TSMC Stock Q3 Earnings Preview: Will Record Sales Be Enough?

TSMC is heading into earnings with a problem most companies would love to have: its sales are so strong that another record may struggle to impress. Monthly disclosures have already revealed a booming third quarter. 

The October 15 results must answer a harder question: how much of that demand survives the cost of new factories and newer chip technology to become lasting earnings and cash?

Let's break down what analysts expect, what could move TSMC stock, and how to judge the results beyond the headline beat.

TSMC Q3 2026 Earnings Date and Analyst Expectations

TSMC reports its July-September quarter on Thursday, October 15, before US markets open. Its earnings conference starts at 2:00 a.m. US Eastern Time, or 11:30 a.m. IST and 2:00 p.m. Taiwan time.

The important distinction is between management's July guidance, current analyst forecasts and revenue already disclosed through monthly reports.

Sources: TSMC Investor Relations, MarketBeat and Reuters. The EPS and net-profit forecasts come from different analyst datasets.

Reuters' profit forecast incorporates 16 analysts. A result near that number would be impressive, but it would broadly meet an already demanding expectation. The more valuable surprise would be stronger operating profitability or an outlook that lifts estimates for future quarters.

TSMC Q3 2026 Revenue: Why Record Sales May Not Surprise Investors

Beyond revenue, TSMC's monthly disclosures allow investors to move further along the chain: the quarter's sales are substantially known, while profitability and guidance remain open questions.

Source: TSMC monthly revenue disclosures for 2026 and 2025; quarterly total and growth calculated from unrounded displayed monthly figures.

That total is approximately 17.6% above Q2. At management's guidance assumption of NT$32 per US dollar, it translates to about $46.70 billion, roughly 2.0% above the top of July's revenue guidance. The company-reported dollar figure will depend on the actual quarterly exchange rate.

The small September decline deserves perspective, too. Sales fell 0.6% from August but remained 54.6% higher than a year earlier. A modest retreat from a very strong month does not, by itself, establish weakening demand.

Investors should give limited weight to an October 15 headline announcing record sales. The sharper question is whether TSMC is earning enough on that larger revenue base, and whether customers need even more capacity next quarter.

TSMC Earnings: The Q2 Profit Adjustment That Changes How Investors Should Read Q3

The comparison with Q2 contains an easily missed complication. TSMC's previous quarter benefited from disposal and valuation gains on its investment in Vanguard International Semiconductor, a separate chipmaker. Those gains were reported below operating income and are distinct from profits earned manufacturing customers' chips.

TSMC disclosed that the gains contributed NT$2.24 to Q2 EPS. Removing that contribution from reported EPS of NT$27.25 gives an illustrative comparison base of NT$25.01. This is our limited adjustment for the disclosed Vanguard contribution, rather than a complete measure of recurring earnings.

Calculations use the LSEG profit forecast and assume Q2's 25.932 billion diluted weighted-average shares remain unchanged in Q3.

This changes the interpretation considerably. A roughly 6% sequential EPS increase could conceal materially stronger growth against a cleaner comparison base. Equally, any new Q3 investment gains should be identified before declaring that chipmaking profits have accelerated.

The same issue affects net margins. Q2's reported net margin was 55.6%; the current profit forecast divided by disclosed Q3 sales implies approximately 49.9%. Comparing those percentages without examining investment gains, tax and operating margins would mix different influences together.

For this release, start with operating profit, then explain the bridge to net profit. That is a more useful test of manufacturing performance than simply celebrating another EPS record.

What Has Changed Since TSMC's Last Earnings Report?

Three developments add useful context to the October call. They concern capacity TSMC may need in future years, rather than immediate additions to Q3 sales.

Sources: Reuters, GlobalFoundries and ASML Investor Relations.

The GlobalFoundries agreement covers silicon interposers, components that help connect chips within advanced packages. Its initial term is five years, and production is expected to begin ramping in the first half of 2028. It supports the longer-term supply chain; it cannot solve tomorrow's capacity constraints.

TSMC had already announced an additional $100 billion Arizona commitment in July, bringing the planned total there to $265 billion. That multi-year commitment must be kept separate from annual capital expenditure. Investors should seek spending schedules, customer commitments and the economics of each expansion.

TSMC Earnings Outlook: AI Demand, 2nm Margins and Q4 Guidance

Whether the 2nm ramp is progressing economically

TSMC's 2nm, or N2, technology is a newer manufacturing generation designed to improve chip performance and power consumption. Yield, the proportion of manufactured chips that work as intended, helps determine how efficiently that technology earns money.

Management warned in July that the steep 2nm ramp would dilute gross margin by about three to four percentage points in the second half. It also expected overseas factories to dilute margins as they expanded. These are existing costs in the investment case, rather than entirely new risks.

