC’s long-term growth, not just a technology headline. N2 production marks TSMC’s entry into the nanosheet node, while N2P and A16 will shape the power efficiency and routing space of next-generation AI/HPC chips. In the Q2 earnings call, you should look at technology progress, customer adoption, yield, capacity, and gross margin impact together. Only when these variables improve together can 2nm translate from “technology leadership” into “financial leadership.”

How Gross Margin, Capital Expenditure, and Overseas Expansion Change the Investment View

The quality of TSMC’s Q2 earnings will depend on whether high utilization and cost improvement can offset early 2nm ramp-up, overseas fab dilution, material costs, and currency movements. High capital expenditure is neither automatically bullish nor bearish. It shows that the company is investing for future demand, but it becomes stronger shareholder value only if future revenue growth can outrun depreciation, construction costs, and overseas operating costs.

In its Q1 earnings call, TSMC said its 2026 capital budget is expected to be near the high end of the USD 52–56 billion range. The 2026 capital budget is tied to AI, HPC, and customer growth opportunities. This figure shows that TSMC has strong confidence in future demand, but it also means depreciation pressure will gradually move into the income statement.

Overseas fabs are another important variable in valuation discussions. Management has said that early overseas fab ramps create around 2–3 percentage points of gross margin dilution, which could later expand to 3–4 percentage points. You should not understand expansion in the United States, Japan, and Germany only as a supply-chain security move. You also need to consider its effects on cost structure, capacity utilization, customer pricing, and free cash flow.

For ordinary investors, margins reveal more about the quality of TSMC’s cycle than revenue does. If revenue growth comes from high-margin advanced nodes and advanced packaging rather than low-price volume growth, TSMC’s valuation base is more stable. If revenue growth requires higher capital expenditure, lower gross margin, or a longer payback cycle, the market may reassess its valuation premium.

Summary: High capital expenditure should be judged together with AI demand, advanced-node capacity, overseas fab dilution, and free cash flow. TSMC’s current strengths are technology leadership, strong customer structure, and scarce capacity. Its risks are new-node ramp-up, overseas fab costs, and material price volatility. If the Q2 earnings call can show that high CapEx corresponds to more visible AI/HPC demand rather than passive expansion, the market may become more tolerant of margin pressure.

How Ordinary Investors Should Read TSMC’s Post-Call Stock Reaction

TSMC’s stock reaction after the earnings call may not move in the same direction as the headline numbers. What you really need to compare is the gap between actual results and market expectations: whether Q2 revenue beats the midpoint of guidance, whether next-quarter guidance remains strong, whether AI demand commentary becomes more positive, whether CoWoS capacity improves, and whether 2nm dilution remains manageable. Stock reactions often come from expectation gaps, not simple year-over-year growth.

You can build three scenarios. In an optimistic scenario, Q2 results come near the high end of guidance, management raises confidence in full-year revenue, CoWoS and N3 demand remains tight, and the N2 ramp goes smoothly. In a neutral scenario, the numbers meet guidance, but management remains cautious on consumer-end demand, material costs, or overseas fab dilution. In a pressure scenario, gross margin falls below expectations, next-quarter guidance weakens, or advanced packaging expansion cannot meet AI customer demand.

In the 24–72 hours after the earnings call, you can review seven questions:

• Did Q2 actual revenue exceed the midpoint of company guidance?
• Did gross margin hold within the market’s expected range?
• Did next-quarter guidance suggest continued AI demand expansion?
• Did management continue to highlight demand from CSP and AI customers?
• Did CoWoS expansion receive a clearer timeline?
• Did the N2/N2P/A16 roadmap remain stable?
• Does CapEx still correspond to future revenue growth?

If you follow TSM ADR, Taiwan-listed 2330, or semiconductor ETFs, you also need to consider trading costs and order execution. During popular earnings windows, price swings can be large, and bid-ask spreads, platform fees, external agency fees, FX conversion, and order types can all affect actual returns. When using Biya to follow U.S. and Hong Kong market assets, you can first use U.S. stock information search to review basic stock information, then check U.S. stock trading fees to understand commissions, platform fees, and external agency fees. Biya charges USD 0 commission for U.S. stock trades, while platform fees, external agency fees, and other charges are subject to the fee center and the order page.

Summary: Post-call trading decisions around TSMC should move beyond “were the earnings good?” and toward “did the results exceed what the market had already priced in?” If revenue, margins, AI demand, CoWoS, and N2 all improve together, the stock is more likely to find support. If revenue is strong but margins or next-quarter guidance are weak, the market may focus more on cost and cycle risk. Earnings trading is not the same as long-term investing. Before trading, you should fully understand order types, fee structure, volatility risk, and your own risk tolerance.

If you treat TSMC’s earnings call as a window into the AI semiconductor supply chain, you can track it alongside Nvidia, ASML, memory chips, HBM, and cloud service provider capital expenditure. Biya supports multi-asset trading across U.S. stocks, Hong Kong stocks, and cryptocurrencies, and can be used to monitor TSM ADR, semiconductor ETFs, and related Hong Kong-listed technology assets. Before using any service, you should confirm whether it is available based on your region, identity verification result, platform rules, and applicable laws and regulations. You should also review the fee details shown on the order page before placing a trade. If you want to track market quotes and earnings windows on mobile, you can use Biya App to manage watchlists and follow relevant market information, while making independent decisions based on public information rather than treating any single earnings call as a trading guarantee.

FAQ

What should ordinary investors focus on in TSMC’s Q2 earnings call?

Ordinary investors should focus on Q2 results, next-quarter guidance, AI/HPC demand, CoWoS capacity, and 2nm progress. Revenue growth is only the first layer. Gross margin, advanced-node utilization, and management tone are more important for market reaction. If these variables are all stronger than expected, the quality of the earnings call is higher.

Why does tight TSMC CoWoS capacity affect AI chip delivery?

Tight CoWoS capacity affects AI chip delivery because high-end AI accelerators usually need advanced packaging to combine compute dies with HBM. Even if wafer capacity is sufficient, constraints in CoWoS, HBM, substrates, or testing can still delay final server chip delivery.

How could TSMC’s 2nm production progress affect long-term valuation?

TSMC’s 2nm production progress affects long-term valuation because it determines customer adoption, pricing power, and profit potential in the next advanced-node cycle. In the short term, 2nm may bring depreciation and yield ramp pressure. Its long-term value depends on whether AI/HPC and high-end smartphone customers adopt the node quickly.

Does high TSMC capital expenditure always support the stock price?

High TSMC capital expenditure does not always support the stock price. It may signal strong future demand, but it can also increase depreciation, cash flow pressure, and gross margin dilution. The key is whether CapEx is backed by visible orders, high utilization, reasonable pricing, and long-term revenue growth.

Will TSM ADR and Taiwan-listed 2330 react the same way to the earnings call?

TSM ADR and Taiwan-listed 2330 are usually directionally related, but their short-term reactions may not be identical. The ADR can also be affected by the U.S. dollar, U.S. tech stock sentiment, overseas risk appetite, and trading hours. Investors should separately check trading prices, FX impact, and local market rules.

How can investors manage risk when trading around TSMC earnings?

Investors can manage risk by controlling position size, understanding order types, confirming fee structures, and avoiding trades based only on a single earnings headline. TSMC is a high-quality semiconductor leader, but it is still affected by valuation, exchange rates, geopolitics, CapEx, and tech stock sentiment. Trading decisions should follow platform rules, billing details, and local regulatory requirements.