TSMC Q1 2026 Profit Jumps 58% to a Record as AI Chip Demand Defies Iran War

Taiwan Semiconductor Manufacturing Co. reported a 58% year-over-year jump in first-quarter net profit last Thursday, April 16, setting a fresh record and blowing past analyst forecasts on runaway demand for AI accelerators. Net profit came in at NT$572.8 billion — roughly $17.6 billion — on revenue of NT$1.134 trillion, according to numbers confirmed by Reuters and CNBC. The company also raised its full-year revenue growth guidance to "above 30%," up from "around 30%" previously, while warning that the ongoing U.S.–Iran war could pressure raw-material pricing.

Because TSMC is the foundry that manufactures Apple Silicon, Nvidia's data-center GPUs, AMD's EPYC and Instinct parts, Qualcomm's Snapdragons and most of the AI accelerator custom silicon the hyperscalers depend on, this quarterly print is a near-direct readout on the health of the entire AI hardware economy. On that measure, the economy is running hot.

Key Numbers From the Quarter

• Net profit: NT$572.8 billion (~$17.6 billion), up 58.3% year over year
• Revenue: NT$1.134 trillion (~$35 billion), up roughly 35% year over year
• Gross margin: 66.2%, beating TSMC's own 63–65% guidance
• Operating profit: NT$659 billion, up 62% year over year
• HPC platform share of revenue: 61%, up 20% YoY
• Smartphone platform revenue: down 11% YoY

The mix shift is as notable as the headline number. Smartphone silicon — historically TSMC's backbone — was down, while high-performance computing, the category TSMC uses to bucket AI accelerators and server CPUs, now represents a clear majority of revenue. The company also disclosed that demand is migrating from 5nm to the more expensive 3nm process node, which is the mechanical reason gross margin expanded.

What TSMC Said About AI Demand

On the earnings call, CEO C.C. Wei told analysts the company has received "strong signals and a positive outlook" from major customers on AI spending through the rest of 2026. That is carefully chosen language — TSMC rarely names customers, and Wei did not this time either. But with Nvidia, AMD, Apple, Broadcom, and the hyperscaler custom-silicon programs all fabbing leading-edge parts at TSMC, "strong signals" from its top customers is effectively a forward read on AI infrastructure capex.

The 66.2% gross margin is the quieter story. Semiconductor analysts have spent two years waiting for AI-chip pricing to normalize as capacity catches up. TSMC's Q2 guidance — gross margin of 65.5% to 67.5% — suggests that normalization is not happening yet. Customers are still willing to pay premium pricing for 3nm capacity, and TSMC is still the only foundry that can deliver it at scale.

The Iran War Warning

The caveat in the release is geopolitical. CFO Wendell Huang told investors the company "expects the Middle East conflict will not affect its supply of critical chipmaking materials like helium in the short term," but added that "a spike in these raw material prices could pressure the company's profitability and the global economy." That is TSMC-speak for a real risk that it is actively modeling.

The U.S.–Iran war, which escalated in late February, has disrupted Persian Gulf shipping and introduced new friction on commodity routes. Helium, neon, and specialty gases used in lithography are global markets; a protracted disruption would hit every foundry, not just TSMC. So far, the AI investment boom has absorbed the pressure — Bloomberg reporting referenced by Asia Financial noted that AI capex has not weakened despite the conflict. That balance is the key thing to watch in Q2.

Why It Matters Beyond Taiwan

Every major AI product launch in 2026 traces back through TSMC's Hsinchu and Arizona fabs. The Nvidia RTX 5080 we reviewed earlier this year is a TSMC 4nm part. The Snapdragon X2 and Apple M5 laptop chips are both 3nm TSMC. Nvidia's Blackwell and Rubin-generation data-center GPUs, AMD's MI400-series accelerators, Google's TPU v6, Amazon's Trainium — all of it is on TSMC leading-edge nodes. When TSMC's 3nm capacity is sold out and priced at 66%-plus gross margin, that is the ceiling on how fast the rest of the industry can ship.

It is also a hedge against the AI-bubble narrative. A company that prints $17.6 billion in quarterly profit on contracted orders from roughly every meaningful chip buyer on earth is a harder story to dismiss than a standalone model company with uncertain unit economics. For context on where that spending is ultimately going, our coverage of AI venture capital hitting $242 billion in Q1 2026 traces the capital stack one layer up.

What's Next

TSMC guided Q2 2026 revenue to a range that implies another sequential record, with gross margin of 65.5% to 67.5%. The company's Arizona N2 ramp — originally scheduled for late 2026 — is on track, and the second Arizona fab is now expected to pull in six months ahead of plan, responding to U.S. customer pressure for domestic leading-edge supply.

The risks are geopolitical, not technical. A widening Middle East conflict, a cross-strait incident, or a renewed round of U.S. export controls on advanced chipmaking equipment could all change the math. Absent those, the read from Hsinchu is straightforward: the AI buildout has not slowed, 3nm is still a seller's market, and the 2nm ramp later this year is already effectively pre-sold.

When TSMC's quarterly print is a bellwether, and the bellwether says "above 30% growth, record margins, strong customer signals," the industry is telling you what to expect from Nvidia, AMD, Broadcom, and Apple earnings over the next six weeks.