Think of a bakery opening a second kitchen. Equipment, staffing and early waste raise the cost of each loaf before production runs smoothly. The investment becomes worthwhile if the kitchen subsequently produces more saleable bread efficiently.

Applied to TSMC, lower margins accompanied by improving yields and committed demand can represent productive investment. Lower margins accompanied by delays, weaker utilisation or persistent cost overruns would be a different story.

Whether AI strength is broad enough to support the larger business

High-performance computing contributed 66% of Q2 revenue, while smartphones contributed 22%. Within wafer revenue, 2nm represented 3% and 3nm represented 30%. Platform revenue and technology revenue are different classifications and should not be added together.

HPC should also not be treated as an exact measure of AI revenue. The earnings call needs to distinguish AI-related demand from the broader computing and consumer cycle. Watch absolute sales and management commentary alongside percentage shares: a category's share can fall even while its revenue rises.

Advanced packaging is another constraint worth watching. CoWoS combines processing chips and high-bandwidth memory so they can work together efficiently. A powerful processor still needs the surrounding connections to deliver useful system performance.

For investors tracking Nvidia, this makes TSMC's commentary on packaging availability especially useful. Broadly manufacturing and packaging demand across competing chip designs provides a stronger investment foundation than dependence on one product launch.

Whether Q4 guidance clears the next hurdle

Reuters reported that Morgan Stanley's Charlie Chan expects full-year revenue growth guidance to move into the low-to-mid 40% range. A higher annual growth headline would be encouraging, but the quarterly numbers underneath it deserve closer attention.

Here is an independent cross-check. The S&P Global annual revenue forecast displayed by StockAnalysis is approximately NT$5.47trillion. Subtracting NT$3.899 trillion of disclosed January-September sales leaves roughly NT$1.571 trillion for Q4, about 5.2% above Q3.

At an assumed NT$32 per dollar, that is approximately $49.1 billion. This is our forecast-implied hurdle, not TSMC guidance or a separate Q4 analyst consensus. Compare the actual Q4 revenue midpoint at management's newly supplied exchange-rate assumption.

A midpoint materially above that comparable benchmark would strengthen the growth case. A lower number would require explanation even if management raises the full-year headline.

Is TSMC's AI Growth Turning Into Enough Free Cash Flow?

Free cash flow is operating cash flow minus capital expenditure. It measures cash remaining after spending on production assets, before distributions and other financing decisions.

TSMC's Q2 results show why this deserves its own examination.

Sources: TSMC Q2 financial statements and management report; cash-flow margins calculated.

Reported profit rose 23.4% sequentially, but free cash flow fell approximately 17.5%. Operating cash generation improved; spending on factories and equipment increased faster. This does not establish a failing business, but it shows that rapid expansion absorbs a meaningful part of the financial reward.

The full-year capital budget is 60 billion-64 billion. An increase would therefore deserve two questions: what additional customer demand requires the spending, and when will the extra capacity begin producing cash?

Our preferred test is whether revenue, operating profitability and cash remaining after investment improve together over several quarters. Tax payments, inventories and customer payment timing can distort a single quarter, so the trend matters more than one isolated cash-flow reading.

There is meaningful financial capacity to fund expansion: TSMC reported NT$2.486 trillion of net cash reserves at Q2 end. That reduces financing pressure; it does not automatically make every proposed factory an attractive investment.

How Should Investors Respond to Different TSMC Earnings Outcomes?

Read the release in three passes: underlying manufacturing performance, the next-quarter outlook, and cash after investment. Then revisit valuation at the price available after the announcement.

These are conditional analytical responses, not predictions of the announcement or immediate stock reaction.

A lower share price becomes interesting when the earnings case remains intact or improves. If forecasts deteriorate faster than the price, apparent cheapness can disappear. Conversely, a higher price can remain reasonable if the release materially improves future earnings visibility.

Author’s View on TSMC Stock Before Earnings

TSMC deserves serious consideration as an AI infrastructure business because its role extends across customers and competing chip designs. Its foundry model manufactures chips designed by other companies, allowing investors to participate in manufacturing demand without choosing a single winning chip brand.

Stance is constructive, with a clear condition: expansion must produce durable operating earnings and stronger cash generation after investment. The current sales evidence supports demand. The cleaner Q2 comparison makes the expected profit progression more encouraging than it first appears. A modelled 19.5 times 2027 earnings can support a credible investment case if those earnings materialise.

We would become more confident with stronger Q4 guidance, manageable 2nm ramp costs and improving cash conversion. Another record EPS figure, on its own, would add less conviction.