Taiwan Semiconductor Manufacturing Company posted the best quarter in its history on July 16, 2026, and the numbers explain why chip stocks moved on the news. Second-quarter revenue hit $40.2 billion, the high end of the company’s own guidance, while net profit jumped 77.4% year over year. TSMC also raised its 2026 capital spending plan to as much as $64 billion and pledged another $100 billion to its Arizona campus, pushing its total committed investment in the state to $265 billion.
For anyone tracking AI chip supply, the TSMC earnings report matters more than almost any other single data point in the industry. Nvidia, AMD, Apple and Qualcomm all depend on TSMC’s advanced nodes to build the processors powering the current AI buildout, and the Q2 2026 results show how much of that demand is real rather than hype. Two-thirds of TSMC’s revenue now comes from high-performance computing chips alone.
This analysis breaks down what drove the beat, what the raised capex guidance signals about AI chip demand heading into 2027, and why the same earnings call that celebrated record profit also flagged the geopolitical risk that could complicate all of it.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
TSMC’s Q2 2026 Earnings, at a Glance
TSMC’s headline figures came in at NT$1.27 trillion in revenue, equal to $40.2 billion once converted at the high end of guidance. Net income reached NT$706.56 billion, or roughly $22.3 billion based on the company’s reported 55.6% net margin. Diluted earnings came to NT$27.25 per common share, or $4.31 per American depositary receipt.
TSMC management summed up the quarter plainly: “We concluded our second quarter with revenue of $40.2 billion at the high end of our guidance in U.S. dollar terms, driven by strong demand for our leading-edge process technologies,” according to a transcript published by Yahoo Finance. Chairman and CEO C.C. Wei struck a similar note on the same call: “We expect our business to be supported by continued strong AI and HPC demand.”
Profit growth outpaced revenue growth by a wide margin. Reuters, quoted via CNBC’s coverage of the earnings release, reported that TSMC “reported a 77.4% jump in second-quarter profit year on year, shattering estimates as the world’s largest contract-chipmaker continues to set consecutive record-breaking milestones.” Quarter over quarter, net income rose 23.4%. Gross margin landed at 67.7% and operating margin at 60.3%, both far above what most chipmakers report even in a strong year.
| Metric | Q2 2026 | Change |
|---|---|---|
| Revenue | $40.2B (NT$1.27 trillion) | +36% YoY |
| Net income | ~$22.3B (NT$706.56B) | +77.4% YoY, +23.4% QoQ |
| Diluted EPS | NT$27.25 ($4.31 per ADR) | N/A |
| Gross margin | 67.7% | N/A |
| Operating margin | 60.3% | N/A |
| Net margin | 55.6% | N/A |
| Q3 2026 revenue guidance | $44.6B–$45.8B | vs. $40.2B in Q2 |
Why Profit Outran Revenue
Revenue grew 36% year over year, a strong number by itself. Profit growing more than twice as fast tells the real story. TSMC is selling a richer mix of chips than it did a year ago. Every wafer that shifts from an older node to 3nm or 2nm carries a higher price tag and a fatter margin, and that mix shift shows up directly in the bottom line.
Management guided Q3 2026 revenue to a range of $44.6 billion to $45.8 billion, which would mark another sequential record if it holds. That guidance, paired with the mix shift already visible in Q2, suggests the margin expansion is not a one-quarter fluke. It reflects TSMC’s pricing power at the leading edge, where it faces essentially no serious competition for the most demanding AI accelerator orders.
The Capex Raise: $64 Billion and Climbing
TSMC lifted its 2026 capital expenditure outlook to a range of $60 billion to $64 billion, up from prior guidance of $52 billion to $56 billion. That’s roughly a $10 billion jump in planned spending inside a single guidance revision, large enough to move the entire semiconductor equipment supply chain that builds TSMC’s tools.
Alongside the capex increase, TSMC raised its full-year 2026 revenue growth outlook to “slightly above 40%,” up from a prior forecast of “above 30%.” Capex increases at this scale typically fund new fab construction, tool installation and capacity expansion for nodes still ramping, which lines up with what TSMC said about 2nm and its overseas buildout on the same call.
Higher capex now usually means higher output 18 to 24 months later. For customers waiting on advanced-node capacity, including AI chip designers who have complained about allocation limits, the raised guidance signals more supply is coming, just not immediately.
