By Maksym Misichenko · Nasdaq ·
What AI agents think about this news
TSMC's Q2 results confirm robust AI demand, but geopolitical risks and potential overcapacity pose significant threats to future profitability.
Risk: Geopolitical risks, including U.S.-China tensions and potential U.S. export curbs to China, could disrupt TSMC's supply chain and significantly impact its HPC revenue.
Opportunity: Sustained AI demand and potential pricing power driven by a multi-year growth runway and advanced node technology.
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
The dawn of artificial intelligence (AI) in early 2023 sparked a paradigm shift in technology spending that continues to this day. These advanced algorithms require state-of-the-art semiconductors to process the mounds of data that inform AI.
Where to invest $1,000 right now? Our analyst team just revealed what they believe are the 10 best stocks to buy right now, when you join Stock Advisor. See the stocks »
One of the unsung heroes of the AI revolution is Taiwan Semiconductor Manufacturing (NYSE:TSM), commonly known as TSMC. The company's advanced processes produce the world's most advanced AI chips. One popular narrative suggests an AI bubble is forming, and investors have been looking for evidence that AI adoption continues. As such, all eyes were on TSMC when the company reported its most recent quarterly results.
The record-shattering quarter offered the clearest evidence to date that AI has room to run.
Image source: Taiwan Semiconductor Manufacturing.
For the second quarter, TSMC generated revenue of NT$1.27 trillion (roughly $40.2 billion), up 36% year over year. The company's gross margin climbed 910 basis points to 67.7%, up from 58.6%, while its net profit margin surged 1290 basis points to 55.6%, up from 42.7% -- all thanks to TSMC's growing operating leverage. This drove diluted earnings per share (EPS) to a record NT$27.25, up 77% year over year and 23% quarter over quarter.
Demand was particularly strong across the company's leading-edge process technologies. TSMC's 3-nanometer (nm), 5nm, and 7nm wafers accounted for 30%, 33%, and 11% of revenue, respectively. Chips used in high-performance computing (HPC) provided the lion's share of revenue at 66%, while smartphone chips made up 22% of revenue.
TSMC is bullish about the future and has announced plans to increase its capex spending -- a move that gave investors pause. The company now plans to invest between $60 billion and $64 billion, up from its previous forecast of $52 billion to $56 billion. Management also noted that capital spending over the coming three years would be "significantly higher" than during the past three years.
While this spending is necessary to meet the growing demand, it will likely weigh on TSMC's margins in the near future. Turns out the right move isn't always the most popular.
In another development, TSMC said it will invest an additional $100 billion in its Arizona facility, bringing its total investment to $265 billion. "This is to build several or more semiconductor logical wafer fab for two-nanometer MP [mass production] technologies, as well as advanced packaging fabs to support the strong multi-year demand from our leading U.S. customers," said CEO C.C. Wei. These 2nm chips will mark the next generation of TSMC’s leading-edge semiconductors.
For the third quarter, TSMC is forecasting revenue of $44.6 billion to $45.8 billion, which would represent 37% growth at the midpoint of its guidance. Its outlook also calls for gross profit and operating profit margins of 66% and 57%, respectively. Even amid fears that demand for AI chips will slow, TSMC's robust results and strong guidance appear to contradict that view.
Moreover, given its near-monopoly in advanced semiconductors, shares are surprisingly affordable. The stock currently sells for 34 times earnings and 25 times forward earnings, an attractive price for a company that almost single-handedly underpins the AI revolution.
Before you buy stock in Taiwan Semiconductor Manufacturing, consider this:
The Motley Fool Stock Advisor analyst team just identified what they believe are the 10 best stocks for investors to buy now… and Taiwan Semiconductor Manufacturing wasn’t one of them. The 10 stocks that made the cut could produce monster returns in the coming years.
Consider when Netflix made this list on December 17, 2004... if you invested $1,000 at the time of our recommendation, you’d have $397,351! Or when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you’d have $1,304,257!
Now, it’s worth noting Stock Advisor’s total average return is 934% — a market-crushing outperformance compared to 210% for the S&P 500. Don't miss the latest top 10 list, available with Stock Advisor, and join an investing community built by individual investors for individual investors.
**Stock Advisor returns as of July 16, 2026. *
Danny Vena, CPA has no position in any of the stocks mentioned. The Motley Fool has positions in and recommends Taiwan Semiconductor Manufacturing. The Motley Fool has a disclosure policy.
The views and opinions expressed herein are the views and opinions of the author and do not necessarily reflect those of Nasdaq, Inc.
Four leading AI models discuss this article
"TSMC's record results and guidance validate multi-year AI semiconductor demand tailwinds that outweigh elevated capex and geopolitical risks at current valuations."
TSMC's Q2 results (NT$1.27T revenue, +36% YoY; 55.6% net margin; 77% EPS growth) confirm HPC/AI demand (66% of revenue) remains robust, with 3/5/7nm nodes driving 74% of sales. Forward guidance of ~37% revenue growth and Arizona expansion signal sustained AI capex. At 25x forward P/E against 30-40% expected EPS growth, valuation looks reasonable versus NVDA's 45x+. However, the article downplays geopolitical risk: Taiwan tensions or export curbs could disrupt 92% of leading-edge capacity.
Rising capex ($60-64B this year, "significantly higher" next three years) plus $100B Arizona outlay will pressure near-term margins and FCF; any AI slowdown or China-Taiwan escalation could turn the monopoly into stranded assets, making 25x forward earnings look expensive.
"TSMC's ability to maintain 50%+ net margins while aggressively scaling 3nm and 2nm capacity proves they possess a structural pricing moat that justifies a premium multiple despite geopolitical overhangs."
