ASM International forecasts second-quarter revenue that beats estimates

ASM International guided second-quarter revenue to around 980 million euros, well above the 883.9 million-euro consensus, after first-quarter revenue also beat estimates at 862.5 million euros versus 828.5 million expected. The company said customers are still investing in leading-edge chip technology and pilot-line work for the 1.4nm node, supporting demand tied to AI-related capacity. The update is positive for ASM shares and sentiment around semiconductor equipment spending.

Analysis

This is less a simple equipment beat than a read-through on where the AI capex cycle is still under-penetrated. The most important signal is not the magnitude of the guide, but that customers are still adding capacity at the leading edge while simultaneously funding pre-production work for the next node; that combination usually means the capex curve is still in the acceleration phase, not the late-cycle replacement phase. For ASML and its peers, that supports both pricing power and utilization, while weakening the bear case that AI spending is merely being re-timed rather than expanded.

The second-order winner is the advanced-node ecosystem: lithography adjacencies, inspection/metrology, process control, and high-end wafer materials should see a more durable demand tail than the headline toolmakers because every new node adds complexity faster than it adds wafer starts. That also tends to tighten the bottleneck around top-tier capacity at the foundry level, which can support premium pricing for leading-edge wafers and reduce bargaining power for customers on future tool orders. In contrast, less differentiated semi-capex names may lag if investors rotate toward companies with direct exposure to the few nodes actually getting funded.

The contrarian risk is that this is a sentiment-positive quarter, not necessarily a clean demand-proof quarter. With bookings disclosure removed, investors lose the best leading indicator for order inflection, so the market may overprice a straight-line upgrade to 2025-26. If AI infrastructure spend pauses for even one quarter, high-multiple equipment names can de-rate quickly because the stock already embeds a multi-year growth runway.

For NVDA and AAPL, the takeaway is nuanced: the memo supports confidence that supply-chain investment is continuing ahead of next-gen launches, but it does not prove end-demand acceleration. If customer capex is being pulled forward mainly to secure future node access, then near-term inventory digestion risk remains, especially for consumer-facing demand that depends on a broader monetization cycle than server spend does.

Key Decisions for Investors

• Go long ASML on a 3-6 month horizon on any post-earnings dip; upside comes from a re-acceleration in 2025-26 capex expectations, while downside is limited unless AI spend visibly slows. Risk/reward favors adding on weakness rather than chasing strength.
• Pair long ASML / short a basket of slower-growing semi-capex names over the next 1-2 quarters to express a quality premium trade. The thesis is that investors will pay up for direct leading-edge exposure and penalize companies without clear node-specific leverage.
• Buy call spreads on NVDA with 6-9 month tenor to express the read-through that foundry and packaging investment is still expanding ahead of next-gen product ramps. Keep size modest: the signal is supportive, but not enough to underwrite aggressive multiple expansion alone.
• Avoid adding to AAPL on this headline alone; treat it as a supply-chain confidence signal, not an end-demand catalyst. If anything, use strength to trim if the stock rallies on AI-capex enthusiasm without a corresponding product-cycle improvement.
• Set a catalyst watchlist for the next quarterly order trends across the equipment group; if investors continue to ignore bookings suppression and the stock still rallies on guides, momentum can extend for another 1-2 quarters. If guides stop beating, the de-rating could be sharp because the market is relying on narrative more than disclosed order data.