Inside the $265 Billion Arizona Commitment
The Arizona news landed alongside the earnings beat. CEO C.C. Wei announced an additional $100 billion investment in the state, bringing TSMC’s total committed Arizona spending to $265 billion, on top of the roughly $165 billion the company had already pledged in prior years.
That figure makes Arizona one of the largest single foreign manufacturing investments in US history. Scale doesn’t equal speed, though. TSMC’s Arizona fab is not expected to produce Apple’s most advanced chip nodes for several more years, meaning the highest-margin, most technically demanding work stays in Taiwan for now. That gap between the size of the financial commitment and the pace of leading-edge output in the US remains one of the more underappreciated details in TSMC’s growth story.
Diversification abroad carries a strategic tradeoff too. The more manufacturing capacity TSMC shifts overseas, the less exclusive leverage Taiwan retains over the global chip supply chain, a dynamic several analysts have flagged as a long-term consideration for Taipei as much as for TSMC’s own shareholders.
AI and HPC Now Drive Two-Thirds of Revenue
High-performance computing, the category that includes AI accelerators and data center processors, accounted for HPC accounted for 66% of TSMC’s Q2 2026 revenue. Smartphones made up 22%, and IoT devices contributed 5%. A year or two ago, smartphones and HPC sat much closer together. Now HPC runs roughly three times the size of the smartphone business inside TSMC’s revenue mix.
That shift mirrors what’s happening across the broader AI infrastructure buildout. Nvidia and SK Hynix signed a $500 billion memory supply agreement earlier in 2026, and data center lease commitments from Meta and Microsoft have topped $850 billion. TSMC sits upstream of nearly all of that spending. Almost every advanced AI chip, whether it’s an Nvidia GPU, an AMD accelerator or a custom silicon design, gets fabricated on a TSMC leading-edge line before it ever reaches a data center rack.
The 2nm Ramp: TSMC’s Next Growth Engine
TSMC’s newest node, N2 (2nm), contributed 3% of wafer revenue in Q2 2026, the first quarter it has shown up as a meaningful commercial contributor rather than early sampling volume. The rest of the node mix broke down as 3nm at 30%, 5nm at 33% and 7nm at 11%. Combined, chips built on processes of 7nm or smaller made up 77% of total wafer revenue.
That 77% figure is the clearest evidence of where TSMC’s pricing power actually lives. Older, larger nodes still run inside TSMC’s fabs, but they contribute a shrinking share of revenue every quarter as customers who need the fastest AI chips pay a premium for capacity on the newest processes. As 2nm output scales through the rest of 2026, it should keep pulling the overall revenue mix toward higher-margin territory.
Three-Year Turnaround: From Chip Glut to Record Profit
TSMC’s Q2 2026 numbers look even more dramatic next to where the company stood three years earlier. Full-year 2023 revenue fell 6.9% to $70.60 billion as the industry worked through a post-pandemic inventory correction, and net income dropped 18.5% to $27.82 billion. That was TSMC’s down year, and it came right before everything changed.
The recovery arrived fast. 2024 revenue climbed 25.4% to $88.34 billion, with net income up 27% to $35.33 billion, as AI accelerator demand pulled the company out of the downturn. By 2025, the acceleration was unmistakable. Revenue jumped 37.99% to $121.91 billion and net income rose 50.8% to $55.05 billion.
| Fiscal Year | Revenue (USD) | Revenue YoY | Net Income (USD) | Net Income YoY |
|---|---|---|---|---|
| 2023 | $70.60B | -6.9% | $27.82B | -18.5% |
| 2024 | $88.34B | +25.4% | $35.33B | +27.0% |
| 2025 | $121.91B | +37.99% | $55.05B | +50.8% |
Q2 2026’s 36% revenue growth and 77.4% profit growth show that momentum hasn’t slowed even as the year-over-year comparison base gets tougher every quarter.
Foundry Market Share: How Far Ahead Is TSMC?
TSMC held 72.3% of the global foundry market in Q1 2026, according to TrendForce. Counterpoint Research put the company’s share of the pure-foundry market at roughly 73% using a slightly different measurement method. Either way, TSMC controls close to three-quarters of a market that includes every company that manufactures chips designed by someone else.