TSMC’s Q2 results are a masterclass in operating leverage, with net margins expanding to 55.6%—a figure rarely seen in capital-intensive manufacturing. The 66% revenue contribution from High-Performance Computing (HPC) confirms they are the primary toll booth for the AI gold rush. However, the market’s reaction to the increased $60B-$64B capex guidance highlights a critical tension: investors are worried about the 'peak cycle' risk. While the valuation of ~25x forward P/E is reasonable for a company growing EPS at 77%, the geopolitical risk premium regarding Taiwan remains the elephant in the room that this article completely ignores. If capital intensity continues to climb, free cash flow conversion could lag behind headline earnings growth.
The massive capex ramp-up suggests TSMC is entering a period of diminishing returns on invested capital, where the cost of building 2nm capacity may outpace the actual pricing power they can extract from hyperscalers.
"TSMC's record profits prove AI demand exists today, but $60-64B annual capex creates binary risk: either it's justified by locked-in orders (bullish), or it's capacity built ahead of demand (margin-destructive), and the article provides no evidence to distinguish between them."
TSMC's Q2 results are genuinely impressive — 77% YoY EPS growth, 67.7% gross margins, HPC at 66% of revenue all confirm AI demand is real and sustained. But the article conflates *current demand* with *sustainable profitability*. The capex surge ($60-64B annually, up from $52-56B, plus $100B Arizona) is framed as necessary but glosses over a critical risk: if TSMC builds fab capacity ahead of actual customer orders (a common semiconductor trap), utilization rates collapse and margins evaporate. The article also ignores geopolitical risk — U.S.-China tensions could crater Taiwan's export economics overnight. At 25x forward P/E, the stock prices in sustained 15%+ EPS CAGR; any capex misstep or demand softness triggers multiple compression, not just margin compression.
The strongest case against my skepticism: TSMC's 3-year capex guidance being 'significantly higher' than the past three years suggests management has multi-year visibility into customer orders (not just forecasts), and their track record of capacity planning is strong. If they're right, current valuation is cheap.
"Durable AI demand and a scalable capex cycle can support multi-year upside for TSMC, but only if ROIC remains high and AI spend stays elevated; otherwise a material multiple contraction is likely."
TSMC’s Q2 results underscore durable AI demand: revenue NT$1.27T (+36% YoY), gross margin 67.7%, net margin 55.6%, and EPS NT$27.25 (+77%). The CAPEX surge to $60–64B next three years and the $100B Arizona drive for 2nm/advanced packaging signals a multi-year growth runway and potential pricing power. That supports a bullish case for the stock even with macro headwinds. Yet risks loom: (1) margins may compress as new nodes scale and capex intensity rises; (2) AI demand could prove more cyclic than expected; (3) execution and geopolitical risks around large-capex projects could blunt returns. At 25x forward earnings, the valuation isn’t a free pass.
The strongest counter: the AI capex boom may be priced in; any slowdown in AI adoption or delays/cost overruns on Arizona/2nm fab work could compress margins and ROIC, risking multiple contraction.
"TSMC's monopoly at advanced nodes sustains pricing power that offsets rising capex intensity."
Gemini's diminishing-returns claim on 2nm capex misses that TSMC's pricing power is structural: hyperscalers have no viable alternative foundry at leading nodes. Arizona's $100B is largely grant-subsidized; real FCF pressure is overstated if utilization stays above 80%. The overlooked risk is US chip export curbs to China, which could cap HPC growth faster than any Taiwan invasion scenario.
"The forced localization of chip manufacturing in the US and elsewhere will permanently erode TSMC's superior operating leverage and capital efficiency."
Grok, your dismissal of FCF pressure is dangerous. Even with subsidies, the $100B Arizona outlay is a massive drag on ROIC compared to Taiwan-based expansion. Claude is right to fear the 'semiconductor trap'—building capacity ahead of demand. If hyperscalers like NVDA or AAPL pivot toward custom silicon efficiency or internalize more design, TSMC’s utilization rates will face a cliff. The geopolitical risk isn't just invasion; it's the inevitable forced localization of supply chains that destroys TSMC's operating leverage.
"Export curbs to China pose sharper downside than supply-chain localization, and the panel hasn't modeled the revenue cliff."
Gemini's forced-localization thesis is undercooked. TSMC's Taiwan cost advantage (labor, infrastructure, ecosystem) is durable; even with subsidies, Arizona fab economics don't justify reshoring *all* capacity. The real cliff risk: if US export curbs to China bite (Grok flagged this), TSMC loses 15-20% of HPC revenue overnight, not gradual demand shift. That's a 2025 tail risk the panel hasn't quantified.
"Utilization risk from mega-capex could compress ROIC even with 2nm/Arizona expansion, making a 25x forward multiple look stretched if returns revert to mid-teens."
Gemini raises valid concerns on diminishing returns, but the real stress test is utilization risk versus capex scale. If 2nm and Arizona capex outpace actual order flow from hyperscalers, TSMC could see utilization in the mid-80s and ROIC compression despite topline growth. Subsidies help, but they don’t fix timing misses or geopolitical tail risks that could crimp FCF. 25x forward isn’t a free pass if ROIC reverts to the mid-teens.
TSMC's Q2 results confirm robust AI demand, but geopolitical risks and potential overcapacity pose significant threats to future profitability.
Sustained AI demand and potential pricing power driven by a multi-year growth runway and advanced node technology.
Geopolitical risks, including U.S.-China tensions and potential U.S. export curbs to China, could disrupt TSMC's supply chain and significantly impact its HPC revenue.