TSMC Reports Record Quarterly Profit Up 69% and Hikes 2026 Guidance on Surging AI Demand

Taiwan Semiconductor Manufacturing Company is now clearing $23.6 billion in pure profit every 90 days—a magnitude that means the foundry nets roughly $260 million every single day it operates. The world's largest contract chipmaker reported a 69 percent year-over-year jump in third-quarter net income, blowing past its own guidance on the back of relentless artificial intelligence infrastructure spending.[1][3]

Revenue reached $46.72 billion for the quarter ending September 30, 2026, representing a 41.2 percent increase in US dollar terms. The figures confirm that the AI build-out is actually shipping physical silicon, not just generating marketing announcements and press releases.[4]

High-performance computing, the segment that includes AI accelerators for companies like Nvidia and AMD, grew 22 percent sequentially. It now accounts for 68 percent of all TSMC revenue, fundamentally transforming the foundry's business model away from consumer electronics.[1]

"AI-related demand is extremely robust and far above our supply capability," Chairman and Chief Executive Officer C.C. Wei told analysts on the October 15 earnings call. "We did not see any sign of slowdown."

That supply gap forced TSMC to raise its full-year 2026 revenue growth forecast. The company now expects sales to expand by approximately 44 percent in US dollar terms, a sharp upward revision from earlier projections that hovered near 30 percent.[2][5]

The cost of the bleeding edge

Despite the record revenue, the cost of manufacturing the world's most advanced silicon is becoming visible. TSMC's gross margin slipped 1.3 percentage points sequentially to 66.4 percent, landing inside guidance but highlighting the friction of progress.[4]

Chief Financial Officer Wendell Huang attributed the margin compression directly to the initial mass production of the company's 2-nanometer technology. Ramping up a new manufacturing node always dilutes profitability before yields stabilize and scale efficiencies take over.[2]

The 2-nanometer process is already shipping in volume, accounting for 8 percent of total wafer revenue in its first year. This distinguishes TSMC's actual capability from competitors who are still announcing future nodes without delivering them to paying customers.[1]

Older, but still advanced, nodes carry the bulk of the business. The 3-nanometer process generated 33 percent of wafer revenue, while 5-nanometer chips accounted for 28 percent. The 7-nanometer process added another 10 percent.[1][4]

Combined, technologies at 7-nanometer and below represented 79 percent of TSMC's total wafer sales. The foundry has effectively abandoned trailing-edge production, leaving legacy chips to rivals while it monopolizes the premium market where margins are highest.[5]

Expanding the global footprint

To maintain this pace, TSMC is spending capital at an unprecedented rate. Management reiterated a 2026 capital expenditure budget of $60 billion to $64 billion, with spending expected to remain elevated through 2028 to meet the backlog of orders.[2]

The company is currently advancing the construction of approximately 20 fabrication facilities worldwide. This massive build-out includes new plants in Taiwan, Japan, and the United States, significantly exceeding its historical pace of building four to five fabs simultaneously.

Overseas expansion brings its own structural costs. Huang warned analysts that overseas fabs are currently diluting gross margins by about 2 percent, a figure expected to widen to 3 or 4 percent in the coming years as more facilities ramp up.[4]

The Arizona facilities, long a focal point of US industrial policy, are part of this margin drag. TSMC is absorbing the higher labor and operational costs of American manufacturing while pushing to bring the fabs online and satisfy political demands for domestic supply chains.

To mitigate these expenses, TSMC is negotiating with both suppliers and customers. The company's strategy relies on its indispensable role in the AI supply chain, ensuring that hyperscalers will pay a premium for guaranteed capacity regardless of where the chips are fabricated.[4][5]

Pricing power and market reality

Because TSMC holds a near-monopoly on reliable advanced manufacturing, it can pass many of these costs onto its clients. The foundry's ability to maintain a 58.7 percent operating margin shows immense pricing power in a constrained market.[1][4]

Fourth-quarter guidance suggests the momentum will not break before the year ends. TSMC projects revenue between $52.3 billion and $54.3 billion for the final three months of 2026, assuming an exchange rate of 31.8 New Taiwan dollars to the US dollar.[2]

If achieved, the midpoint of that guidance would represent a 58 percent year-over-year jump. The company expects gross margins to hold steady between 65 and 67 percent during the fourth quarter, even as the 2-nanometer ramp accelerates.[4]

The broader smartphone and consumer electronics markets remain sluggish, challenged by rising component prices and economic uncertainty. Smartphone revenue grew just 10 percent sequentially, accounting for only 21 percent of TSMC's total sales.

Automotive revenue climbed 23 percent sequentially but still represents just 4 percent of the business. Internet-of-things revenue rose 5 percent, while digital consumer electronics declined 12 percent, shrinking to a mere 1 percent of total revenue.

This divergence highlights how completely AI infrastructure is carrying the semiconductor sector. Without the massive capital expenditures from hyperscalers buying AI accelerators, the foundry's growth would look entirely different, tied to a stagnant consumer hardware cycle.[4][5]

The ultimate test for TSMC is not whether it can manufacture the chips, but whether its customers can generate enough return on their AI investments to keep buying them. For now, the orders are signed, the silicon is shipping, and the foundry is collecting the toll.[4][5]