Samsung Foundry held 6.5% of the market in the same period, down from approximately 9.8% in the fourth quarter of 2025 by TrendForce’s accounting. SMIC, China’s largest contract chipmaker, held 5.1%, and GlobalFoundries came in at 3.3%. Intel Foundry’s share was not broken out separately in the available data, a gap that says something on its own about how the market currently ranks Intel’s foundry ambitions against the other names in the field.
| Foundry | Market Share | Source |
|---|---|---|
| TSMC | 72.3% (73% pure-foundry basis) | TrendForce, Counterpoint |
| Samsung Foundry | 6.5% | TrendForce |
| SMIC | 5.1% | TrendForce |
| GlobalFoundries | 3.3% | TrendForce |
| Intel Foundry & others | Not separately disclosed | N/A |
Can Samsung or Intel Foundry Close the Gap?
Not anytime soon, based on the current numbers. Samsung’s foundry share actually shrank between the fourth quarter of 2025 and the first quarter of 2026, moving the wrong direction while TSMC’s grew. Samsung keeps investing in its own advanced nodes and has picked up some allocation from customers looking to diversify away from full TSMC dependence, but it hasn’t landed the kind of marquee, high-volume AI customer commitment that would meaningfully dent TSMC’s lead.
Intel Foundry remains the wildcard. The division sits at the center of Intel’s turnaround plan and has talked up its 18A process node, but it doesn’t yet show up with a standalone market-share figure in the leading tracking firms’ 2026 data, which suggests its external customer volume is still too small to register. For now, the foundry market remains close to a one-company story, with four rivals splitting what’s left over.
The Customer Concentration Question
TSMC doesn’t break out revenue by individual customer, but its node mix and platform mix both point to heavy reliance on a small group of fabless chip designers. Nvidia’s AI accelerators, Apple’s iPhone and Mac processors, AMD’s data center chips like the MI350P, and Qualcomm’s data center CPU push with its Dragonfly C1000 line all depend on TSMC capacity at the leading edge. Even Nvidia’s newer RTX Spark platform runs through the same advanced-node pipeline.
That concentration cuts both ways. It’s what lets TSMC command premium pricing and keep utilization high across its newest fabs. It also means a slowdown in AI capital spending, a customer’s move to build its own advanced-node capacity, or a surprise decision by a rival foundry to poach a major account would hit TSMC’s revenue mix faster than it would a foundry with a more spread-out customer base.
What Wall Street Is Saying About TSM Stock
Brokerage XTB valued TSMC shares at roughly $465 based on its own model, against a cited trading price of about $326, implying close to 43% potential upside. That figure reflects one analyst house’s estimate rather than a Wall Street consensus target, and investors should treat it accordingly given how fast estimates move after a quarter like this one.
TSMC’s own management offered a longer-range data point on the earnings call, saying the global semiconductor market could reach $1.5 trillion by 2030. That’s a market-wide forecast rather than a TSMC-specific figure, but given the company’s current 72%-plus foundry share, TSMC would likely capture an outsized portion of whatever that growth actually looks like.
The Geopolitical Risk Wall Street Keeps Discounting
Every dollar of TSMC’s raised guidance sits on top of a risk that never shows up in a quarterly earnings table. Reuters has described TSMC as sitting in the line of fire, noting that a conflict over Taiwan could destroy the company’s fabs outright or, if the facilities survived, sever them from the global supply chains they depend on to operate.
Taiwan Strait Risk in Numbers
Forecasters at the Swift Centre put the odds of a Chinese blockade of Taiwan by mid-2027 at 9%, with estimates across different models ranging from 4.9% to 15%. The same forecasting effort estimated only a 3.7% chance that any Taiwan-based fab gets destroyed or knocked offline for 90 days or more by 2027 under normal conditions. That probability jumps to 59% in the scenario where a blockade actually happens. A separate 2025 academic study on Taiwan’s semiconductor supply chain, tracked by policy researchers at CSIS, concluded that Beijing is increasingly likely to favor a quarantine strategy over a full invasion, arguing the supply chain is especially exposed to that kind of coercive pressure before 2027.
Export Controls and the China Question
US export controls already restrict TSMC’s ability to serve Chinese customers with its most advanced AI and supercomputing chips, part of the broader technology split between Washington and Beijing. Those restrictions raise compliance costs and narrow TSMC’s addressable market in China even though the company isn’t the direct target of most individual rules. Diversification to Arizona, along with reported plans in Japan and Europe, functions as a hedge against both the military risk and the export-control risk at once, spreading manufacturing capacity across jurisdictions with different exposure to a Taiwan Strait crisis.
