TSMC Earnings Call Preview: Can AI Demand Push Guidance Even Higher?
Claire Weston
TSMC holds its Q2 earnings call on July 16. The market is betting the company will lift its full-year USD revenue growth target from above 30% to roughly 35% — the answer hinges on how far AI demand has spread and whether capacity can keep up.
What is the market betting on?
TSMC raised its full-year USD revenue growth forecast from near 30% to above 30% in April. Some investors now expect a further lift to around 35%.
This means → the bet is not just "good quarter or not" but whether TSMC will raise guidance twice in one year — historically uncommon.
Supply-chain sources say multiple major clients plan Q3 flagship launches, raising the odds of a second upgrade.
Where exactly has AI demand spread?
AI chip demand has expanded beyond Nvidia GPUs into ASICs (chips custom-built for specific tasks), CPUs, high-speed networking chips, HBM (high-bandwidth memory), optical communications, and silicon photonics.
In plain terms = Nvidia used to be the sole engine pulling TSMC forward. Now the entire AI infrastructure chain is placing orders — more diverse, less dependent on any single customer.
This reflects a shift from "buy GPUs, build a cluster" to full-scale data-center overhauls, deepening reliance on TSMC's advanced nodes and packaging.
Is there enough capacity?
4 nm / 3 nm lines are running full. Clients keep adding wafer starts; the early expansion of 3 nm capacity confirms AI demand has far outrun original plans.
Once 2 nm ramps, capacity at the Hsinchu and Kaohsiung fabs is expected to be booked almost immediately.
This means → the key question has flipped from "Can TSMC win orders?" to "Can TSMC supply enough capacity?"
Are rivals taking share?
Recent reports cite Google shifting some TPU work to Intel's EMIB packaging, Apple exploring Intel options, and Anthropic evaluating Samsung 2 nm for in-house chips.
Sources characterize these moves as overflow from TSMC's capacity crunch — limited, non-core transfers, not real order losses.
In plain terms = clients are not leaving by choice. TSMC's lines are full, and what spills over is peripheral work.
Will 2 nm drag down margins?
TSMC has disclosed that the 2 nm ramp will dilute gross margin by roughly 2–3 percentage points in 2026. Overseas fabs will add further pressure after that.
Sources still expect margins to hold near 65%, supported by the large share of high-priced advanced nodes, strong utilization, and pricing power.
This means → near-term pressure is real, but TSMC's pricing power and product mix can offset the cost drag of new lines.
What should investors listen for on July 16?
The signal investors want most is confirmation that AI demand is durable, especially after recent rumors of Nvidia product delays and slowing AI compute needs.
Chairman C.C. Wei is expected to argue that rapid AI-platform iteration will only deepen demand for TSMC's process, packaging, and manufacturing capabilities.
This reflects the real focus of the call: not Q2 results themselves, but whether TSMC will revise the AI up-cycle outlook higher once more.
Content is for reference only, not financial advice.