Everyone expected the AI bellwethers to rocket on record beats. TSMC and ASML delivered exactly that—stronger revenue, fatter margins, raised full-year guidance—yet the stocks dropped 2-6% post-earnings. Consensus calls it the first crack in the AI spending frenzy. You should hear it for what it really is: classic profit-taking after monster rallies, not evidence that hyperscalers are suddenly hitting pause. The factories tell the opposite story.
TSMC just posted its fourth straight record quarter. Q1 revenue hit $35.9 billion, up 40.6% year-over-year, while net profit surged 58.3% to a record NT$572.5 billion. Gross margin expanded to 66.2%. They guided Q2 revenue to $39-40.2 billion and raised full-year 2026 growth above 30% in USD terms, pushing capex toward the high end of $52-56 billion. The stock still sold off. That's the deadpan fact bomb: record results, higher guidance, and the market treated it like bad news.
This isn't fading AI demand. It's persistent supply constraints that demand keeps outrunning. TSMC's 3nm and advanced packaging capacity is already oversubscribed, with management flagging "extremely strong" demand for AI accelerators and leading-edge nodes. Advanced technologies (7nm and below) made up 74% of wafer revenue in Q1, HPC platform hit 61% of total revenue, and they're ramping 2nm while converting lines to chase orders locked in through 2026 and beyond. When the company powering Nvidia, AMD, and the rest says capacity is sold out and they're accelerating spend, the message is clear: silicon shortage isn't easing—it's structural.
ASML echoes it from the equipment side. Q1 net sales reached €8.8 billion, beating expectations, with gross margin landing at 53.0% on the high end and net income of €2.8 billion. They raised full-year 2026 sales guidance to €36-40 billion from the prior €34-39 billion, citing robust logic and DRAM AI demand plus strong uptake of High-NA EUV systems. The 5-6% drop came amid China export noise, but the order book and margin performance confirm the lithography bottleneck remains the gatekeeper for every cutting-edge chip. Hyperscalers aren't slowing their buildouts; they're fighting for slots.
The variant perception here is straightforward: the market is lazily reading these dips as early warnings of an AI top because it's simpler than grappling with multi-year supply tightness. TSMC's advanced node demand already exceeds what they can physically build, even as they push capex higher. ASML's raised outlook shows customers continuing to invest aggressively in capacity. Those post-earnings drops after TSMC's ~150% run and ASML's strong YTD performance? Pure position squaring. Fundamentals point to AI-related growth accelerating faster than new supply can arrive.
Connect it to the broader chip complex and the implication snaps into focus. When your two most critical upstream suppliers both beat, raise, and reaffirm sold-out capacity while stepping up investment, downstream revenue isn't peaking—it's still ramping. Near-term weakness from sentiment and profit-taking simply creates the setup. The orders are in hand. The fabs are running flat out. The only thing that got ahead of itself is the fear trade.
This thesis isn't blind cheerleading. It has clear, measurable kill criteria: if TSMC or ASML cuts FY2026 revenue guidance or Q2/Q3 numbers below current raised levels by July earnings, the supply-constraint story fractures. Same if TSMC confirms 2026 capex below $50 billion or customer order pauses appear in reports. Broader chip sector revenue growth (SOXX constituents or Nvidia/AMD) slowing below 15% YoY in Q2 reports with weakening AI commentary would shift the narrative. Or if ASML China restrictions force more than a 10% downward revision by October. Those are the falsifiable triggers that would prove the market right.
Bottom line: the sell-the-news reaction in TSMC and ASML isn't the start of an AI top. It's the market handing you a cleaner entry point while capacity constraints keep biting hard. Ignore the day-one ticker noise. Watch the factories, the order books, and the capex ramps. Dips here are a buying setup for the multi-year AI buildout still ahead. Position accordingly before the tightness drives the next leg.