What This Means for AI Chip Prices and Supply
For buyers of AI infrastructure, the practical takeaway from the TSMC earnings report is that advanced-node capacity stays tight enough to support premium pricing, and it will stay that way for a while. The raised capex guidance adds supply, but on an 18-to-24-month lag, so the current allocation crunch for 2nm and 3nm capacity isn’t going away in 2026. For a broader look at how that scarcity is reshaping the wider industry, see our ongoing coverage of AI chips in 2026.
That has direct consequences for GPU and AI accelerator pricing. When Nvidia, AMD or any other TSMC customer pays more for leading-edge wafers or waits longer for allocation, that cost and that delay tend to show up downstream in accelerator pricing and in how quickly cloud providers can add new AI capacity. Enterprises planning AI infrastructure budgets for 2027 should expect the current supply-constrained pricing environment to persist through at least the next several quarters.
5 Predictions for TSMC and the Foundry Market Through 2027
- 2nm output climbs fast. Expect N2’s share of wafer revenue to move from 3% toward 15% to 20% within two to three quarters as Apple, Nvidia and AMD ramp orders on the new node.
- Capex gets raised again. If AI accelerator demand keeps outpacing supply into late 2026, a third capex revision when TSMC reports Q3 results would not be a surprise.
- Arizona stays behind Taiwan on leading edge. The most advanced nodes should keep shipping from Taiwan first for at least the next few years, regardless of how much money flows into Arizona.
- Samsung and Intel Foundry keep losing ground unless one lands a marquee account. Neither rival has shown the kind of volume win in 2026 that would meaningfully shift the foundry market-share numbers.
- Geopolitical hedging accelerates. Expect more announcements tying TSMC’s Japan, Germany and US expansion to explicit risk-diversification language, as customers and governments price in Taiwan Strait risk.
Frequently Asked Questions
What was TSMC’s revenue in Q2 2026?
TSMC reported $40.2 billion in revenue for the second quarter of 2026, at the high end of its own guidance, up 36% from the same quarter a year earlier.
How much did TSMC raise its 2026 capex guidance?
TSMC raised its 2026 capital expenditure outlook to a range of $60 billion to $64 billion, up from a prior range of $52 billion to $56 billion.
How big is TSMC’s Arizona investment now?
TSMC’s total committed investment in Arizona reached $265 billion after CEO C.C. Wei announced an additional $100 billion pledge alongside the Q2 2026 earnings report.
What percentage of TSMC’s revenue comes from AI and HPC chips?
High-performance computing, which includes AI accelerators, made up 66% of TSMC’s revenue in Q2 2026, compared with 22% for smartphones and 5% for IoT devices.
What is TSMC’s market share in the global foundry industry?
TSMC held approximately 72.3% of the global foundry market in the first quarter of 2026, according to TrendForce, compared with 6.5% for Samsung Foundry and 5.1% for SMIC.
Is TSMC’s 2nm (N2) node in production yet?
Yes. TSMC’s N2 process contributed 3% of wafer revenue in Q2 2026, its first quarter as a meaningful commercial contributor rather than early sampling volume.
What is the biggest risk to TSMC’s growth?
The clearest risk is geopolitical. Forecasters at the Swift Centre put the odds of a Chinese blockade of Taiwan by mid-2027 at around 9%, a scenario that would raise the chance of a Taiwan fab going offline for 90 or more days to 59%, compared with 3.7% under normal conditions.
Why did TSMC’s profit grow faster than its revenue?
A shift toward higher-margin, leading-edge nodes drove the gap. Chips built on 7nm or smaller processes made up 77% of Q2 2026 wafer revenue, and those advanced nodes carry richer pricing than TSMC’s older process technologies.
Related Coverage
- Nvidia, SK Hynix Ink $500B AI Memory Deal [2026]
- China’s Rare Earth Curbs Put $6.5T of Tech at Risk [2026]
- South Korea’s $880B Chip Plan Hits Power Wall [2026]
- AMD MI350P vs Nvidia H200: 40% Faster, $30K Price [2026]
- Qualcomm’s Dragonfly C1000 Lands Meta, Eyes $15B [2026]
- Data Center Leases Hit $850B, Meta, Microsoft Lead [2026]
- Nvidia RTX Spark Debuts: 128GB Memory, 1 Petaflop AI [2026]
